1 2017-12-18 Jeff Law <law@redhat.com>
4 2017-11-19 Jeff Law <law@redhat.com>
6 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
7 of degenerates resulting from ignoring an edge.
9 2017-12-18 Martin Sebor <msebor@redhat.com>
12 PR tree-optimization/78450
13 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
14 (handle_builtin_strlen): Call it.
16 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
18 PR rtl-optimization/83424
19 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
21 2017-12-18 Marek Polacek <polacek@redhat.com>
24 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
25 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
28 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
30 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
31 (umaddsidi4, umaddsidi_split): Likewise.
33 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
35 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
38 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41 * config/s390/s390.c (s390_option_override): Avoid strncpy.
43 2017-12-18 Richard Biener <rguenther@suse.de>
45 PR tree-optimization/81877
46 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
47 (outermost_indep_loop): Adjust.
48 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
49 (can_sm_ref_p): Adjust.
51 2017-12-18 Richard Biener <rguenther@suse.de>
54 * tree.c (array_at_struct_end_p): Return true if the underlying
55 object has space for at least one element in excess of what
56 the array domain specifies.
58 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
60 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
61 example to avoid overfull hbox.
62 * doc/invoke.texi (Option Summary): Add missing @gol.
63 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
66 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
67 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
69 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
70 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
71 (AArch64 Options): Document -mverbose-cost-dump.
72 (ARM Options): Likewise, plus -mflip-thumb.
74 2017-12-17 Martin Sebor <msebor@redhat.com>
77 * gimple-ssa-warn-restrict.c
78 (builtin_memref::offset_out_of_bounds): Correct the handling of
81 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
83 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
85 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
87 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
89 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
91 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
94 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
96 * doc/invoke.texi: Fix some typos.
98 2017-12-16 Martin Sebor <msebor@redhat.com>
100 PR tree-optimization/78918
101 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
102 * builtins.c (check_sizes): Rename...
103 (check_access): ...to this. Rename function arguments for clarity.
104 (check_memop_sizes): Adjust names.
105 (expand_builtin_memchr, expand_builtin_memcpy): Same.
106 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
107 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
108 (check_strncat_sizes, expand_builtin_strncat): Same.
109 (expand_builtin_strncpy, expand_builtin_memset): Same.
110 (expand_builtin_bzero, expand_builtin_memcmp): Same.
111 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
112 (maybe_emit_sprintf_chk_warning): Same.
113 (expand_builtin_strcpy): Adjust.
114 (expand_builtin_stpcpy): Same.
115 (expand_builtin_with_bounds): Detect out-of-bounds accesses
116 in pointer-checking forms of memcpy, memmove, and mempcpy.
117 (gcall_to_tree_minimal, max_object_size): Define new functions.
118 * builtins.h (max_object_size): Declare.
119 * calls.c (alloc_max_size): Call max_object_size instead of
120 hardcoding ssizetype limit.
121 (get_size_range): Handle new argument.
122 * calls.h (get_size_range): Add a new argument.
123 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
124 * doc/invoke.texi (-Wrestrict): Adjust, add example.
125 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
127 (gimple_fold_builtin_memory_chk): Same.
128 (gimple_fold_builtin_stxcpy_chk): New function.
129 * gimple-ssa-warn-restrict.c: New source.
130 * gimple-ssa-warn-restrict.h: New header.
131 * gimple.c (gimple_build_call_from_tree): Propagate location.
132 * passes.def (pass_warn_restrict): Add new pass.
133 * tree-pass.h (make_pass_warn_restrict): Declare.
134 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
136 (handle_builtin_strcat): Same.
137 (strlen_optimize_stmt): Rename...
138 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
139 stpncpy, strncpy, and their checking forms.
141 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
143 PR rtl-optimization/82849
144 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
145 and get_max_loop_iterations_int.
147 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
148 Alan Hayward <alan.hayward@arm.com>
149 David Sherwood <david.sherwood@arm.com>
151 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
152 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
153 size as a poly_uint64.
154 (mode_for_vector, mode_for_int_vector): Take the number of vector
155 elements as a poly_uint64.
156 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
157 size as a poly_uint64.
158 (mode_for_vector, mode_for_int_vector): Take the number of vector
159 elements as a poly_uint64.
161 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
162 Alan Hayward <alan.hayward@arm.com>
163 David Sherwood <david.sherwood@arm.com>
165 * machmode.h (MACRO_MODE): New macro.
166 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
167 * targhooks.c (default_libcall_value, default_secondary_reload)
168 (default_memory_move_cost, default_register_move_cost)
169 (default_class_max_nregs): Likewise.
171 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
172 Alan Hayward <alan.hayward@arm.com>
173 David Sherwood <david.sherwood@arm.com>
175 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
176 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
178 * genautomata.c (main): Likewise.
179 * genconditions.c (write_header): Likewise.
180 * genemit.c (main): Likewise.
181 * genextract.c (print_header): Likewise.
182 * genopinit.c (main): Likewise.
183 * genoutput.c (output_prologue): Likewise.
184 * genpeep.c (main): Likewise.
185 * genpreds.c (write_insn_preds_c): Likewise.
186 * genrecog.c (writer_header): Likewise.
187 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
188 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
189 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
190 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
191 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
192 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
193 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
194 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
195 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
196 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
197 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
198 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
199 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
200 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
201 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
202 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
203 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
204 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
205 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
206 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
207 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
208 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
209 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
210 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
211 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
212 * config/darwin.c (IN_TARGET_CODE): Likewise.
213 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
214 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
215 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
216 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
217 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
218 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
219 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
220 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
221 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
222 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
223 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
224 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
225 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
226 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
227 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
228 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
229 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
230 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
231 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
232 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
233 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
234 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
235 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
236 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
237 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
238 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
239 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
240 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
241 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
242 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
243 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
244 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
245 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
246 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
247 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
248 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
249 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
250 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
251 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
252 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
253 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
254 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
255 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
256 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
257 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
258 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
259 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
260 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
261 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
262 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
263 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
264 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
265 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
266 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
267 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
268 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
269 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
270 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
271 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
272 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
273 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
274 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
275 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
276 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
277 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
278 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
279 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
280 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
281 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
282 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
283 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
284 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
285 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
286 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
287 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
288 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
289 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
290 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
291 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
292 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
293 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
294 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
295 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
296 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
297 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
298 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
299 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
300 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
301 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
302 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
303 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
304 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
305 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
306 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
307 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
308 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
309 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
310 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
311 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
312 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
313 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
314 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
315 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
316 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
317 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
318 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
320 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
321 Alan Hayward <alan.hayward@arm.com>
322 David Sherwood <david.sherwood@arm.com>
324 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
327 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
328 Alan Hayward <alan.hayward@arm.com>
329 David Sherwood <david.sherwood@arm.com>
331 * doc/generic.texi (VEC_SERIES_EXPR): Document.
332 * doc/md.texi (vec_series@var{m}): Document.
333 * tree.def (VEC_SERIES_EXPR): New tree code.
334 * tree.h (build_vec_series): Declare.
335 * tree.c (build_vec_series): New function.
336 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
337 * tree-pretty-print.c (dump_generic_node): Likewise.
338 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
339 * tree-inline.c (estimate_operator_cost): Likewise.
340 * expr.c (expand_expr_real_2): Likewise.
341 * optabs-tree.c (optab_for_tree_code): Likewise.
342 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
343 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
344 * expmed.c (make_tree): Handle VEC_SERIES.
345 * optabs.def (vec_series_optab): New optab.
346 * optabs.h (expand_vec_series_expr): Declare.
347 * optabs.c (expand_vec_series_expr): New function.
348 * tree-vect-generic.c (expand_vector_operations_1): Check that
349 the operands also have vector type.
351 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
352 Alan Hayward <alan.hawyard@arm.com>
353 David Sherwood <david.sherwood@arm.com>
355 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
356 (VEC_COND_EXPR): Add missing @tindex.
357 * doc/md.texi (vec_duplicate@var{m}): Document.
358 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
359 * tree.c (build_vector_from_val): Add stubbed-out handling of
360 variable-length vectors, using VEC_DUPLICATE_EXPR.
361 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
362 * cfgexpand.c (expand_debug_expr): Likewise.
363 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
364 * tree-inline.c (estimate_operator_cost): Likewise.
365 * tree-pretty-print.c (dump_generic_node): Likewise.
366 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
367 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
368 (test_vec_duplicate_folding): New function.
369 (fold_const_c_tests): Call it.
370 * optabs.def (vec_duplicate_optab): New optab.
371 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
372 * optabs.h (expand_vector_broadcast): Declare.
373 * optabs.c (expand_vector_broadcast): Make non-static. Try using
375 * expr.c (store_constructor): Try using vec_duplicate_optab for
377 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
379 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
382 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
383 div/mod latencies a bit.
385 2017-12-15 Jeff Law <law@redhat.com>
387 PR tree-optimization/36550
388 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
389 (mark_threaded_blocks): Rewrite code to avoid block copying when
390 optimizing for size. Don't pessimize blocks which will be
391 copied, but all the statements will be dead.
393 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
395 PR tree-optimization/81165
396 * tree-ssa-threadupdate.c (uses_in_bb): New.
397 (estimate_threading_killed_stmts): New.
398 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
399 * tree-ssa-threadedge.c
400 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
403 2017-12-15 Jeff Law <law@redhat.com>
405 PR tree-optimization/83410
406 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
407 threads when parallelizing loops.
409 2017-12-15 Jakub Jelinek <jakub@redhat.com>
411 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
413 * config/alpha/alpha.c (vms_attribute_table): Swap
414 affects_type_identity and handler fields, adjust comments.
415 * config/mips/mips.c (mips_attribute_table): Likewise.
416 * config/visium/visium.c (visium_attribute_table): Likewise.
417 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
418 * config/microblaze/microblaze.c (microblaze_attribute_table):
420 * config/spu/spu.c (spu_attribute_table): Likewise.
421 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
422 * config/arc/arc.c (arc_attribute_table): Likewise.
423 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
424 * config/v850/v850.c (v850_handle_interrupt_attribute,
425 v850_handle_data_area_attribute): Formatting fixes.
426 (v850_attribute_table): Swap affects_type_identity and handler
427 fields, adjust comments.
428 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
429 * config/arm/arm.c (arm_attribute_table): Likewise.
430 * config/avr/avr.c (avr_attribute_table): Likewise.
431 * config/s390/s390.c (s390_attribute_table): Likewise.
432 * config/sh/sh.c (sh_attribute_table): Likewise.
433 * config/i386/i386.c (ix86_handle_cconv_attribute,
434 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
435 (ix86_attribute_table): Swap affects_type_identity and handler
436 fields, adjust comments.
437 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
438 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
439 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
440 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
441 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
442 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
443 * config/rx/rx.c (rx_attribute_table): Likewise.
444 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
445 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
446 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
447 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
448 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
449 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
450 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
451 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
452 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
453 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
454 * doc/plugins.texi (user_attr): Likewise. Add NULL for
456 * attribs.c (empty_attribute_table): Swap affects_type_identity and
458 (register_scoped_attributes, decl_attributes): Formatting fixes.
460 PR tree-optimization/83269
461 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
462 subtraction in arg0's type if type is signed and arg0 is unsigned.
466 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
467 convert? on @0 instead of convert. Check type of @1, not @0.
468 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
469 convert? on @0 instead of convert. Check type of @1, not @0.
470 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
471 only optimize if either both @1 and @2 types are narrower
472 precision, or both are wider or equal precision, and in the former
473 case only if both have undefined overflow.
475 2017-12-15 Richard Biener <rguenther@suse.de>
478 * internal-fn.def (IFN_NOP): Add.
479 * internal-fn.c (expand_NOP): Do nothing.
480 * lto-streamer-in.c (input_function): Instead of removing
481 sanitizer calls replace them with IFN_NOP calls.
483 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
484 Alan Hayward <alan.hayward@arm.com>
485 David Sherwood <david.sherwood@arm.com>
487 * dse.c (store_info, read_info_type): Replace begin and end with
489 (print_range): New function.
490 (set_all_positions_unneeded, any_positions_needed_p)
491 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
493 (record_store): Likewise. Optimize the case in which all positions
495 (get_stored_val): Replace read_begin and read_end with read_offset
497 (replace_read): Update call accordingly.
499 2017-12-15 Bin Cheng <bin.cheng@arm.com>
501 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
502 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
503 (loop_cand::loop_cand): Initialize above members.
504 (loop_cand::supported_operations): Delete.
505 (loop_cand::can_interchange_p): Inline above function.
506 (loop_cand::classify_simple_reduction): Record number of constant
507 initialized simple reductions.
508 (should_interchange_loops): New parameters. Check stmt cost of loops
510 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
511 Update call to should_interchange_loops.
512 (should_interchange_loop_nest): Update call to
513 should_interchange_loops.
515 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
518 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
519 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
520 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
522 2017-12-15 Julia Koval <julia.koval@intel.com>
524 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
525 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
526 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
527 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
528 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
530 2017-12-15 Julia Koval <julia.koval@intel.com>
532 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
533 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
534 * config/i386/sse.md (vaesenc_<mode>): New pattern.
535 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
536 _mm_aesenc_epi128): New intrinsics.
538 2017-12-15 Julia Koval <julia.koval@intel.com>
540 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
541 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
542 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
543 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
544 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
546 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
548 * gimple-ssa-strength-reduction.c (analyze_increments):
549 Distinguish replacement costs for constant strides from those for
552 2017-12-14 Jakub Jelinek <jakub@redhat.com>
554 * var-tracking.c (variable_tracking_main_1): Formatting fix.
556 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
558 * doc/invoke.texi: Document -Wcast-function-type.
559 * recog.h (stored_funcptr): Change signature.
560 * tree-dump.c (dump_node): Avoid warning.
561 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
563 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
566 * gimple-fold.c (get_range_strlen): Add the handling of non-member
569 2017-12-14 David Malcolm <dmalcolm@redhat.com>
571 PR tree-optimization/83312
572 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
573 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
574 "val" param, and to cope with arbitrary basic blocks.
575 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
576 handle NULL_TREE for "val", dropping "bb" param.
577 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
578 drop "bb" param. Handle NULL_TREE for "val".
579 (find_case_label_for_value): Make "switch_stmt" param const.
580 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
582 (vrp_prop::check_all_array_refs): Reimplement as...
583 (check_array_bounds_dom_walker::before_dom_children): ...this new
584 vfunc. Replace linear search through BB block list, excluding
585 those with non-executable in-edges via dominator walk.
587 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
589 * config/arm/arm.opt (mverbose-cost-dump): New option.
590 * config/arm/arm.c (arm_rtx_costs): Use it.
592 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
595 * reload1.c (emit_input_reload_insns): Skip debug markers.
597 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
599 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
603 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
604 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
605 * config/sh/sh.c: Skip debug insns besides notes.
606 * config/sh/sh.md: Likewise.
607 * config/sh/sh_treg_combine.cc: Likewise.
608 * config/sh/sync.md: Likewise.
610 2017-12-14 Tom de Vries <tom@codesourcery.com>
612 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
613 Add item for weak_undefined.
615 2017-12-14 Richard Biener <rguenther@suse.de>
617 PR tree-optimization/67842
618 * sese.h (bb_in_region): Remove #if 0'ed code.
620 2017-12-14 Richard Biener <rguenther@suse.de>
622 PR tree-optimization/83326
623 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
624 may_be_zero parameter and handle it by not marking the first
625 peeled copy as not exiting the loop.
626 (try_peel_loop): Likewise.
627 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
628 to handle the case of constant or zero iterations and perform
629 loop header copying on-the-fly.
631 2017-12-14 Richard Biener <rguenther@suse.de>
633 PR tree-optimization/83418
634 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
635 Instead of asserting we don't get unfolded comparisons deal with
638 2017-12-14 Jakub Jelinek <jakub@redhat.com>
641 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
642 basic blocks. Assert debug bind insns don't appear outside of bbs,
643 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
646 PR tree-optimization/83198
647 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
648 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
649 value if arg is a REAL_CST with incompatible type.
651 2017-12-14 Sudakshina Das <sudi.das@arm.com>
652 Bin Cheng <bin.cheng@arm.com>
655 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
657 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
660 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
662 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
663 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
665 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
666 Alan Hayward <alan.hayward@arm.com>
667 David Sherwood <david.sherwood@arm.com>
669 * poly-int.h: New file.
670 * poly-int-types.h: Likewise.
671 * coretypes.h: Include them.
672 (POLY_INT_CONVERSION): Define.
673 * target.def (estimated_poly_value): New hook.
674 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
675 * doc/tm.texi: Regenerate.
676 * doc/poly-int.texi: New file.
677 * doc/gccint.texi: Include it.
678 * doc/rtl.texi: Describe restrictions on subreg modes.
679 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
680 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
681 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
682 * targhooks.h (default_estimated_poly_value): Declare.
683 * targhooks.c (default_estimated_poly_value): New function.
684 * target.h (estimated_poly_value): Likewise.
685 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
686 (wi::unary_traits): Delete.
687 (wi::binary_traits::signed_shift_result_type): Define for
688 offset_int << HOST_WIDE_INT, etc.
689 (generic_wide_int::operator <<=): Define for all types and use
690 wi::lshift instead of <<.
691 (wi::hwi_with_prec): Add a default constructor.
692 (wi::ints_for): New class.
693 (operator <<): Define for all wide-int types.
694 (operator /): New function.
695 (operator %): Likewise.
696 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
697 (ASSERT_MAYBE_NE_AT): New macros.
699 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
700 Dominik Vogt <vogt@linux.vnet.ibm.com>
703 * emit-rtl.c (init_emit): Remove ??? comment.
704 * explow.c (get_dynamic_stack_size): Take known alignment of stack
705 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
706 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
707 alignment of 3 virtual registers to BITS_PER_WORD.
709 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
711 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
713 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
714 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
716 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
719 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
721 2017-12-13 Jakub Jelinek <jakub@redhat.com>
723 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
724 with the exception of debug begin stmt markers appear before
728 * final.c (rest_of_handle_final): Call variable_tracking_main only
729 if !flag_var_tracking.
731 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
732 Jakub Jelinek <jakub@redhat.com>
736 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
737 labels debug stmts that can only appear after labels.
739 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
741 PR rtl-optimization/82398
742 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
743 EXPR_USEFULNESS in priority comparison.
745 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
747 PR rtl-optimization/83393
748 * combine.c (move_deaths): If reg_stat points to a too new insn in
749 last_death, do not use it: find the proper insn instead.
751 2017-12-12 Jeff Law <law@redhat.com>
753 PR tree-optimization/83298
754 PR tree-optimization/83362
755 PR tree-optimization/83383
756 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
757 push_value_range a public interface. Add new argument to
758 record_ranges_from_stmt.
759 * gimple-ssa-evrp-analyze.c
760 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
761 Update comments. Handle recording temporary equivalences.
762 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
763 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
764 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
765 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
766 gimple-ssa-evrp-analyze.h.
767 (record_temporary_equivalences_from_phis): Add new argument. When
768 the PHI arg is an SSA_NAME, set the result's range to the range
770 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
772 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
773 Pass it down to children as needed.
774 (thread_outgoing_edges): Likewise.
775 (thread_across_edge): Likewise. Push/pop range state as needed.
776 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
778 2017-12-12 Julia Koval <julia.koval@intel.com>
780 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
781 (PTA_CANNONLAKE): Remove PTA_CLWB.
783 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
786 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
787 (vcondv2div2df): Likewise.
788 (vconduv2dfv2di): Likewise.
790 2017-12-12 Jakub Jelinek <jakub@redhat.com>
792 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
793 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
794 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
795 ATTR_MATHFN_FPROUNDING_ERRNO.
797 2017-12-12 Richard Biener <rguenther@suse.de>
799 PR tree-optimization/83385
800 * graphite-scop-detection.c (get_order, order): Remove.
801 (bb_to_rpo): New global.
803 (build_scops): Sort pbbs in RPO order.
805 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
807 * combine.c (simplify_set): Do not transform subregs to zero_extends
808 if the destination mode is a vector mode.
810 2017-12-12 Jakub Jelinek <jakub@redhat.com>
812 PR tree-optimization/83359
813 * tree-cfg.h (fold_loop_internal_call): Declare.
814 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
815 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
816 (find_loop_dist_alias): New function.
817 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
818 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
821 PR tree-optimization/80631
822 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
823 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
824 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
825 hardcoding zero as the value if COND_EXPR is never true. For
826 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
827 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
828 hardcoding MAX_EXPR as the reduction operation.
829 (is_nonwrapping_integer_induction): Allow negative step.
830 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
831 vect_create_epilog_for_reduction, if no value is suitable, don't
832 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
834 2017-12-12 Richard Biener <rguenther@suse.de>
836 PR tree-optimization/81889
837 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
838 range info from the non-wrapping IV instead of just the range
841 2017-12-12 Julia Koval <julia.koval@intel.com>
843 * config.gcc: Add vaesintrin.h.
844 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
846 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
847 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
849 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
850 * config/i386/immintrin.h: Include vaesintrin.h.
851 * config/i386/sse.md (vaesdec_<mode>): New pattern.
852 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
853 _mm_aesdec_epi128): New intrinsics.
855 2017-12-12 Julia Koval <julia.koval@intel.com>
857 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
858 OPTION_MASK_ISA_VAES_UNSET): New.
859 (ix86_handle_option): Handle -mvaes.
860 * config/i386/cpuid.h: Define bit_VAES.
861 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
862 * config/i386/i386-c.c (__VAES__): New.
863 * config/i386/i386.c (ix86_target_string): Add -mvaes.
864 (ix86_valid_target_attribute_inner_p): Ditto.
865 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
866 * config/i386/i386.opt: Add -mvaes.
867 * doc/invoke.texi: Ditto.
869 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
871 * debug.h (gcc_debug_hooks): Add inline_entry.
872 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
873 * debug.c (do_nothing_debug_hooks): Likewise.
874 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
875 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
876 (dwarf2_lineno_debug_hooks): Likewise.
878 * common.opt (gstatement-frontiers): New, setting
879 debug_nonbind_markers_p.
880 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
881 * toplev.c (process_options): Autodetect value for debug statement
883 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
884 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
886 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
887 markers. Integrate source bind into debug stmt expand loop.
888 (pass_expand::execute): Check debug marker limit. Avoid deep
889 TER and expand debug locations for debug bind insns only.
890 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
892 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
893 * final.c (reemit_insn_block_notes): Take current block from
894 nonbind markers. Declare note where it's first set.
895 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
896 begin stmt markers if enabled.
897 (notice_source_line): Handle nonbind markers. Fail if their
898 location is unknown or that of builtins.
899 (rest_of_handle_final): Convert begin stmt markers to notes if
900 var-tracking didn't run.
901 (rest_of_clean_state): Skip begin stmt markers.
902 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
904 * function.c (allocate_struct_function): Set begin_stmt_markers.
905 * function.h (struct function): Add debug_marker_count counter
906 and debug_nonbind_markers flag.
907 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
908 * gimple-low.c (lower_function_body): Adjust
909 debug_nonbind_markers.
910 (lower_stmt): Drop or skip gimple debug stmts.
911 (lower_try_catch): Skip debug stmts.
912 * gimple.c (gimple_build_debug_begin_stmt): New.
913 (gimple_copy): Increment debug_marker_count if copying one.
914 * gimple.h (gimple_build_debug_begin_stmt): Declare.
915 * gimplify.c (rexpr_location): New.
916 (rexpr_has_location): New.
917 (warn_switch_unreachable_r): Handle gimple debug stmts.
918 (shortcut_cond_r): Call expr_location.
920 (find_goto_label): New.
921 (shortcut_cond_expr): Call expr_has_location, expr_location, and
923 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
925 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
926 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
927 (LANG_HOOKS_INITIALIZER): ... this.
928 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
929 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
931 (update_ebb_live_info): Skip debug insn markers.
932 * lra.c (debug_insn_static_data): Rename to...
933 (debug_bind_static_data): ... this.
934 (debug_marker_static_data): New.
935 (lra_set_insn_recog_data): Select one of the above depending
937 (lra_update_isn_regno_info): Don't assume debug insns have
939 (push_insns): Skip debug insns.
940 * lto-streamer-in.c (input_function): Drop debug stmts
941 depending on active options. Adjust debug_nonbind_markers.
942 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
943 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
944 begin stmt marker notes.
945 (print_insn): Likewise.
946 * recog.c (extract_insn): Recognize rtl for debug markers.
947 * rtl.def (DEBUG_MARKER): New.
948 * tree-inline.c: Include params.h.
949 (remap_gimple_stmt): Handle nonbind markers.
950 (maybe_move_debug_stmts_to_successors): Likewise.
951 (copy_debug_stmt): Likewise.
952 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
953 markers regardless of no side effects.
954 (tsi_link_before): Don't update container's side effects when adding
956 (tsi_link_after): Likewise.
957 (expr_first): Skip begin stmt markers.
958 (expr_last): Likewise.
959 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
960 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
961 Disregard nonbind markers.
962 * tree.c (make_node_stat): Don't set side effects for begin stmt
964 (build1_stat): Likewise.
965 * tree.def (DEBUG_BEGIN_STMT): New.
966 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
967 * var-tracking.c (delete_debug_insns): Renamed to...
968 (delete_vta_debug_insns): ... this.
969 (reemit_marker_as_note): New.
970 (vt_initialize): Reemit markers.
971 (delete_vta_debug_insns): Likewise.
972 (vt_debug_insns_local): Reemit or delete markers.
973 (variable_tracking_main_1): Likewise.
974 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
975 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
976 (gimple_debug_nonbind_marker_p): New.
977 (gimple_build_debug_bind): Adjust.
978 (gimple_build_debug_begin_stmt): New.
979 * doc/invoke.texi (max-debug-marker-count): New param.
980 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
981 (debug_parameter_ref, debug_marker): New.
982 (NOTE_INSN_BEGIN_STMT): New.
983 (DEBUG_INSN): Describe begin stmt markers.
985 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
986 without debug insns, we wouldn't, but clean up debug insns
987 after a control flow insn nevertheless.
989 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
990 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
991 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
992 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
993 (rtl_tidy_fallthru_edge): Likewise.
994 (rtl_verify_fallthru): Likewise.
995 (rtl_verify_bb_layout): Likewise.
996 (skip_insns_after_block): Likewise.
997 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
998 * dwarf2out.c: Include print-rtl.h.
999 (dwarf2out_next_real_insn): New.
1000 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
1001 Dump debug binds in asm comments.
1002 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
1003 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
1004 callers to use gsi_start_nondebug_bb instead.
1005 (gsi_after_labels): Skip gimple debug stmts.
1006 (gsi_start_nondebug): New.
1007 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
1008 (proper_loop_form_for_interchange): Adjust.
1009 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
1010 * gimple.h (gimple_seq_last_nondebug_stmt): New.
1011 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
1012 (collect_fallthrough_labels): Likewise.
1013 (should_warn_for_implicit_fallthrough): Likewise.
1014 (warn_implicit_fallthrough_r): Likewise.
1015 (expand_FALLTHROUGH_r): Likewise.
1016 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
1017 (graphite_copy_stmts_from_block): Skip nonbind markers.
1018 * haifa-sched.c (sched_extend_bb): Skip debug insns.
1019 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
1020 * jump.c (clean_barriers): Skip debug insns.
1021 * omp-expand.c (expand_parallel_call): Skip debug insns.
1022 (expand_task_call): Likewise.
1023 (remove_exit_barrier): Likewise.
1024 (expand_omp_taskreg): Likewise.
1025 (expand_omp_for_init_counts): Likewise.
1026 (expand_omp_for_generic): Likewise.
1027 (expand_omp_for_static_nochunk): Likewise.
1028 (expand_omp_for_static_chunk): Likewise.
1029 (expand_omp_simd): Likewise.
1030 (expand_omp_taskloop_for_outer): Likewise.
1031 (expand_omp_taskloop_for_inner): Likewise.
1032 (expand_oacc_for): Likewise.
1033 (expand_omp_sections): Likewise.
1034 (expand_omp_single): Likewise.
1035 (expand_omp_synch): Likewise.
1036 (expand_omp_atomic_load): Likewise.
1037 (expand_omp_atomic_store): Likewise.
1038 (expand_omp_atomic_fetch_op): Likewise.
1039 (expand_omp_atomic_pipeline): Likewise.
1040 (expand_omp_atomic_mutex): Likewise.
1041 (expand_omp_target): Likewise.
1042 (grid_expand_omp_for_loop): Likewise.
1043 (grid_expand_target_grid_body): Likewise.
1044 (build_omp_regions_1): Likewise.
1045 * omp-low.c (check_combined_parallel): Skip debug stmts.
1046 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
1047 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
1048 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
1050 * tree-cfg.c (make_blobs_1): Skip debug stmts.
1051 (make_edges): Likewise.
1052 (cleanup_dead_labels): Likewise.
1053 (gimple_can_merge_blocks_p): Likewise.
1054 (stmt_starts_bb_p): Likewise.
1055 (gimple_block_label): Likewise.
1056 (gimple_redirect_edge_and_branch): Likewise.
1057 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
1059 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
1061 * tree-pretty-print (print_declaration): Omit initializer in slim
1063 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
1065 (eliminate_unnecessary_stmts): Stabilize block removal order.
1066 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
1067 * var-tracking.c (get_first_insn): New.
1068 (vt_emit_notes): Call it.
1069 (vt_initialize): Walk any insns before the first BB.
1070 (delete_debug_insns): Likewise.
1072 * gimple.h (enum gimple_debug_subcode): Add
1073 GIMPLE_DEBUG_BEGIN_STMT.
1074 (gimple_debug_begin_stmt_p): New.
1075 (gimple_debug_nonbind_marker_p): New.
1076 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
1077 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
1078 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
1079 * insn-notes.def (BEGIN_STMT): New.
1080 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
1081 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
1082 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
1083 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
1084 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
1085 (INSN_DEBUG_MARKER_KIND): New.
1086 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
1087 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
1088 (INSN_VAR_LOCATION_PTR): New.
1089 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
1091 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
1092 deps only if debug bind insns are enabled.
1093 (pass_expand::execute): Avoid deep TER and expand
1094 debug locations for debug bind insns only.
1095 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
1096 debug stmts special handling down to debug bind stmts.
1097 * combine.c (try_combine): Narrow debug insns special handling
1098 down to debug bind insns.
1099 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
1100 Narrow debug insns preexisting special handling down to debug
1102 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
1103 handling down to debug bind insns.
1104 * function.c (instantiate_virtual_regs): Skip debug markers,
1105 adjust handling of debug binds.
1106 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
1107 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
1108 * haifa-sched.c (schedule_insn): Narrow special handling of debug
1109 insns to debug bind insns.
1110 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
1111 special handling of debug stmts to debug bind stmts.
1112 * ipa-split.c (split_function): Likewise.
1113 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
1114 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
1116 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
1117 * regrename.c (build_def_use): Likewise.
1118 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
1119 (pass_cprop_hardreg): Narrow special casing of debug insns to
1121 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
1122 * reload1.c (reload): Likewise.
1123 * sese.c (sese_insert_phis_for_liveouts): Narrow special
1124 casing of debug stmts to debug bind stmts.
1125 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
1126 * ssa-iterators.h (num_imm_uses): Likewise.
1127 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
1128 debug stmts to debug bind stmts.
1129 * tree-inline.c (tree_function_versioning): Narrow special casing
1130 of debug stmts to debug bind stmts.
1131 * tree-loop-distribution.c (generate_loops_for_partition):
1132 Narrow special casing of debug stmts to debug bind stmts.
1133 * tree-sra.c (analyze_access_subtree): Narrow special casing
1134 of debug stmts to debug bind stmts.
1135 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
1136 stmts to debug bind stmts.
1137 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
1138 casing of debug stmts to debug bind stmts.
1139 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
1140 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
1141 casing of debug stmts to debug bind stmts.
1142 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
1144 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
1145 debug stmts to debug bind stmts.
1146 (gimple_replace_ssa_lhs): Likewise.
1147 (insert_debug_temp_for_var_def): Likewise.
1148 (insert_debug_temps_for_defs): Likewise.
1149 (reset_debug_uses): Likewise.
1150 * tree-ssanames.c (release_ssa_name_fn): Likewise.
1151 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
1152 (adjust_debug_stmts): Likewise.
1153 (adjust_phi_and_debug_stmts): Likewise.
1154 (vect_do_peeling): Likewise.
1155 * tree-vect-loop.c (vect_transform_loop): Likewise.
1156 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
1157 * var-tracking.c (adjust_mems): Narrow special casing of debug
1158 insns to debug bind insns.
1159 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
1160 (compute_bb_dataflow, vt_find_locations): Likewise.
1161 (vt_expand_loc, emit_notes_for_changes): Likewise.
1162 (vt_init_cfa_base): Likewise.
1163 (vt_emit_notes): Likewise.
1164 (vt_initialize): Likewise.
1165 (vt_finalize): Likewise.
1167 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
1168 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
1169 (next_nonnote_nondebug_insn_bb): New.
1170 (prev_nonnote_nondebug_insn_bb): New.
1171 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
1172 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
1173 (prev_nonnote_nondebug_insn_bb): Declare.
1174 (next_nonnote_nondebug_insn_bb): Declare.
1175 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
1176 * cfgrtl.c (get_last_bb_insn): Likewise.
1177 * lra.c (push_insns): Likewise.
1179 2017-12-11 David Malcolm <dmalcolm@redhat.com>
1182 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
1184 * selftest.c (selftest::test_runner::test_runner): New ctor.
1185 (selftest::test_runner::~test_runner): New dtor.
1186 * selftest.h (class selftest::test_runner): New class.
1188 2017-12-11 Carl Love <cel@us.ibm.com>
1190 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
1191 vec_extract_fp32_from_shortl]): Add #defines.
1192 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
1193 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
1194 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
1195 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
1196 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
1197 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
1198 * doc/extend.texi: Add documentation for the added builtins.
1200 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
1202 PR rtl-optimization/80693
1203 PR rtl-optimization/81019
1204 PR rtl-optimization/81020
1205 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
1206 are not mentioned in i3. Place the REG_UNUSED note on i2,
1207 possibly modified to REG_DEAD, if it did not originate in i3.
1209 2017-12-11 Jakub Jelinek <jakub@redhat.com>
1211 * recog.c (store_data_bypass_p_1): New function.
1212 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
1213 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
1215 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
1217 PR rtl-optimization/83361
1218 * ifcvt.c (if_convert): Call fixup_partitions.
1220 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
1222 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1223 early folding of splat_u{8,16,32}.
1225 2017-12-11 Jakub Jelinek <jakub@redhat.com>
1227 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
1228 output_operand_lossage first argument with capital letter.
1229 (aarch64_override_options): Don't start error and sorry first argument
1230 with capital letter.
1232 2017-12-11 Andi Kleen <ak@linux.intel.com>
1234 PR gcov-profile/83355
1235 * auto-profile.c (string_table::get_index_by_decl): Don't
1236 recurse when abstract origin points to itself.
1238 2017-12-11 Bin Cheng <bin.cheng@arm.com>
1240 PR tree-optimization/83320
1241 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
1242 (prune_datarefs_not_in_loop): Ditto.
1244 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
1246 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
1248 2017-12-10 Jakub Jelinek <jakub@redhat.com>
1250 PR tree-optimization/83337
1251 * gimple-loop-interchange.cc (compute_access_stride): Handle
1252 bitfield DRs properly.
1254 2017-12-09 Jakub Jelinek <jakub@redhat.com>
1256 PR tree-optimization/83338
1257 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
1258 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
1259 vector of unsigned integers to vector of signed integers.
1261 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
1263 PR rtl-optimization/83317
1264 * lra-constraints.c (process_address_1): Add insn code check.
1266 2017-12-08 Michael Matz <matz@suse.de>
1268 Fix PR tree-optimization/83323
1269 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
1270 head-controlled loops and loop BBs.
1271 * common.opt (funroll-and-jam): Remove, instead ...
1272 (floop-unroll-and-jam): ... reuse this option.
1273 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
1274 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
1275 (-floop-unroll-and-jam): ... this option.
1277 2017-12-08 Jakub Jelinek <jakub@redhat.com>
1279 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
1280 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
1281 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
1283 2017-12-08 Richard Biener <rguenther@suse.de>
1286 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
1287 handle accesses outside of zero-sized vars.
1289 2017-12-08 Martin Jambor <mjambor@suse.cz>
1291 PR tree-optimization/83141
1292 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
1293 test for MEM_REFs implicitely changing types with padding. Remove
1295 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
1297 2017-12-08 Jakub Jelinek <jakub@redhat.com>
1299 * config/arc/arc.c (arc_attribute_table): Add exclusions to
1301 * config/avr/avr.c (avr_attribute_table): Likewise.
1302 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
1303 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
1304 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
1305 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
1306 of struct attribute_spec.
1307 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1309 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
1312 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
1314 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
1317 * profile-count.c (profile_count::from_gcov_type): Move from
1318 profile-count.h; handle overflow.
1319 * profile-count.h (profile_count::from_gcov_type): Move offline.
1321 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
1323 PR rtl-optimization/83304
1324 * combine.c (move_deaths): If we do not know where a register died,
1327 2017-12-08 Richard Biener <rguenther@suse.de>
1329 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
1330 Provide -fopt-info-loop feedback when we interchange in a nest.
1332 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
1334 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
1335 for armv6 ARM CPU IDs.
1337 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
1339 * common/config/arm/arm-common.c: Include <algorithm>.
1340 (INCLUDE_VECTOR): Define.
1341 (compare_opt_names): New function.
1342 (arm_rewrite_selected_arch): Only strip out extensions that can be
1343 expressed through -mfpu. Sort the remaining extensions
1346 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
1348 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
1349 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
1350 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
1351 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
1353 2017-06-08 Tristan Gingold <gindold@adacore.com>
1356 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
1357 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
1359 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1361 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
1362 of struct attribute_spec.
1364 2017-12-08 Julia Koval <julia.koval@intel.com>
1366 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
1367 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
1368 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
1369 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
1370 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
1371 _mm_maskz_dpwssds_epi32): Ditto.
1373 2017-12-08 Richard Biener <rguenther@suse.de>
1375 PR tree-optimization/81303
1376 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
1377 conditions try to create a comparison vector type matching
1378 the data vector type.
1379 (vectorizable_condition): Adjust.
1380 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
1381 Leave invariant conditions alone in case we can vectorize those.
1383 2017-12-08 Julia Koval <julia.koval@intel.com>
1385 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
1386 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
1387 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
1388 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
1389 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
1391 2017-12-08 Julia Koval <julia.koval@intel.com>
1393 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
1394 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
1395 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
1396 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
1397 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
1398 _mm_maskz_dpbusds_epi32): New intrinsics.
1400 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
1402 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
1404 (ld<bh>io_signed): Likewise.
1405 (st<bhw>io): Likewise.
1406 * config/nios2/predicates.md (ldstio_memory_operand): Allow
1407 SMALL_INT12 constant integer operand.
1409 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
1411 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
1413 (nios2_large_constant_allowed): ...this. Adjust uses.
1414 (nios2_plus_symbolic_constant_p): Rename to...
1415 (nios2_plus_large_constant_p): ...this. Adjust uses.
1416 (nios2_legitimate_address_p): Correct CONST_INT handling.
1417 (nios2_symbolic_memory_operand_p): Rename to...
1418 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
1419 (nios2_large_constant_p): Check for large constant integers too.
1420 (nios2_split_large_constant): Handle constant integers.
1421 (nios2_split_symbolic_memory_operand): Rename to...
1422 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
1423 (nios2_legitimize_constant_address): Handle constant integers.
1424 (r0rel_constant_p): Handle small constant integers.
1425 (nios2_print_operand_address): Handle r0-relative integer addresses.
1426 * config/nios2/nios2-protos.h: Adjust for renamed functions.
1427 * config/nios2/nios2.md: Adjust for renamed functions.
1429 2017-12-07 Andrew Waterman <andrew@sifive.com>
1431 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
1432 (TARGET_HAVE_SRODATA_SECTION): New define.
1433 (riscv_select_section): New function.
1435 2017-12-08 Joseph Myers <joseph@codesourcery.com>
1436 Alexander Monakov <amonakov@ispras.ru>
1437 Jakub Jelinek <jakub@redhat.com>
1440 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
1442 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1444 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
1445 a bool success value. Don't call output_operand_lossage here.
1446 (aarch64_print_ldpstp_address): Return a bool success value.
1447 (aarch64_print_operand_address): Call output_addr_const if
1448 aarch64_print_address_internal fails.
1449 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
1450 'y'; call output_operand_lossage instead. Call output_operand_lossage
1451 if aarch64_print_ldpstp_address fails.
1453 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1455 * tree-vector-builder.h
1456 (tree_vector_builder::binary_encoded_nelts): Declare.
1457 * tree-vector-builder.c
1458 (tree_vector_builder::binary_encoded_nelts): New function.
1459 * fold-const.c (negate_expr_p): Likewise.
1460 (operand_equal_p, fold_checksum_tree): Likewise.
1461 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
1462 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
1463 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
1464 (uniform_vector_p): Likewise.
1465 * varasm.c (const_hash_1, compare_constant): Likewise.
1466 * tree-ssa-ccp.c: Include tree-vector-builder.h.
1467 (valid_lattice_transition): Operate directly on the VECTOR_CST
1469 * ipa-icf.c: Include tree-vector-builder.h.
1470 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
1471 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
1473 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1475 * tree.c (build_vector): Delete.
1476 * tree.h (build_vector): Make static and move into the self-testing
1479 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1481 * vector-builder.h (vector_builder::derived): New const overload.
1482 (vector_builder::elt): New function.
1483 * tree-vector-builder.h (tree_vector_builder::type): New function.
1484 (tree_vector_builder::apply_step): Declare.
1485 * tree-vector-builder.c (tree_vector_builder::apply_step): New
1487 * gimple-fold.h (tree_vector_builder): Declare.
1488 (gimple_build_vector): Take a tree_vector_builder instead of a
1489 type and vector of elements.
1490 * gimple-fold.c (gimple_build_vector): Likewise.
1491 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
1493 (get_initial_defs_for_reduction): Likewise.
1494 (vectorizable_induction): Likewise.
1496 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1498 * tree-vector-builder.h
1499 (tree_vector_builder::new_binary_operation): Declare.
1500 * tree-vector-builder.c
1501 (tree_vector_builder::new_binary_operation): New function.
1502 * fold-const.c (fold_relational_const): Use it.
1503 (const_binop): Likewise. Check that both input vectors have
1504 the same number of elements, thus excluding things like WIDEN_SUM.
1505 Check whether it is possible to operate directly on the encodings
1508 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1510 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
1511 new_unary_operation, operating only on the encoded elements.
1512 (const_unop): Likewise.
1513 (exact_inverse): Likewise.
1514 (distributes_over_addition_p): New function.
1515 (const_binop): Use tree_vector_builder and new_unary_operation
1516 for combinations of VECTOR_CST and INTEGER_CST. Operate only
1517 on the encoded elements unless the encoding is strided and the
1518 operation does not distribute over addition.
1519 (fold_convert_const): Use tree_vector_builder and
1520 new_unary_operation. Operate only on the encoded elements
1521 for truncating integer conversions, or for non-stepped encodings.
1523 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1525 * config/sparc/sparc.c: Include tree-vector-builder.h.
1526 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
1527 * expmed.c: Include tree-vector-builder.h.
1528 (make_tree): Use tree_vector_builder instead of build_vector.
1529 * fold-const.c: Include tree-vector-builder.h.
1530 (const_binop): Use tree_vector_builder instead of build_vector.
1531 (const_unop): Likewise.
1532 (native_interpret_vector): Likewise.
1533 (fold_vec_perm): Likewise.
1534 (fold_ternary_loc): Likewise.
1535 * gimple-fold.c: Include tree-vector-builder.h.
1536 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
1538 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
1539 (simplify_vector_constructor): Use tree_vector_builder instead
1541 * tree-vect-generic.c: Include tree-vector-builder.h.
1542 (add_rshift): Use tree_vector_builder instead of build_vector.
1543 (expand_vector_divmod): Likewise.
1544 (optimize_vector_constructor): Likewise.
1545 * tree-vect-loop.c: Include tree-vector-builder.h.
1546 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
1547 of build_vector. Explicitly use a stepped encoding for
1549 * tree-vect-slp.c: Include tree-vector-builder.h.
1550 (vect_get_constant_vectors): Use tree_vector_builder instead
1552 (vect_transform_slp_perm_load): Likewise.
1553 (vect_schedule_slp_instance): Likewise.
1554 * tree-vect-stmts.c: Include tree-vector-builder.h.
1555 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
1556 (vect_gen_perm_mask_any): Likewise.
1557 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
1558 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
1560 (build_vector_from_val): Likewise. Explicitly use a duplicate
1563 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
1565 * doc/generic.texi (VECTOR_CST): Describe new representation of
1567 * vector-builder.h: New file.
1568 * tree-vector-builder.h: Likewise.
1569 * tree-vector-builder.c: Likewise.
1570 * Makefile.in (OBJS): Add tree-vector-builder.o.
1571 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
1572 * tree-core.h (tree_base): Add a vector_cst field to the u union.
1573 (tree_vector): Change the number of elements to
1574 vector_cst_encoded_nelts.
1575 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
1576 (VECTOR_CST_ELTS): Delete.
1577 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
1578 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
1579 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
1580 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
1581 (VECTOR_CST_ENCODED_ELT): Likewise.
1582 (vector_cst_encoded_nelts): New function.
1583 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
1584 VECTOR_CST_NELTS_PER_PATTERN as arguments.
1585 (vector_cst_int_elt, vector_cst_elt): Declare.
1586 * tree.c: Include tree-vector-builder.h.
1587 (tree_code_size): Abort if passed VECTOR_CST.
1588 (tree_size): Update for new VECTOR_CST layout.
1589 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
1590 VECTOR_CST_NELTS_PER_PATTERN as arguments.
1591 (build_vector): Use tree_vector_builder.
1592 (vector_cst_int_elt, vector_cst_elt): New functions.
1593 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
1594 encoded elements and then create the vector in the canonical form.
1595 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
1596 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
1597 (tree_c_tests): Call test_vector_cst_patterns.
1598 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
1600 (hash_tree): Likewise.
1601 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
1602 (streamer_write_tree_header): Likewise.
1603 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
1604 (streamer_alloc_tree): Likewise. Update call to make_vector.
1605 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
1607 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1609 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
1610 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
1611 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
1613 2017-12-07 Bin Cheng <bin.cheng@arm.com>
1614 Richard Biener <rguenther@suse.de>
1616 PR tree-optimization/81303
1617 * Makefile.in (gimple-loop-interchange.o): New object file.
1618 * common.opt (floop-interchange): Reuse the option from graphite.
1619 * doc/invoke.texi (-floop-interchange): Ditto. New document for
1620 -floop-interchange and mention it for -O3.
1621 * opts.c (default_options_table): Enable -floop-interchange at -O3.
1622 * gimple-loop-interchange.cc: New file.
1623 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
1624 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
1625 * passes.def (pass_linterchange): New pass.
1626 * timevar.def (TV_LINTERCHANGE): New time var.
1627 * tree-pass.h (make_pass_linterchange): New declaration.
1628 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
1629 interchange. Record IV before/after increment in new parameters.
1630 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
1631 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
1633 (check_reduction_path): ...New function here.
1634 * tree-vectorizer.h (check_reduction_path): New declaration.
1636 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
1639 PR rtl-optimization/80818
1640 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
1641 always early clobbered.
1642 * lra-lives.c (process_bb_lives): Check input hard regs for early
1643 clobbered non-operand hard reg.
1645 2017-12-07 Jakub Jelinek <jakub@redhat.com>
1648 * tree-cfg.c (verify_gimple_assign_binary): Don't require
1649 types_compatible_p, just that TYPE_MODE is the same.
1651 2017-12-07 Martin Sebor <msebor@redhat.com>
1654 * attribs.c (empty_attribute_table): Initialize new member of
1655 struct attribute_spec.
1656 (decl_attributes): Add argument. Handle mutually exclusive
1657 combinations of attributes.
1658 (selftests::test_attribute_exclusions): New function.
1659 (selftests::attribute_c_tests): Ditto.
1660 * attribs.h (decl_attributes): Add default argument.
1661 * selftest.h (attribute_c_tests): Declare.
1662 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
1663 * tree-core.h (attribute_spec::exclusions, exclude): New type and
1665 * doc/extend.texi (Common Function Attributes): Update const and pure.
1666 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
1667 of struct attribute_spec.
1668 * config/arc/arc.c (arc_attribute_table): Same.
1669 * config/arm/arm.c (arm_attribute_table): Same.
1670 * config/avr/avr.c ( avr_attribute_table): Same.
1671 * config/bfin/bfin.c (bfin_attribute_table): Same.
1672 * config/cr16/cr16.c (cr16_attribute_table): Same.
1673 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
1674 * config/h8300/h8300.c (h8300_attribute_table): Same.
1675 * config/i386/i386.c (ix86_attribute_table): Same.
1676 * config/ia64/ia64.c (ia64_attribute_table): Same.
1677 * config/m32c/m32c.c (m32c_attribute_table): Same.
1678 * config/m32r/m32r.c (m32r_attribute_table): Same.
1679 * config/m68k/m68k.c (m68k_attribute_table): Same.
1680 * config/mcore/mcore.c (mcore_attribute_table): Same.
1681 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
1682 * config/mips/mips.c (mips_attribute_table): Same.
1683 * config/msp430/msp430.c (msp430_attribute_table): Same.
1684 * config/nds32/nds32.c (nds32_attribute_table): Same.
1685 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
1686 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
1687 * config/rl78/rl78.c (rl78__attribute_table): Same.
1688 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
1689 * onfig/rx/rx.c (rx_attribute_table): Same.
1690 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
1691 * config/sh/sh.c (sh_attribute_table): Same.
1692 * config/sparc/sparc.c (sparc_attribute_table): Same.
1693 * config/spu/spu.c (spu_attribute_table): Same.
1694 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
1695 * config/v850/v850.c (v850_attribute_table): Same.
1696 * config/visium/visium.c (visium_attribute_table): Same.
1698 2017-12-07 Tamar Christina <tamar.christina@arm.com>
1701 * config/arm/arm.c (INCLUDE_STRING): Define.
1702 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
1703 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
1706 2017-12-07 Michael Matz <matz@suse.de>
1708 Add unroll and jam pass
1710 * gimple-loop-jam.c: New file.
1711 * Makefile.in (OBJS): Add gimple-loop-jam.o.
1712 * common.opt (funroll-and-jam): New option.
1713 * opts.c (default_options_table): Add unroll-and-jam at -O3.
1714 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
1715 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
1716 * passes.def: Add pass_loop_jam.
1717 * timevar.def (TV_LOOP_JAM): Add.
1718 * tree-pass.h (make_pass_loop_jam): Declare.
1719 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
1720 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
1721 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
1722 to flow_loop_tree_node_add.
1723 (duplicate_subloops, copy_loops_to): Append to sibling list.
1724 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
1725 * doc/invoke.texi (-funroll-and-jam): Document new option.
1726 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
1728 2017-12-07 Richard Biener <rguenther@suse.de>
1730 PR tree-optimization/83296
1731 PR tree-optimization/67769
1732 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
1733 flow sensitive info in an unrelated BB.
1734 (value_replacement): Use reset_flow_sensitive_info.
1735 (minmax_replacement): Reset flow sensitive info on the def
1736 we move. Do not reset flow sensitive info in the whole BB
1737 we move the stmt to.
1738 (abs_replacement): Likewise.
1740 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1743 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1744 rs6000_cpu to the given -mcpu=, or to the default processor.
1746 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1748 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
1749 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
1750 instead of rs6000_cpu_attr.
1751 (rs6000_adjust_cost): Ditto.
1752 (is_microcoded_insn): Ditto.
1753 (rs6000_adjust_priority): Ditto.
1754 (rs6000_issue_rate): Ditto.
1755 (rs6000_use_sched_lookahead): Ditto.
1756 (rs6000_use_sched_lookahead_guard): Ditto.
1757 (rs6000_sched_reorder): Ditto.
1758 (force_new_group): Ditto.
1759 * config/rs6000/rs6000.md (cpu attribute): Ditto.
1760 (group_ending_nop): Ditto.
1762 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1764 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
1765 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
1766 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
1767 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
1769 (rs6000_reassociation_width): Ditto.
1770 (rs6000_emit_epilogue): Ditto.
1771 (rs6000_adjust_cost): Ditto.
1772 (is_microcoded_insn): Ditto.
1773 (is_cracked_insn): Ditto.
1774 (rs6000_adjust_priority): Ditto.
1775 (rs6000_sched_reorder): Ditto.
1776 (rs6000_sched_reorder2): Ditto.
1777 (insn_must_be_first_in_group): Ditto.
1778 (insn_must_be_last_in_group): Ditto.
1779 (rs6000_register_move_cost): Ditto.
1780 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
1783 2017-12-07 Julia Koval <julia.koval@intel.com>
1785 * config.gcc: Add vaesintrin.h.
1786 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
1788 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
1789 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
1791 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1792 * config/i386/immintrin.h: Include vaesintrin.h.
1793 * config/i386/sse.md (vaesdec_<mode>): New pattern.
1794 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
1795 _mm_aesdec_epi128): New intrinsics.
1797 2017-12-06 David Malcolm <dmalcolm@redhat.com>
1799 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
1800 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
1801 c-family/c-spellcheck.cc.
1802 (best_macro_match::best_macro_match): Likewise.
1804 (struct edit_distance_traits<cpp_hashnode *>): Move to
1805 c-family/c-spellcheck.h.
1806 (class best_macro_match): Likewise.
1808 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1810 PR tree-optimization/83293
1811 * gimple-ssa-strength-reduction.c (insert_initializers): Use
1812 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
1813 might insert into empty bb.
1816 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
1817 simplify for plus with :c added, and pointer_plus without that.
1818 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
1819 with undefined overflow and the conversion is not widening,
1820 perform negation in utype and only convert to type afterwards.
1821 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
1822 simplify for plus with :c added, and pointer_plus without that.
1823 If type is integral with undefined overflow and the conversion is
1824 not widening, perform minus in utype and only convert to type
1825 afterwards. Move the last pointer_diff_expr simplify into the
1828 2017-12-06 Martin Sebor <msebor@redhat.com>
1830 PR tree-optimization/82646
1831 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
1832 strncpy, not maxlen.
1834 2017-12-06 Martin Sebor <msebor@redhat.com>
1836 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
1839 PR tree-optimization/83075
1840 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
1841 strncat/strncpy don't change length of source string.
1843 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
1846 2017-11-29 Martin Aberg <maberg@gaisler.com>
1848 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1849 to prevent b2bst errata sequence.
1850 (sqrtdf2_fix): Likewise.
1852 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1854 PR tree-optimization/81945
1855 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
1856 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
1857 to dest_cfun has orig_loop_num set, either remap it to the new
1858 loop number if the loop got moved too, or clear it.
1860 2017-12-05 Steve Ellcey <sellcey@cavium.com>
1862 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
1864 (thunderx2t99_nothing): New insn reservation.
1865 (thunderx2t99_mrs): New insn reservation.
1866 (thunderx2t99_multiple): New insn reservation.
1867 (thunderx2t99_alu_basi): Add bfx to reservation.
1868 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
1870 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1873 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
1875 2017-12-05 Bin Cheng <bin.cheng@arm.com>
1877 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
1878 tree-ssa-pre.c::remove_dead_inserted_code.
1879 * tree-ssa-dce.h: New file.
1880 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
1881 (remove_dead_inserted_code): Move and rename to function
1882 tree-ssa-dce.c::simple_dce_from_worklist.
1883 (pass_pre::execute): Update use.
1885 2017-12-05 Richard Biener <rguenther@suse.de>
1887 PR tree-optimization/83277
1888 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
1889 to code-gen liveout vars.
1891 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
1893 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
1894 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
1895 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
1898 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
1900 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
1902 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
1903 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
1906 2017-12-05 Richard Biener <rguenther@suse.de>
1908 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
1909 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
1910 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
1911 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
1913 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1915 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
1917 2017-12-05 Martin Liska <mliska@suse.cz>
1918 Jakub Jelinek <jakub@redhat.com>
1920 * doc/invoke.texi: Document the options.
1921 * flag-types.h (enum sanitize_code): Add
1922 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1923 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
1924 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1925 * opts.c: Define new sanitizer options.
1926 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
1927 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
1929 2017-12-05 Julia Koval <julia.koval@intel.com>
1931 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
1932 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
1933 (ix86_handle_option): Handle -mavx512vnni.
1934 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
1935 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
1936 * config/i386/i386-c (__AVX512VNNI__): New.
1937 * config/i386/i386.c (ix86_target_string): Handle new option.
1938 (ix86_valid_target_attribute_inner_p): Handle new option.
1939 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
1940 * config/i386/i386.opt (mavx512vnni): New option.
1942 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1945 * athlon.md: Disable for generic.
1946 * haswell.md: Enable for generic.
1947 * i386.c (ix86_sched_init_global): Add core hooks for generic.
1948 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
1950 (ix86_adjust_cost): Move generic to haswell path.
1952 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
1954 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
1955 instead of MEM_P in a couple more places. Fix formatting issues.
1957 2017-12-04 Jim Wilson <jimw@sifive.com>
1959 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
1960 instead of GP_REG_LAST-1.
1961 (riscv_adjust_libcall_cfi_prologue): Likewise.
1962 (riscv_adjust_libcall_cri_epilogue): Likewise.
1963 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
1966 2017-12-04 Luis Machado <luis.machado@linaro.org>
1968 * ipa-pure-const.c (check_decl): Add missing newline.
1969 (state_from_flags): Likewise.
1971 2017-12-04 Jeff Law <law@redhat.com>
1973 PR tree-optimizatin/78496
1974 * gimple-ssa-evrp-analyze.h
1975 (evrp_range_analyzer::get_vr_values): Simplify.
1976 * gimple-ssa-evrp-analyze.c: Corresponding changes.
1977 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
1978 and gimple-ssa-evrp-analyze.h.
1979 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
1980 (simplify_stmt_for_jump_threading): Copy a blob of code from
1981 tree-vrp.c to use ranges to simplify statements.
1982 (dom_opt_dom_walker::before_dom_children): Call
1983 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
1984 (dom_opt_dom_walker::after_dom_children): Similarly for
1985 evrp_range_analyzer::leave.
1986 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
1989 * gimple-ssa-evrp-analyze.c
1990 (evrp_range_analyzer::extract_range_from_stmt): Always use
1991 vr_values::update_value_range so preexisting range info is
1992 medged with new range info, even if the new range is VR_VARYING.
1994 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1996 * combine.c: Adjust comment.
1997 (use_crosses_set_p): Delete.
1998 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
1999 (try_combine): Ditto.
2001 2017-12-04 Richard Biener <rguenther@suse.de>
2003 PR tree-optimization/83255
2004 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
2005 Re-add zero-iteration check.
2007 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
2009 PR rtl-optimization/83245
2010 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
2011 hard registers as earlyclobber, also if not in an asm.
2013 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
2017 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2020 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
2021 rs6000_cpu based on cpu_index, not tune_index.
2023 2017-12-04 Richard Biener <rguenther@suse.de>
2025 PR tree-optimization/83238
2026 * graphite-scop-detection.c (scop_detection::merge_sese): Make
2027 code match comment, rejecting invalid SESE regions.
2029 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
2031 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
2032 require base operand is a REG_POINTER prior to reload on targets
2033 with non-equivalent space registers.
2035 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
2037 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
2038 (update_specialized_profile): Fix updating of counts.
2039 (perhaps_add_new_callers): Likewise.
2041 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
2044 * x86-tune.def: Remove obsolette FIXMEs.
2045 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
2046 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
2047 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
2049 (X86_TUNE_PAD_RETURNS): Disable for generic.
2051 2017-12-02 Jakub Jelinek <jakub@redhat.com>
2053 PR tree-optimization/83170
2054 PR tree-optimization/83241
2055 * gimple-ssa-store-merging.c
2056 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
2057 gimple_vuse (ins_stmt) in case it has changed.
2058 (imm_store_chain_info::output_merged_store): Likewise.
2060 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
2064 * tree-cfg.c (pass_warn_function_return::execute): Handle
2065 __builtin_ubsan_handle_missing_return like __builtin_unreachable
2066 with BUILTINS_LOCATION.
2070 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
2071 is BLKmode for vector field with vector raw mode, use TYPE_MODE
2072 instead of DECL_MODE.
2074 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
2075 last argument to rtx pointer.
2076 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
2077 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
2078 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
2079 depending on the chosen ISAs.
2080 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
2081 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
2082 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
2084 * config/i386/sse.md (mov<mode>_internal): Likewise.
2085 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
2087 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2089 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
2091 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2093 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
2094 parameters from prototype.
2095 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
2096 parameters. Don't print a comment.
2097 (emit_fusion_gpr_load): Adjust.
2098 (emit_fusion_load_store): Adjust.
2099 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
2100 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
2101 comment on the second line.
2103 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2106 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
2107 rs6000_cpu based on cpu_index, not tune_index.
2109 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2111 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
2112 which_alternative instead of which_alternative + 1.
2113 (output_asm_insn): Print an extra tab if the template is short.
2115 2017-12-01 Jim Wilson <jimw@sifive.com>
2117 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
2119 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
2120 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
2122 * doc/tm.texi: Regenerate.
2124 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
2127 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
2128 whether we can allocate pseudos before trying to fix an address.
2129 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
2130 memory address is indexed or indirect.
2131 (floatuns_<mode>si2_hw2): Likewise.
2133 2017-12-01 Jason Merrill <jason@redhat.com>
2135 * Makefile.in (TAGS): Add c-family/*.cc.
2137 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
2139 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
2140 (call_value_insn): Likewise.
2141 (sibcall_insn): Likewise.
2142 (sibcall_value_insn): Likewise.
2143 (movsi_aarch64): Likewise.
2144 (movdi_aarch64): Likewise.
2145 (add_losym_): Likewise.
2146 (ldr_got_small_): Likewise.
2147 (ldr_got_small_sidi): Likewise.
2148 (ldr_got_small_28k_): Likewise.
2149 (ldr_got_small_28k_sidi): Likewise.
2150 * config/aarch64/aarch64.c (aarch64_print_address_internal):
2151 Move output_addr_const to symbolic case. Add error check.
2153 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2155 * config/s390/predicates.md (plus16_Q_operand): New predicate.
2156 * config/s390/s390.md: Disable MVC merging peephole if it would
2157 disable operand forwarding.
2158 (new peephole2): Split MVCs if it would turn them into up to 2
2161 2017-12-01 Richard Biener <rguenther@suse.de>
2163 PR tree-optimization/83232
2164 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
2165 detection of same access. Instead of breaking the group here
2166 do not consider the duplicate. Add comment explaining real fix.
2168 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
2170 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
2172 2017-12-01 Sudakshina Das <sudi.das@arm.com>
2174 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
2176 2017-12-01 Jakub Jelinek <jakub@redhat.com>
2178 * function.h (struct function): Remove cilk_frame_decl,
2179 is_cilk_function and calls_cilk_spawn fields.
2180 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
2182 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
2183 cilk_elemental field.
2184 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
2185 * target.def: Adjust comment.
2186 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
2187 Don't test cilk_elemental.
2189 PR tree-optimization/83233
2190 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
2191 bswap_stat name for the struct.
2194 * tree.h (SWITCH_BREAK_LABEL_P): Define.
2195 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
2196 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
2197 SWITCH_BREAK_LABEL_P set on the label.
2198 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
2199 added for default case if it was missing and not all cases covered.
2200 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
2201 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
2203 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
2204 to avoid -Wimplicit-fallthrough warning.
2205 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
2206 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
2208 PR tree-optimization/83221
2209 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
2211 (init_reassoc): Formatting fix.
2214 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
2215 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
2219 * tree-cfg.c: Include asan.h.
2220 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
2222 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2224 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
2226 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2228 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
2229 vpcmpeqd instruction.
2231 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2233 * config/i386/i386.c (standard_sse_constant_opcode): Fix
2234 registers type for 128bit mode.
2236 2017-11-30 Jakub Jelinek <jakub@redhat.com>
2238 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
2240 * gengtype-state.c (read_state_pair): Likewise.
2241 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
2242 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
2243 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
2244 * ifcvt.c (noce_try_store_flag_constants): Likewise.
2245 * tree-ssa-ccp.c (ccp_finalize): Likewise.
2246 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
2247 * builtins.c (fold_builtin_3): Likewise.
2248 * graphite-scop-detection.c
2249 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
2250 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
2253 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
2256 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
2257 parameter. Return true if ignore_decl_rtl_set_p is true and other
2258 conditions are satisfied.
2259 * asan.h (asan_protect_global): Add new parameter.
2260 * varasm.c (categorize_decl_for_section): Pass true as second parameter
2261 to asan_protect_global calls.
2263 2017-11-30 Jim Wilson <jimw@sifive.com>
2265 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
2266 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
2267 -mstrict-align, add info on default value. Delete redundant lines for
2268 -mabi. Add missing -mexplicit-relocs docs.
2270 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2272 * config/arc/arc.md (trap): New pattern.
2274 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2276 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
2277 instruction to end into a delay slot.
2278 * config/arc/arc.md (cond_delay_insn): Check if the instruction
2279 can be placed into a delay slot against reg_note.
2281 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2283 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
2284 labels number of usages.
2286 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2288 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
2290 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
2292 2017-11-30 Jakub Jelinek <jakub@redhat.com>
2295 * internal-fn.c (expand_mul_overflow): Optimize unsigned
2296 multiplication by power of 2 constant into two shifts + comparison.
2298 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
2301 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
2303 2017-11-30 Richard Biener <rguenther@suse.de>
2305 PR tree-optimization/83202
2306 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2307 allow_peel argument and guard peeling.
2308 (canonicalize_loop_induction_variables): Likewise.
2309 (canonicalize_induction_variables): Pass false.
2310 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
2311 peeling from cunrolli.
2313 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
2315 * combine.c (try_combine): Print a message to dump file whenever
2316 I0, I1, or I2 cannot be combined into I3.
2318 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
2320 PR rtl-optimization/83156
2321 PR rtl-optimization/82621
2322 * combine.c (try_combine): Don't split an I2 if one of the dests is
2323 set again before I3. Allow unused dests.
2325 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
2327 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
2329 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
2331 PR rtl-optimization/80818
2332 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
2333 recursively. Use insn code for clobber.
2334 (lra_set_insn_recog_data): Pass the new arg to
2335 collect_non_operand_hard_regs.
2336 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
2338 (lra_update_insn_regno_info): Pass insn to
2339 add_regs_to_insn_regno_info.
2341 2017-11-29 Jim Wilson <jimw@sifive.com>
2342 Andrew Waterman <andrew@sifive.com>
2344 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
2345 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
2346 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
2347 use SINGLE_SHIFT_COST.
2348 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
2349 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
2351 2017-11-29 Julia Koval <julia.koval@intel.com>
2353 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
2354 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
2355 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
2356 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
2357 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
2358 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
2359 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
2360 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
2361 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
2362 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
2363 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
2364 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
2365 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
2366 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
2367 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
2368 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
2369 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
2370 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
2371 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
2372 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
2373 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
2374 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
2375 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
2376 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
2377 __builtin_ia32_vpshldv_v2di_mask,
2378 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
2379 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
2380 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
2382 2017-11-29 Julia Koval <julia.koval@intel.com>
2384 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
2385 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
2386 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
2387 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
2388 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
2389 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
2390 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
2391 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
2392 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
2393 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
2394 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
2395 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
2396 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
2397 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
2398 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
2399 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
2400 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
2401 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
2402 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2403 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
2404 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
2406 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2408 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
2409 movsi_pic_gotdata_op instruction as a load for the UT699 errata
2412 2017-11-29 Martin Aberg <maberg@gaisler.com>
2414 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
2415 to prevent b2bst errata sequence.
2416 (sqrtdf2_fix): Likewise.
2418 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2420 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
2421 (div_sqrt_insn_p): New function.
2422 (sparc_do_work_around_errata): Insert NOP instructions to
2423 prevent sequences that could trigger the TN-0013 errata for
2424 certain LEON3 processors.
2425 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
2426 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
2427 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
2428 (in_branch_delay): Prevent div and sqrt in delay slot if
2430 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
2432 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2434 * config/sparc/sparc.c (atomic_insn_p): New function.
2435 (sparc_do_work_around_errata): Insert NOP instructions to
2436 prevent sequences that could trigger the TN-0010 errata for
2438 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
2439 instruction referable in atomic_insns_p.
2441 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2443 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
2444 (atomic_compare_and_swap_leon3_1): Likewise.
2447 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2449 * config/sparc/sparc.c (fpop_insn_p): New function.
2450 (sparc_do_work_around_errata): Insert NOP instructions to
2451 prevent sequences that could trigger the TN-0012 errata for
2453 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
2454 * config/sparc/sparc.md (fix_gr712rc): New attribute.
2455 (in_branch_annul_delay): Prevent floating-point instructions
2456 in delay slot of annulled integer branch.
2458 2017-11-29 Richard Biener <rguenther@suse.de>
2460 PR tree-optimization/83202
2461 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
2463 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
2464 nodes vectorized to the same stmts multiple times.
2465 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
2466 (vect_analyze_slp_instance): Adjust.
2467 (scalar_stmts_to_slp_tree_map_t): New typedef.
2468 (vect_schedule_slp_instance): Add a map recording the SLP node
2469 representing the vectorized stmts for a set of scalar stmts.
2470 Avoid code-generating redundancies.
2471 (vect_schedule_slp): Allocate map and pass it down.
2473 2017-11-29 Nathan Sidwell <nathan@acm.org>
2476 * tree.c (build_complex_type): Fix canonicalization. Only fill in
2479 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
2481 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
2482 cases for printing LDP/STP memory addresses.
2483 (aarch64_print_address_internal): Renamed from
2484 aarch64_print_operand_address, added parameter, add Pmode check.
2485 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
2486 (aarch64_print_operand_address): Indirect to
2487 aarch64_print_address_internal.
2488 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
2489 'y' operand output specifier.
2491 2017-11-29 Jakub Jelinek <jakub@redhat.com>
2494 * tree.c (build_simple_mem_ref_loc): Handle
2495 get_addr_base_and_unit_offset returning a MEM_REF.
2498 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
2501 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
2504 2017-11-28 David Malcolm <dmalcolm@redhat.com>
2506 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
2508 (layout::move_to_column): Likewise.
2510 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2513 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
2514 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
2516 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2517 Martin Jambor <mjambor@suse.cz>
2520 * tree.h (expr_type_first_operand_type_p): Declare
2521 * tree.c (expr_type_first_operand_type_p): New function.
2522 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
2523 (ipa_value_from_jfunc): Adjust declaration.
2524 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
2525 Use it as result type for arithmetics, unless it is NULL in which case
2526 be more conservative.
2527 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
2528 ipa_get_jf_pass_through_result.
2529 (propagate_vals_across_pass_through): Likewise.
2530 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
2531 is to propagate_vals_across_pass_through.
2532 (propagate_constants_across_call): Pass PARM_TYPE to
2533 propagate_scalar_across_jump_function.
2534 (find_more_scalar_values_for_callers_subset): Pass parameter type to
2535 ipa_value_from_jfunc.
2536 (cgraph_edge_brings_all_scalars_for_node): Likewise.
2537 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
2538 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
2539 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
2540 target_type, pass it to ipa_value_from_jfunc.
2541 (update_indirect_edges_after_inlining): Pass parameter type to
2542 try_make_edge_direct_simple_call.
2544 2017-11-28 Jeff Law <law@redhat.com>
2546 * gimple-ssa-evrp-analyze.c
2547 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
2548 refine ranges if scev_initialized_p returns true.
2549 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
2551 2017-11-28 Julia Koval <julia.koval@intel.com>
2553 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
2554 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
2555 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
2556 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
2557 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
2558 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
2559 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
2560 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
2561 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
2562 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
2563 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
2564 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
2565 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
2566 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
2567 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
2568 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
2569 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
2570 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
2571 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
2572 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
2573 __builtin_ia32_vpshrd_v2di_mask): New builtins.
2574 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
2576 2017-11-28 Julia Koval <julia.koval@intel.com>
2578 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
2579 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
2580 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
2581 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
2582 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
2583 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
2584 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
2585 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
2586 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
2587 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
2588 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
2589 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
2590 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
2591 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
2592 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
2593 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
2594 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
2595 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
2596 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
2597 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
2598 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
2599 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
2600 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
2601 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
2602 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
2603 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
2604 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
2605 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
2606 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
2607 __builtin_ia32_vpshld_v2di_mask): New builtins.
2608 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2609 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
2611 2017-11-28 Richard Biener <rguenther@suse.de>
2613 PR tree-optimization/80776
2614 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
2616 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
2618 (evrp_range_analyzer::record_ranges_from_incoming_edges):
2619 If the incoming edge is an effective fallthru because the other
2620 edge only reaches a __builtin_unreachable () then record ranges
2621 derived from the controlling condition in SSA info.
2622 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
2623 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
2625 2017-11-28 Olivier Hainque <hainque@adacore.com>
2627 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
2628 on mingw build hosts.
2630 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2633 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
2635 2017-11-28 Richard Biener <rguenther@suse.de>
2638 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
2639 copies generated from memcpy use a character array as reference
2642 2017-11-28 Julia Koval <julia.koval@intel.com>
2643 Sebastian Peryt <sebastian.peryt@intel.com>
2645 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
2646 c-family/c-cilkplus.o, c-family/array-notation-common.o,
2647 cilk-common.o, cilk.h, cilk-common.c): Remove.
2649 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
2650 * builtins.c (is_builtin_name): Remove cilkplus condition.
2651 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
2652 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
2653 cilk-builtins.def, cilkplus.def): Remove.
2654 * cif-code.def (CILK_SPAWN): Remove.
2655 * cilk-builtins.def: Delete.
2656 * cilk-common.c: Ditto.
2658 * cilkplus.def: Ditto.
2659 * config/darwin.h (fcilkplus): Delete.
2660 * cppbuiltin.c: Ditto.
2661 * doc/extend.texi: Remove cilkplus doc.
2662 * doc/generic.texi: Ditto.
2663 * doc/invoke.texi: Ditto.
2664 * doc/passes.texi: Ditto.
2665 * gcc.c (fcilkplus): Remove.
2666 * gengtype.c (cilk.h): Remove.
2667 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
2669 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
2671 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
2673 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
2674 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
2675 cilkplus conditions.
2676 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
2677 inline_read_section): Ditto.
2678 * ipa-inline-analysis.c (cilk.h): Remove.
2679 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
2680 * lto-wrapper.c (merge_and_complain, append_compiler_options,
2681 append_linker_options): Remove condition for fcilkplus.
2682 * lto/lto-lang.c (cilk.h): Remove.
2683 (lto_init): Remove condition for fcilkplus.
2684 * omp-expand.c (expand_cilk_for_call): Delete.
2685 (expand_omp_taskreg, expand_omp_for_static_chunk,
2686 expand_omp_for): Remove cilkplus
2688 (expand_cilk_for): Delete.
2689 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
2690 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
2691 execute_lower_omp, diagnose_sb_0): Ditto.
2692 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
2693 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2694 * tree-nested.c: Ditto.
2695 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
2696 (dump_generic_node): Ditto.
2697 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2698 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
2700 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
2702 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2704 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
2705 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
2706 for vector mode and !TARGET_SIMD.
2708 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2710 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
2712 * tree.h (SWITCH_LABELS): Remove.
2713 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
2714 assert SWITCH_BODY is non-NULL.
2715 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
2717 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
2719 PR tree-optimization/80788
2720 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
2721 has TREE_OVERFLOW set, call drop_tree_overflow.
2723 2017-11-28 Richard Biener <rguenther@suse.de>
2725 PR tree-optimization/83158
2726 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
2728 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
2731 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
2732 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
2734 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
2736 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
2737 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
2739 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2742 * varasm.c (compare_constant): Take the mode of the constants into
2743 account when comparing floating point constants.
2745 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
2747 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
2748 from explicit instantiation of debug_helper.
2749 * vec.h (DEFINE_DEBUG_VEC): Ditto.
2751 2017-11-27 Richard Biener <rguenther@suse.de>
2753 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
2756 2017-11-27 Richard Biener <rguenther@suse.de>
2758 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
2759 (type_hash_canon): Also clear int_cst_hash_table entry for
2761 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
2763 2017-11-27 Tamar Christina <tamar.christina@arm.com>
2765 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2766 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2767 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2769 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2771 * hash-map.h (gt_cleare_cache): Avoid UB.
2773 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
2775 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
2776 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
2777 (decide_unroll_runtime_iterations): Likewise.
2778 (decide_unroll_stupid): Likewise.
2780 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2783 * config/i386/i386.md: Add a loop with incssp.
2785 2017-11-27 Martin Jambor <mjambor@suse.cz>
2787 PR tree-optimization/81248
2788 * tree-sra.c (splice_param_accesses): Remove size check.
2789 (decide_one_param_reduction): Fix size check.
2790 * gimple-pretty-print.c (dump_profile): Silence warning.
2791 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
2793 2017-11-27 Jakub Jelinek <jakub@redhat.com>
2796 * dbxout.c (lastlineno): New variable.
2797 (dbx_debug_hooks): Use dbxout_switch_text_section as
2798 switch_text_section debug hook.
2799 (dbxout_function_end): Switch to current_function_section
2800 rather than function_section. If crtl->has_bb_partition,
2801 output just one N_FUN, depending on in_cold_section_p.
2802 (dbxout_source_line): Remember last lineno in lastlineno.
2803 (dbxout_switch_text_section): New function.
2804 (dbxout_function_decl): Adjust dbxout_block caller.
2805 (dbx_block_with_cold_children): New function.
2806 (dbxout_block): Return true if any LBRAC/RBRAC have been
2807 emitted. Use dbx_block_with_cold_children at depth == 0
2808 in second partition. Add PARENT_BLOCKNUM argument, pass
2809 it optionally adjusted to children. Output LBRAC/RBRAC
2810 around recursive call only if the block is in the current
2811 partition, if not and anything was output, emit empty
2813 * final.c (final_scan_insn): Compute cold_function_name
2814 before calling switch_text_section debug hook. Call
2815 that hook even if dwarf2out_do_frame if not emitting
2819 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
2820 TREE_READONLY decls.
2822 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2824 PR rtl-optimization/82488
2825 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
2827 2017-11-26 Julia Koval <julia.koval@intel.com>
2829 * config/i386/i386.c (processor_target_table): Add skylake_cost for
2831 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
2834 2017-11-26 Julia Koval <julia.koval@intel.com>
2836 * config/i386/driver-i386.c (host_detect_local_cpu):
2837 Detect skylake-avx512.
2839 2017-11-26 Julia Koval <julia.koval@intel.com>
2841 * config.gcc: Add -march=cannonlake.
2842 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
2843 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
2844 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
2845 (PTA_CANNONLAKE): New.
2846 (processor_target_table): Add cannonlake.
2847 (ix86_option_override_internal): Ditto.
2848 (fold_builtin_cpu): Ditto.
2849 (get_builtin_code_for_version): Handle cannonlake.
2850 (M_INTEL_COREI7_CANNONLAKE): New.
2851 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
2852 * doc/invoke.texi: Add -march=cannonlake.
2854 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
2856 * plugin.c (add_new_plugin): Use platform-specific library extensions.
2857 (try_init_one_plugin): Alternative implementation for MinGW.
2858 * Makefile.in (plugin_implib): New.
2859 (gengtype-lex.c): Fix broken AIX workaround.
2860 * configure: Regenerate.
2861 * doc/plugins.texi: Document support for MinGW.
2863 2017-11-25 Jakub Jelinek <jakub@redhat.com>
2865 PR rtl-optimization/81553
2866 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
2867 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
2868 is a shift where C1 has different mode than the whole shift, use C1's
2869 mode for MULT rather than the shift's mode.
2872 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
2873 builtins not enabled in the currently selected ISA.
2875 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
2877 PR tree-optimization/71026
2878 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
2879 (insert_reciprocals): Change to insert reciprocals before a division
2880 by a square and to insert the square of a reciprocal.
2881 (execute_cse_reciprocals_1): Change to consider division by a square.
2882 (register_division_in): Add importance parameter.
2884 2017-11-24 Richard Biener <rguenther@suse.de>
2886 PR tree-optimization/82402
2887 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
2888 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
2890 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
2892 * match.pd (0-ptr): New transformation.
2894 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
2897 * ipa-inline.c (inline_small_functions): Set current badnes correctly
2898 when skipping checking.
2900 2017-11-24 Richard Biener <rguenther@suse.de>
2902 PR tree-optimization/83128
2903 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
2904 (vn_reference_lookup_3): Likewise.
2906 2017-11-24 Jakub Jelinek <jakub@redhat.com>
2909 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
2910 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
2913 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
2914 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
2915 check for the latter with gimple_call_builtin_p. Do not handle
2916 BUILT_IN_STPNCPY_CHK which is not a pass through call.
2918 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
2920 * config/arm/arm_neon.h: Fix pragma GCC push_options before
2923 2017-11-23 Julia Koval <julia.koval@intel.com>
2925 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
2926 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
2927 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
2928 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
2929 _mm512_maskz_expandloadu_epi16): New intrinsics.
2930 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
2931 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
2932 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
2933 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
2934 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
2935 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
2936 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
2937 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
2938 _mm256_maskz_expandloadu_epi8): New intrinsics.
2939 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
2940 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
2941 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
2942 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
2943 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
2944 * config/i386/sse.md (VI248_VLBW): New iterator.
2945 (expand<mode>_mask, expand<mode>_maskz): New patterns.
2947 2017-11-23 Julia Koval <julia.koval@intel.com>
2949 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
2950 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
2951 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
2952 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
2953 _mm512_mask_compressstoreu_epi16): New.
2954 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
2955 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
2956 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
2957 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
2958 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
2959 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
2960 _mm256_mask_compressstoreu_epi8): New.
2961 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
2962 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
2963 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
2964 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
2965 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
2966 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
2967 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
2968 __builtin_ia32_compresshi128_mask,
2969 __builtin_ia32_compressstoreuqi512_mask,
2970 __builtin_ia32_compressstoreuhi512_mask,
2971 __builtin_ia32_compressstoreuqi256_mask,
2972 __builtin_ia32_compressstoreuqi128_mask,
2973 __builtin_ia32_compressstoreuhi256_mask,
2974 __builtin_ia32_compressstoreuhi128_mask): New builtins.
2975 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
2977 (ix86_expand_special_args_builtin): Handle new types.
2978 (s4fma_expand): Handle new builtin array.
2979 * config/i386/immintrin.h: Include new headers.
2980 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
2981 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
2983 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2985 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
2986 fixes. Declare temp and g variables at the top in order to avoid
2987 {} in most of the cases.
2989 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
2991 * match.pd (ptr-0): New transformation.
2993 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2995 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
2996 (struct addr_mode_cost_table): New.
2997 (struct tune_params): Add field addr_mode_costs.
2998 * config/arm/arm.c (generic_addr_mode_costs): New.
2999 (arm_slowmul_tune): Initialise addr_mode_costs field.
3000 (arm_fastmul_tune): Likewise.
3001 (arm_strongarm_tune): Likewise.
3002 (arm_xscale_tune): Likewise.
3003 (arm_9e_tune): Likewise.
3004 (arm_marvell_pj4_tune): Likewise.
3005 (arm_v6t2_tune): Likewise.
3006 (arm_cortex_tune): Likewise.
3007 (arm_cortex_a8_tune): Likewise.
3008 (arm_cortex_a7_tune): Likewise.
3009 (arm_cortex_a15_tune): Likewise.
3010 (arm_cortex_a35_tune): Likewise.
3011 (arm_cortex_a53_tune): Likewise.
3012 (arm_cortex_a57_tune): Likewise.
3013 (arm_exynosm1_tune): Likewise.
3014 (arm_xgene1_tune): Likewise.
3015 (arm_cortex_a5_tune): Likewise.
3016 (arm_cortex_a9_tune): Likewise.
3017 (arm_cortex_a12_tune): Likewise.
3018 (arm_cortex_a73_tune): Likewise.
3019 (arm_v7m_tune): Likewise.
3020 (arm_cortex_m7_tune): Likewise.
3021 (arm_v6m_tune): Likewise.
3022 (arm_fa726te_tune): Likewise.
3023 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
3025 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
3027 * config/arm/arm.c (arm_mem_costs): New function.
3028 (arm_rtx_costs_internal): Use arm_mem_costs.
3030 2017-11-23 Mark Wielaard <mark@klomp.org>
3032 * dwarf2out.c (init_sections_and_labels): Use generation to create
3033 unique ranges_section_label and ranges_base_label. Return generation.
3034 (output_rnglists): Add generation argument. Use generation to create
3035 unique ranges labels.
3036 (dwarf2out_finish): Get generation from init_sections_and_labels
3037 and pass generation to output_rnglists.
3039 2017-11-23 Mike Stump <mikestump@comcast.net>
3040 Eric Botcazou <ebotcazou@adacore.com>
3042 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
3043 * cfgloop.h (struct loop): Add unroll field.
3044 * function.h (struct function): Add has_unroll bitfield.
3045 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
3046 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
3047 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
3048 (pass_rtl_unroll_loops::gate): Likewise.
3049 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
3050 for which loop->unroll==1.
3051 (decide_unroll_constant_iterations): Use note for consistency and
3052 take loop->unroll into account. Return early if loop->unroll is set.
3053 Fix thinko in existing test.
3054 (decide_unroll_runtime_iterations): Use note for consistency and
3055 take loop->unroll into account.
3056 (decide_unroll_stupid): Likewise.
3057 * lto-streamer-in.c (input_cfg): Read loop->unroll.
3058 * lto-streamer-out.c (output_cfg): Write loop->unroll.
3059 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
3061 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
3062 (print_loop): Print loop->unroll if set.
3063 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
3064 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
3065 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
3067 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
3068 loop->unroll is set and smaller than the trip count. Otherwise bypass
3069 entirely the heuristics if loop->unroll is set. Remove dead note.
3070 Fix off-by-one bug in other note.
3071 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
3072 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
3074 (tree_unroll_loops_completely): Make static.
3075 (pass_complete_unroll::execute): Use correct type for variable.
3076 (pass_complete_unrolli::execute): Fix formatting.
3077 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
3079 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
3081 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
3082 enable when TARGET_PREFER_AVX128 is set.
3084 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3086 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
3088 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
3089 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
3091 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3093 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
3094 frequencies. Fix estimation of aggregate parameters.
3096 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3098 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
3101 2017-11-23 Tom de Vries <tom@codesourcery.com>
3103 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
3105 2017-11-23 Tom de Vries <tom@codesourcery.com>
3107 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
3110 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
3113 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
3114 sibcall_value_pcrel_fdpic): Use local variable instead of
3116 (calli_tbr_rel): Add missing operand 2.
3117 (call_valuei_tbr_rel): Add missing operand 3.
3119 2017-11-23 Jakub Jelinek <jakub@redhat.com>
3122 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
3123 bitpos/bitsize covering the whole destination, use store_expr only if
3124 the complex mode is the same. Otherwise, use expand_normal and if
3125 it returns CONCAT, subreg each part separately instead of trying to
3126 subreg the whole result.
3128 2017-11-23 Richard Biener <rguenther@suse.de>
3130 PR tree-optimization/23094
3131 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
3132 come here from walking over backedges in the first iteration.
3133 (vn_reference_lookup_3): Skip clobbers that store the same value.
3135 2017-11-23 Richard Biener <rguenther@suse.de>
3137 PR tree-optimization/81403
3138 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
3139 a block we need a leader relative to.
3140 (phi_translate_1): For nary processing require a leader from
3141 get_representative_for given we run expression simplification
3142 using match-and-simplify. Remove previous fix.
3144 2017-11-22 Jeff Law <law@redhat.com>
3146 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
3147 Use new method allocate_value_range rather than accessing the
3148 vrp_value_range_pool data member directly.
3149 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
3150 to use extract_range_from_stmt method to avoid need for
3151 extract_range_from_assignment method.
3152 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
3153 method rather than setting values_propgated data member directly.
3154 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
3155 and values propagated data members and extract_range_from_assignment
3156 method. Reorder private data members to conform to standards.
3157 Add new methods set_lattice_propagation_complete and
3158 allocate_value_range.
3160 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
3162 PR rtl-optimization/83030
3163 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
3164 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
3165 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
3166 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
3167 CROSSING_JUMP_P flag is already set before setting it.
3168 * cfgrtl.c (fixup_partition_crossing): Likewise.
3169 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
3172 2017-11-22 Jakub Jelinek <jakub@redhat.com>
3174 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
3175 Handle the case where both arguments are using gen_const_vec_series.
3177 2017-11-22 David Malcolm <dmalcolm@redhat.com>
3180 * pretty-print.c (pp_format): Move quoting implementation to
3181 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
3182 to pass address of "quote" local.
3183 (pp_begin_quote): New function.
3184 (pp_end_quote): New function.
3185 * pretty-print.h (printer_fn): Convert penultimate param from bool
3187 (pp_begin_quote): New decl.
3188 (pp_end_quote): New decl.
3189 * tree-diagnostic.c (default_tree_printer): Convert penultimate
3190 param from bool to bool *.
3191 * tree-diagnostic.h (default_tree_printer): Likewise.
3193 2017-11-22 Jeff Law <law@redhat.com>
3195 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
3196 Initialize vr_values.
3197 (evrp_range_analyzer::try_find_new_range): Call methods attached to
3198 vr_values via vr_values class instance rather than delegators.
3199 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
3200 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
3201 (evrp_range_analyzer::push_value_range): Likewise.
3202 (evrp_range_analyzer::pop_value_range): Likewise.
3203 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
3204 most delegators. Those remaining are exposed as public interfaces.
3205 Make vr_values a pointer and private.
3206 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
3208 (evrp_range_analyzer::get_vr_value): New method.
3209 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
3210 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
3211 (class evrp_dom_walker): Attach evrp_folder class, initialize
3212 it in the ctor. Remove temporary delegators.
3213 (evrp_dom_walker::before_dom_children): Call methods in attached
3214 evrp_range_analyzer class via class instance pointer. Use
3215 free value_range_constant_singleton to remove need for
3216 op_with_constant_singleton_value delegator method. Do not
3217 create a vrp_prop class instance for every call! Narrow
3218 scope of a couple variables.
3219 (evrp_dom_walker::cleanup): Call methods in attached
3220 evrp_range_analyzer class via class instance pointer.
3221 * vr-values.h (class vr_values): Privatize many methods and
3224 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3226 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
3227 * cfgexpand.c (expand_debug_expr): Remove handling for them.
3228 * expr.c (expand_expr_real_2): Likewise.
3229 * fold-const.c (const_unop): Likewise.
3230 * optabs-tree.c (optab_for_tree_code): Likewise.
3231 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3232 * tree-inline.c (estimate_operator_cost): Likewise.
3233 * tree-pretty-print.c (dump_generic_node): Likewise.
3234 (op_code_prio): Likewise.
3235 (op_symbol_code): Likewise.
3236 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
3237 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
3238 * internal-fn.c (direct_internal_fn_optab): New function.
3239 (direct_internal_fn_array, direct_internal_fn_supported_p
3240 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
3241 * fold-const-call.c (fold_const_reduction): New function.
3242 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
3244 * tree-vect-loop.c: Include internal-fn.h.
3245 (reduction_code_for_scalar_code): Rename to...
3246 (reduction_fn_for_scalar_code): ...this and return an internal
3248 (vect_model_reduction_cost): Take an internal_fn rather than
3250 (vect_create_epilog_for_reduction): Likewise. Build calls rather
3252 (vectorizable_reduction): Use internal functions rather than tree
3253 codes for the reduction operation. Update calls to the functions
3255 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
3256 Use calls to internal functions rather than REDUC tree codes.
3257 * config/aarch64/aarch64-simd.md: Update comment accordingly.
3259 2017-11-22 Olivier Hainque <hainque@adacore.com>
3261 * config/vxworks.c (vxworks_override_options): Pick default
3262 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
3263 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
3264 DWARF_GNAT_ENCODINGS_DEFAULT.
3265 * config/vxworksae.h: Likewise.
3267 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
3269 PR tree-optimization/83104
3270 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
3273 2017-11-22 Marek Polacek <polacek@redhat.com>
3274 H.J. Lu <hongjiu.lu@intel.com>
3275 Jason Merrill <jason@redhat.com>
3280 * calls.c (initialize_argument_information): Call
3281 warn_parameter_passing_abi target hook.
3282 (store_one_arg): Use 0 for empty record size. Don't push 0 size
3283 argument onto stack.
3284 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
3285 * common.opt: Update -fabi-version description.
3286 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
3287 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
3289 (ix86_is_empty_record): New function.
3290 (ix86_warn_parameter_passing_abi): New function.
3291 (TARGET_EMPTY_RECORD_P): Redefine.
3292 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
3293 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
3294 * doc/tm.texi: Regenerated.
3295 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
3296 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
3297 * dwarf2out.c (get_ultimate_context): Move to tree.c.
3298 * explow.c (hard_function_value): Call arg_int_size_in_bytes
3299 instead of int_size_in_bytes.
3300 * expr.c (copy_blkmode_to_reg): Likewise.
3301 * function.c (aggregate_value_p): Return 0 for empty types.
3302 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
3304 (locate_and_pad_parm): Call arg size_in_bytes instead
3306 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
3307 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
3308 * target.def (empty_record_p, warn_parameter_passing_abi): New target
3310 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
3311 (std_gimplify_va_arg_expr): Skip empty records. Call
3312 arg_size_in_bytes instead size_in_bytes.
3313 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
3314 * tree-core.h (tree_type_common): Add empty_flag.
3315 (tree_decl_common): Update comments.
3316 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
3318 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
3319 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
3321 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
3322 * tree.c (default_is_empty_type): New function.
3323 (default_is_empty_record): New function.
3324 (arg_int_size_in_bytes): New function.
3325 (arg_size_in_bytes): New function.
3326 (get_ultimate_context): New function.
3327 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
3328 TRANSLATION_UNIT_WARN_EMPTY_P.
3329 (default_is_empty_record, arg_int_size_in_bytes,
3330 arg_size_in_bytes, get_ultimate_context): Declare.
3332 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
3334 * config/arm/arm.c (cmse_clear_registers): New function.
3335 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
3336 code by call to cmse_clear_registers.
3337 (cmse_nonsecure_entry_clear_before_return): Likewise.
3339 2017-11-22 Tamar Christina <tamar.christina@arm.com>
3341 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
3342 (vdot_s32, vdotq_s32): New.
3343 (vdot_lane_u32, vdotq_lane_u32): New.
3344 (vdot_lane_s32, vdotq_lane_s32): New.
3347 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3350 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
3351 for unsigned values with fewer HWIs than the precision.
3352 (test_overflow): New function.
3353 (wide_int_cc_tests): Call it.
3355 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3357 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
3358 has the same mode class as Pmode.
3360 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
3362 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
3363 padding_bits_to_clear_ptr.
3364 (cmse_nonsecure_entry_clear_before_return): Likewise.
3366 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
3368 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
3369 auto_sbitap instead of integer bitfield to control register needing
3372 2017-11-22 Jakub Jelinek <jakub@redhat.com>
3374 PR tree-optimization/83044
3375 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
3376 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
3377 that need the upper bound. Subtract offset from
3378 get_addr_base_and_unit_offset only if positive and subtract it
3379 before division by eltsize rather than after it.
3382 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
3383 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
3384 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
3388 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
3389 expand_binop with *mul_widen_optab, make sure at least one of the
3390 operands doesn't have VOIDmode.
3393 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
3395 PR rtl-optimization/82044
3396 PR tree-optimization/82042
3397 * dse.c (record_store): Check for overflow.
3398 (check_mem_read_rtx): Properly check for overflow if width == -1, call
3399 add_wild_read instead of clear_rhs_from_active_local_stores on
3400 overflow and log it into dump_file.
3402 2017-11-22 Richard Biener <rguenther@suse.de>
3404 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
3405 fake edges to exit when looking for a place to insert.
3406 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
3408 (insert_into_preds_of_block): Commit edge insertion immediately,
3409 assert that doesn't require new BBs.
3410 (fini_pre): Release expressions.
3411 (pass_pre::execute): Shuffle things around a bit, if the fn
3412 is too large do not compute AVAIL either as this is really the
3415 2017-11-22 Richard Biener <rguenther@suse.de>
3417 PR tree-optimization/83089
3418 * tree-if-conv.c (pass_if_conversion::execute): If anything
3419 changed reset SCEV and free the number of iteration estimates.
3421 2017-11-21 Martin Sebor <msebor@redhat.com>
3423 PR tree-optimization/82945
3424 * calls.h (warn_nonstring_bound): Remove unused function.
3426 2017-11-21 Martin Sebor <msebor@redhat.com>
3428 PR tree-optimization/82945
3429 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
3430 * calls.h (maybe_warn_nonstring_arg): Declare new function.
3431 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
3433 (initialize_argument_information): Call maybe_warn_nonstring_arg.
3434 * calls.h (get_attr_nonstring_decl): Declare new function.
3435 * doc/extend.texi (attribute nonstring): Update.
3436 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
3437 get_attr_nonstring_decl and handle it.
3438 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
3439 detection of nul-termination.
3440 (strlen_to_stridx): Change to a pointer.
3441 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
3442 (pass_strlen::execute): Same.
3444 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
3446 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
3447 for the new option -mprefer-vector-width=[none|128|256|512].
3448 * config/i386/i386.c (ix86_target_string): remove old style options
3449 -mprefer-avx256 and make -mprefer-avx128 as alias.
3450 (ix86_option_override_internal): Apply defaults for the
3451 -mprefer-vector-width=[128|256] option.
3452 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
3453 Implement macros to work with -mprefer-vector-width=.
3454 * config/i386/i386.opt: Implemented option
3455 -mprefer-vector-width=[none|128|256|512].
3456 * doc/invoke.texi: Documentation for
3457 -mprefer-vector-width=[none|128|256|512].
3459 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
3461 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
3463 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
3465 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
3467 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
3469 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
3471 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
3472 * expr.c (expand_expr_real_2): Likewise.
3473 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
3474 fold_binary_loc): Likewise.
3475 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
3476 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
3477 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
3478 MINUS_EXPR transformations.
3479 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
3480 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
3481 * tree-inline.c (estimate_operator_cost): Likewise.
3482 * tree-pretty-print.c (dump_generic_node, op_code_prio,
3483 op_symbol_code): Likewise.
3484 * tree-vect-stmts.c (vectorizable_operation): Likewise.
3485 * vr-values.c (extract_range_from_binary_expr): Likewise.
3486 * varasm.c (initializer_constant_valid_p_1): Likewise.
3487 * tree.def: New tree code POINTER_DIFF_EXPR.
3489 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
3491 * config/i386/i386.md (*bswap<mode>2_movbe): Add
3492 integer suffix to movbe mnemonic.
3493 (*bswaphi2_movbe): Ditto.
3494 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
3496 2017-11-21 Jakub Jelinek <jakub@redhat.com>
3499 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
3500 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
3501 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
3503 2017-11-21 Martin Liska <mliska@suse.cz>
3505 * tree-inline.c (expand_call_inline): Remove not needed
3508 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
3509 Jakub Jelinek <jakub@redhat.com>
3512 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
3513 Remove static keyword from f variable.
3515 2017-11-21 Jakub Jelinek <jakub@redhat.com>
3517 PR tree-optimization/83086
3518 * gimple-ssa-store-merging.c
3519 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
3520 rather than n.base_addr.
3522 2017-11-21 Martin Liska <mliska@suse.cz>
3524 PR rtl-optimization/82044
3525 PR tree-optimization/82042
3526 * dse.c (check_mem_read_rtx): Check for overflow.
3528 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
3530 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
3533 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
3535 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
3536 and its corresponding call arg location note.
3538 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
3539 Andrew Burgess <andrew.burgess@embecosm.com>
3541 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
3543 (arc_return_slot_offset): Likewise.
3544 (arc_eh_return_address_location): New declaration.
3545 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
3546 (MUST_SAVE_REGISTER): Add exception handler case.
3547 (MUST_SAVE_RETURN_ADDR): Likewise.
3548 (arc_frame_pointer_required): Likewise.
3549 (arc_frame_pointer_needed): New function.
3550 (arc_compute_frame_size): Changed.
3551 (arc_expand_prologue): Likewise.
3552 (arc_expand_epilogue): Likewise.
3553 (arc_initial_elimination_offset): Likewise.
3554 (arc_return_slot_offset): Delete.
3555 (arc_eh_return_address_location): New function.
3556 (arc_builtin_setjmp_frame_value): Likewise.
3557 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
3558 (EH_RETURN_STACKADJ_RTX): Define.
3559 (EH_RETURN_HANDLER_RTX): Likewise.
3560 * config/arc/arc.md (eh_return): Delete.
3562 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
3564 * print-tree.h (debug_vec_tree): Remove prototype.
3565 * gdbinit.in (pvt): Remove macro.
3567 2017-11-21 Jakub Jelinek <jakub@redhat.com>
3569 PR tree-optimization/83047
3570 * gimple-ssa-store-merging.c
3571 (imm_store_chain_info::output_merged_store): If the loads with the
3572 same vuse are in different basic blocks, for load_gsi pick a load
3573 location that is dominated by the other loads.
3576 * config/i386/i386.c (ix86_memmodel_check): Start
3577 -Winvalid-memory-model diagnostics with lowercase letter.
3580 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
3581 set in_cold_section_p to the partition containing loc_list->first.
3582 When seeing loc_list->last_before_switch node, update secname and
3583 perform range_across_switch second partition handling only after that.
3586 * run-rtl-passes.c: Include debug.h.
3587 (run_rtl_passes): Call debug_hooks->assembly_start.
3588 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
3592 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
3593 OPTAB_DIRECT in calls to expand_simple_binop.
3595 2017-11-20 David Malcolm <dmalcolm@redhat.com>
3598 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
3600 2017-11-20 David Malcolm <dmalcolm@redhat.com>
3603 * spellcheck.h (best_match::blithely_get_best_candidate): New
3606 2017-11-20 Jakub Jelinek <jakub@redhat.com>
3608 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
3609 with lower case letter.
3611 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
3613 * config/i386/i386.md (bswaphi2): New expander.
3614 (*bswaphi2_movbe): New insn pattern.
3615 (bswaphi -> rorhi pepehole2): New peephole pattern.
3617 2017-11-20 Jeff Law <law@redhat.com>
3619 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
3620 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
3621 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
3622 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
3623 gimple-ssa-evrp-analyze.h.
3625 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
3626 set BB_VISITED here.
3627 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
3629 2017-11-20 Bin Cheng <bin.cheng@arm.com>
3631 * tree-predcom.c: Add general comment on Store-Store chains.
3632 (split_data_refs_to_components): Postpone clearing eliminate_store_p
3634 (get_chain_last_ref_at): Rename into...
3635 (get_chain_last_write_at): ...this.
3636 (get_chain_last_write_before_load): New function.
3637 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
3638 CT_STORE_STORE when write reference is added.
3639 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
3640 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
3641 (initialize_root_vars_store_elim_1): Ditto.
3642 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
3643 definition is created.
3644 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
3645 chain by replacing it with dominant stored value.
3647 2017-11-20 Bin Cheng <bin.cheng@arm.com>
3649 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
3651 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
3653 * vr-values.c (extract_range_from_binary_expr): Use a full range
3656 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
3658 * config/arm/arm.md (R4_REGNUM): Define constant.
3659 (nonsecure_call_internal): Remove r4 clobber.
3660 (nonsecure_call_value_internal): Likewise.
3661 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
3662 clobber and resequence match_operands.
3663 (nonsecure_call_value_reg_thumb1_v5): Likewise.
3664 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
3665 (nonsecure_call_value_reg_thumb2): Likewise.
3667 2017-11-20 Jakub Jelinek <jakub@redhat.com>
3669 PR tree-optimization/78821
3670 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
3671 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
3672 to the address of the base rather than the base itself.
3673 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
3674 (find_bswap_or_nop_finalize): New function.
3675 (find_bswap_or_nop): Use it.
3676 (bswap_replace): Return a tree rather than bool, change first
3677 argument from gimple * to gimple_stmt_iterator, allow inserting
3678 into an empty sequence, allow ins_stmt to be NULL - then emit
3679 all stmts into gsi. Fix up MEM_REF address gimplification.
3680 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
3681 (struct store_immediate_info): Add N and INS_STMT non-static
3683 (store_immediate_info::store_immediate_info): Initialize them
3684 from newly added ctor args.
3685 (merged_store_group::apply_stores): Formatting fixes. Sort by
3687 (stmts_may_clobber_ref_p): For stores call also
3688 refs_anti_dependent_p.
3689 (gather_bswap_load_refs): New function.
3690 (imm_store_chain_info::try_coalesce_bswap): New method.
3691 (imm_store_chain_info::coalesce_immediate_stores): Use it.
3692 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
3693 (imm_store_chain_info::output_merged_store): Fail if number of
3694 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
3695 and NOP_EXPR rhs_code.
3696 (pass_store_merging::process_store): Compute n and ins_stmt, if
3697 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
3698 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
3700 (pass_store_merging::execute): Calculate dominators.
3702 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
3703 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
3704 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
3705 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3706 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
3707 class pass_optimize_bswap, bswap_replace,
3708 pass_optimize_bswap::execute): Moved to ...
3709 * gimple-ssa-store-merging.c: ... this file.
3710 Include optabs-tree.h.
3711 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
3712 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3713 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
3714 anonymous namespace, remove static keywords.
3715 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
3716 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
3718 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
3721 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
3723 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
3725 * vec.h (debug_helper): New function.
3726 (DEFINE_DEBUG_VEC): New macro.
3727 * hash-set.h (debug_helper): New function.
3728 (DEFINE_DEBUG_HASH_SET): New macro.
3729 * cfg.c (debug_slim (edge)): New function.
3730 Call DEFINE_DEBUG_VEC for edges.
3731 Call DEFINE_DEBUG_HASH_SET for edges.
3732 * cfghooks.c (debug_slim (basic_block)): New function.
3733 Call DEFINE_DEBUG_VEC for basic blocks.
3734 Call DEFINE_DEBUG_HASH_SET for basic blocks.
3735 * print-tree.c (debug_slim): New function to handle trees.
3736 Call DEFINE_DEBUG_VEC for trees.
3737 Call DEFINE_DEBUG_HASH_SET for trees.
3738 (debug (vec<tree, va_gc>) &): Remove.
3739 (debug (<vec<tree, va_gc>) *): Remove.
3740 * print-rtl.c (debug_slim): New function to handle const_rtx.
3741 Call DEFINE_DEBUG_VEC for rtx_def.
3742 Call DEFINE_DEBUG_VEC for rtx_insn.
3743 Call DEFINE_DEBUG_HASH_SET for rtx_def.
3744 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
3745 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
3746 (debug (vec<rtx_insn *> *ptr): Remove.
3747 (debug_insn_vector): Remove.
3748 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
3750 2017-11-20 Tom de Vries <tom@codesourcery.com>
3752 PR rtl-optimization/82020
3753 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
3754 IF_THEN_ELSE condition.
3756 2017-11-19 Jeff Law <law@redhat.com>
3758 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
3759 of degenerates resulting from ignoring an edge.
3761 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3764 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
3766 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3769 * profile-count.c (profile_count::to_sreal_scale): Fix return value
3770 for uninitialied counts.
3772 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3775 * tree-inline.c (estimate_num_insns): Set to 1 at least.
3777 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3780 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
3782 2017-11-19 Tom de Vries <tom@codesourcery.com>
3784 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
3787 2017-11-19 Tom de Vries <tom@codesourcery.com>
3789 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
3792 2017-11-19 Tom de Vries <tom@codesourcery.com>
3794 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
3797 2017-11-19 Tom de Vries <tom@codesourcery.com>
3799 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
3801 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
3803 2017-11-19 Tom de Vries <tom@codesourcery.com>
3805 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
3807 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
3808 semicolon after MCORE_EXPORT_NAME call.
3810 2017-11-19 Tom de Vries <tom@codesourcery.com>
3813 * vmsdbgout.c (vmsdbgout_early_finish): New function.
3814 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
3816 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
3818 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
3819 (cgraph_edge::clone): Cleanup updating of profile.
3820 * ipa-cp.c (update_profiling_info): Likewise.
3821 * ipa-inline-transform.c (inline_transform): Likewise.
3822 * ipa-inline.c (inline_small_functions): Add missing space to dump.
3823 * ipa-split.c (execute_split_functions): Do not split when function
3825 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
3826 * profile-count.c (profile_count::dump): Add global0.
3827 (profile_count::to_cgraph_frequency): Do not ICE when entry is
3829 (profile_count::to_sreal_scale): Likewise.
3830 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
3831 (profile_count::combine_with_ipa_count): New function.
3832 * profile-count.h (profile_guessed_global0adjusted): New.
3833 (profile_count::adjusted_zero): New.
3834 (profile_count::global0adjusted): New.
3835 (profile_count::combine_with_ipa_count): New.
3836 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
3837 correct profile of return block of split functions.
3838 (copy_cfg_body): Remove unused profile_count.
3839 (copy_body): Likewise.
3840 (expand_call_inline): Update.
3841 (tree_function_versioning): Update.
3843 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
3845 * hash-set.h (hash_set::empty): New.
3846 * tree-ssa-threadbackward.h: Delete.
3847 * tree-ssa-threadbackward.c (class thread_jumps): New.
3848 Move max_threaded_paths into class.
3849 (fsm_find_thread_path): Remove arguments that are now in class.
3850 (profitable_jump_thread_path): Rename to...
3851 (thread_jumps::profitable_jump_thread_path): ...this.
3852 (convert_and_register_jump_thread_path): Rename to...
3853 (thread_jumps::convert_and_register_current_path): ...this.
3854 (check_subpath_and_update_thread_path): Rename to...
3855 (thread_jumps::check_subpath_and_update_thread_path): ...this.
3856 (register_jump_thread_path_if_profitable): Rename to...
3857 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
3858 (handle_phi): Rename to...
3859 (thread_jumps::handle_phi): ...this.
3860 (handle_assignment): Rename to...
3861 (thread_jumps::handle_assignment): ...this.
3862 (fsm_find_control_statement_thread_paths): Rename to...
3863 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
3864 (find_jump_threads_backwards): Rename to...
3865 (thread_jumps::find_jump_threads_backwards): ...this.
3866 Initialize path local data.
3867 (pass_thread_jumps::execute): Call find_jump_threads_backwards
3868 from within thread_jumps class.
3869 (pass_early_thread_jumps::execute): Same.
3871 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3873 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
3875 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3877 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
3878 to canonical location.
3880 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3883 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
3885 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
3887 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
3889 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
3890 rcrt1.o%s/grcrt1.o%s for -static-pie.
3892 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3894 * i386.c (ix86_multiplication_cost, ix86_division_cost,
3895 ix86_shift_rotate_cost): Break out from ...
3896 (ix86_rtx_costs): ... here.
3897 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
3900 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3902 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
3903 when entry block was promoted unlikely.
3904 (estimate_bb_frequencies): Increase frequency scale.
3905 * profile-count.h (profile_count): Export precision info.
3907 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3909 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
3910 disturbin profile of entry block.
3912 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3914 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
3917 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3919 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
3920 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
3922 2017-11-17 Jeff Law <law@redhat.com>
3924 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
3925 from evrp_dom_walker class. Various methods moved into new class.
3926 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
3927 (evrp_range_analyzer::enter): New method.
3928 (evrp_range_analyzer::leave): New method.
3929 (evrp_dom_walker): Remove delegators no longer needed by this class.
3930 Replace vr_values data member with evrp_range_analyzer
3932 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
3933 method extracted from evrp_dom_walker::before_dom_children.
3934 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
3935 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
3937 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
3938 Add private copy constructor and move assignment operators.
3939 Privatize methods and class data where trivially possible.
3940 (evrp_dom_walker::cleanup): New function, extracted from
3941 execute_early_vrp. Simplify access to class data.
3943 * vr-values.h (get_output_for_vrp): Prototype.
3944 * vr-values.c (get_output_for_vrp): New function extracted from
3945 vrp_visit_assignment_or_call and extract_range_from_stmt.
3946 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
3948 2017-11-17 Luis Machado <luis.machado@linaro.org>
3950 * config/aarch64/aarch64.c
3951 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
3952 (qdf24xx_tunings) <autoprefetcher_model>: Set to
3953 tune_params::AUTOPREFETCHER_WEAK.
3955 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3958 * config/arm/arm.c (arm_valid_target_attribute_rec):
3959 Parse "arch=" and "+<ext>".
3960 (arm_valid_target_attribute_tree): Re-init global options.
3961 (arm_option_override): Make non-static.
3962 (arm_options_perform_arch_sanity_checks): Make errors fatal.
3963 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
3964 (__ARM_FEATURE_CRC32): Support undef.
3965 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
3966 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
3968 2017-11-17 David Malcolm <dmalcolm@redhat.com>
3970 * gdbinit.in (break-on-diagnostic): New command.
3972 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3974 * config/i386/i386.c (ix86_expand_epilogue): Change simple
3975 return to indirect jump for EH return if control-flow
3976 protection is enabled. Change explicit 'false' argument in
3977 pro_epilogue_adjust_stack with a value of
3979 * config/i386/i386.md (simple_return_indirect_internal):
3980 Remove SImode restriction to support 64-bit.
3982 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3984 * combine.c (added_notes_insn): New.
3985 (try_combine): Handle added_notes_insn like added_links_insn.
3986 Rewrite return value code.
3987 (distribute_notes): Set added_notes_insn to the earliest insn we added
3990 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3992 PR rtl-optimization/82621
3993 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
3994 dest of one of those SETs is unused.
3996 2017-11-17 Richard Biener <rguenther@suse.de>
3999 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
4000 * tree-pretty-print.c (dump_generic_node): Handle
4001 annot_expr_parallel_kind.
4002 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
4003 * gimplify.c (gimple_boolify): Likewise.
4005 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4007 * config.gcc (extra_headers): Add cet.h for x86 targets.
4008 * config/i386/cet.h: New file.
4009 * doc/install.texi: Add --enable-cet/--disable-cet.
4011 2017-11-17 Richard Biener <rguenther@suse.de>
4013 PR tree-optimization/83017
4014 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
4015 (gen_parallel_loop): Properly count iterations.
4016 (parallelize_loops): Handle loop->can_be_parallel independent
4017 of flag_loop_parallelize_all. Make static profitability test match
4019 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
4020 * invoke.texi (parloops-min-per-thread): Document.
4022 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
4024 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
4025 upstreaming review comments.
4027 2017-11-17 Tamar Christina <tamar.christina@arm.com>
4029 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
4030 with fast unaligned access.
4031 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
4033 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
4035 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
4036 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
4038 2017-11-17 Richard Biener <rguenther@suse.de>
4040 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
4041 folding of references.
4043 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
4046 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
4047 of replacing call to strncmp with corresponding call to strcmp when
4050 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
4052 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
4053 option prefer-avx256 for skylake-avx512 configuration.
4054 * config/i386/i386.c (ix86_option_override_internal): Ditto.
4055 (get_builtin_code_for_version): Ditto.
4057 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
4058 Monk Chiang <sh.chiang04@gmail.com>
4060 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
4061 (FIXED_REGISTERS): Reserve more register numbers.
4062 (CALL_USED_REGISTERS): Likewise.
4063 (REG_ALLOC_ORDER): Likewise.
4064 (REG_CLASS_CONTENTS): Likewise.
4065 (REGISTER_NAMES): Likewise.
4067 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
4068 Kito Cheng <kito.cheng@gmail.com>
4070 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
4072 * config/nds32/iterators.md: Add vector mode iterators and attributes.
4074 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
4076 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
4077 parameter list for vec_splats.
4079 2017-11-16 Joseph Myers <joseph@codesourcery.com>
4081 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
4083 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
4085 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
4088 * doc/install.texi: Document incompatibility of Perl >=5.6.26
4089 with the required version of automake 1.11.6.
4091 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
4093 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
4095 (DU_C2_3_power9): Correct reservation combinations.
4096 (FP_DIV_power9, VEC_DIV_power9): New.
4097 (power9-alu): Split out rotate/shift...
4098 (power9-rot): ...to here, correct dispatch resource.
4099 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
4101 (power9-fp): Correct latency.
4102 (power9-sdiv): Add div/sqrt resource.
4103 (power9-ddiv): Correct latency, add div/sqrt resource.
4104 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
4105 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
4107 (power9-qpdiv, power9-qpmul): Adjust resource usage.
4109 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
4111 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
4112 switch statement mapping KF built-ins to TF built-ins if we don't
4113 have the proper ISA 3.0 assembler support.
4115 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4117 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
4118 (lower_emutls_function_body): Do not compute it.
4120 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4122 * ipa-split.c (split_bb_info): Turn time to sreal.
4123 (split_point): Likewise.
4124 (dump_split_point): Likewise.
4125 (fine_split_points): Likewise.
4126 (execute_split_functions): Only zero split_bbs; turn time to sreals.
4128 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4130 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
4133 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4135 * predict.c (combine_predictions_for_bb): Preserve zero predicted
4137 (expensive_function_p): Remove useless assert.
4138 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
4140 2017-11-16 Martin Sebor <msebor@redhat.com>
4142 PR tree-optimization/82588
4143 PR tree-optimization/82583
4144 * tree-vrp.c (check_array_ref): Handle flexible array members,
4145 string literals, and inner indices.
4146 (search_for_addr_array): Add detail to diagnostics.
4148 2017-11-16 Nathan Sidwell <nathan@acm.org>
4152 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
4153 (SET_DECL_ASSEMBLER_NAME): Forward to
4154 overwrite_decl_assembler_name.
4155 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
4156 (overwrite_decl_assembler_name): Declare.
4157 * tree.c (overwrite_decl_assembler_name): New.
4158 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
4159 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
4160 (LANG_HOOKS_INITIALIZER): Add it.
4161 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
4162 * langhooks.c (lhd_set_decl_assembler_name): Use
4163 SET_DECL_ASSEMBLER_NAME.
4164 (lhd_overwrite_decl_assembler_name): Default implementation.
4166 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
4167 Jackson Woodruff <jackson.woodruff@arm.com>
4169 PR tree-optimization/71026
4170 * match.pd: Canonicalize constant multiplies in division.
4172 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
4174 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
4175 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
4176 Remove OPT_fomit_frame_pointer entry.
4177 * common/config/arc/arc-common.c: Likewise.
4178 * common/config/arm/arm-common.c: Likewise.
4179 * common/config/avr/avr-common.c: Likewise.
4180 * common/config/c6x/c6x-common.c: Likewise.
4181 * common/config/cr16/cr16-common.c: Likewise.
4182 * common/config/cris/cris-common.c: Likewise.
4183 * common/config/epiphany/epiphany-common.c: Likewise.
4184 * common/config/fr30/fr30-common.c: Likewise.
4185 * common/config/frv/frv-common.c: Likewise.
4186 * common/config/ia64/ia64-common.c: Likewise.
4187 * common/config/iq2000/iq2000-common.c: Likewise.
4188 * common/config/lm32/lm32-common.c: Likewise.
4189 * common/config/m32r/m32r-common.c: Likewise.
4190 * common/config/mcore/mcore-common.c: Likewise.
4191 * common/config/microblaze/microblaze-common.c: Likewise.
4192 * common/config/mips/mips-common.c: Likewise.
4193 * common/config/mmix/mmix-common.c: Likewise.
4194 * common/config/mn10300/mn10300-common.c: Likewise.
4195 * common/config/nios2/nios2-common.c: Likewise.
4196 * common/config/pa/pa-common.c: Likewise.
4197 * common/config/pdp11/pdp11-common.c: Likewise.
4198 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
4199 * common/config/riscv/riscv-common.c: Likewise.
4200 * common/config/rs6000/rs6000-common.c: Likewise.
4201 * common/config/rx/rx-common.c: Likewise.
4202 * common/config/s390/s390-common.c: Likewise.
4203 * common/config/sh/sh-common.c: Likewise.
4204 * common/config/sparc/sparc-common.c: Likewise.
4205 * common/config/tilegx/tilegx-common.c: Likewise.
4206 * common/config/tilepro/tilepro-common.c: Likewise.
4207 * common/config/v850/v850-common.c: Likewise.
4208 * common/config/visium/visium-common.c: Likewise.
4209 * common/config/xstormy16/xstormy16-common.c: Likewise.
4210 * common/config/xtensa/xtensa-common.c: Likewise.
4211 * invoke.texi (-fomit-frame-pointer): Update documentation.
4213 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4215 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
4217 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4219 * cfg.c (scale_bbs_frequencies_int,
4220 cale_bbs_frequencies_gcov_type): Remove.
4221 * cfg.h (scale_bbs_frequencies_int,
4222 cale_bbs_frequencies_gcov_type): Remove.
4224 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4226 * tree-ssa-loop-manip.c
4227 (scale_dominated_blocks_in_loop): Update to profile counts.
4228 (tree_transform_and_unroll_loop): Likewise.
4230 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4232 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
4233 scale_bbs_frequencies_int.
4235 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4237 * final.c (compute_alignments): Use counts rather than frequencies.
4239 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4241 * cfgloopanal.c: Include sreal.h
4242 (average_num_loop_insns): Use counts and sreal for accounting.
4244 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4246 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
4249 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4250 Kito Cheng <kito.cheng@gmail.com>
4252 * config/nds32/constraints.md: Provide more constraints.
4253 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
4254 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
4255 support constraints usage.
4257 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4259 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
4261 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4262 Kito Cheng <kito.cheng@gmail.com>
4264 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
4265 * config/nds32/nds32.opt: Refine the layout.
4266 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
4267 TARGET_EXT_STRING): Support new options.
4268 * config/nds32/nds32.h: Likewise.
4269 * config/nds32/nds32.md: Likewise.
4270 * config/nds32/nds32-predicates.c: Likewise.
4271 * config/nds32/constraints.md: Likewise.
4272 * common/config/nds32/nds32-common.c: Likewise.
4274 2017-11-16 Julia Koval <julia.koval@intel.com>
4277 * config/i386/gfniintrin.h: Add sse check.
4278 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
4280 2017-11-16 Julia Koval <julia.koval@intel.com>
4282 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
4283 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
4284 (ix86_handle_option): Handle -mavx512vbmi2.
4285 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
4286 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4287 * config/i386/i386-c.c (__AVX512VBMI2__): New.
4288 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
4289 (ix86_valid_target_attribute_inner_p): Ditto.
4290 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
4291 * config/i386/i386.opt (mavx512vbmi2): New option.
4292 * doc/invoke.texi: Add new option.
4294 2017-11-16 Julia Koval <julia.koval@intel.com>
4296 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
4297 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
4298 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
4299 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
4300 _mm512_gf2p8mul_epi8): New intrinsics.
4301 * config/i386/i386-builtin-types.def
4302 (V64QI_FTYPE_V64QI_V64QI): New type.
4303 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
4304 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
4305 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
4306 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
4307 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
4308 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4310 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
4312 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
4313 explicitly as a stream of bytes.
4315 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4317 * config/rs6000/altivec.h (vec_xst_be): New #define.
4318 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
4319 and externalize from *altivec_vperm_<mode>_internal.
4320 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
4322 (XL_BE_V8HI): Likewise.
4323 (XL_BE_V4SI): Likewise.
4324 (XL_BE_V4SI): Likewise.
4325 (XL_BE_V2DI): Likewise.
4326 (XL_BE_V4SF): Likewise.
4327 (XL_BE_V2DF): Likewise.
4328 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
4329 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
4330 all array entries with these keys: VSX_BUILTIN_VEC_XL,
4331 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
4332 VSX_BUILTIN_VEC_XST_BE.
4333 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
4334 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
4336 (altivec_init_builtins): Replace conditional calls to def_builtin
4337 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
4338 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
4339 with unconditional calls. Remove calls to def_builtin for
4340 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
4341 __builtin_vec_xst_be.
4342 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
4343 to define_expand, and add alternate RTL generation for P8.
4344 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
4345 vsx_ld_elemrev_v8hi.
4346 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
4347 add alternate RTL generation for P8.
4348 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
4349 vsx_ld_elemrev_v16qi.
4350 (vsx_st_elemrev_v8hi): Convert define_insn
4351 to define_expand, and add alternate RTL generation for P8.
4352 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
4353 vsx_st_elemrev_v8hi.
4354 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
4355 add alternate RTL generation for P8.
4356 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
4357 vsx_st_elemrev_v16qi.
4359 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
4362 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
4363 TARGET_AVX512ER check.
4364 (ix86_option_override_internal): Set MASK_VZEROUPPER if
4365 neither -mzeroupper nor -mno-zeroupper is used and
4366 TARGET_EMIT_VZEROUPPER is set.
4367 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
4368 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
4370 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
4372 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4373 folding of vector compares.
4374 (fold_build_vec_cmp): New helper function.
4375 (fold_compare_helper): New helper function.
4376 (builtin_function_type): Add compare builtins to the list of functions
4377 having unsigned arguments. Cosmetic updates to comment indentation.
4378 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
4379 the not+eq combination.
4381 2017-11-15 Bin Cheng <bin.cheng@arm.com>
4383 PR tree-optimization/82726
4384 PR tree-optimization/70754
4385 * tree-predcom.c (order_drefs_by_pos): New function.
4386 (combine_chains): Move code setting has_max_use_after to...
4387 (try_combine_chains): ...here. New parameter. Sort combined chains
4388 according to position information.
4389 (tree_predictive_commoning_loop): Update call to above function.
4390 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
4392 2017-11-15 Bin Cheng <bin.cheng@arm.com>
4394 PR tree-optimization/82726
4396 2017-01-23 Bin Cheng <bin.cheng@arm.com>
4398 PR tree-optimization/70754
4399 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
4400 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
4401 combined stmt before it if not NULL.
4402 (combine_chains): Process refs reversely and compute dominance point
4406 2017-02-23 Bin Cheng <bin.cheng@arm.com>
4408 PR tree-optimization/79663
4409 * tree-predcom.c (combine_chains): Process refs in reverse order
4410 only for ZERO length chains, and add explaining comment.
4412 2017-11-15 Tamar Christina <tamar.christina@arm.com>
4414 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
4415 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
4416 * doc/invoke.texi (ARM Options): Add armv8.3-a.
4418 2017-11-15 Tamar Christina <tamar.christina@arm.com>
4420 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
4422 2017-11-15 Martin Liska <mliska@suse.cz>
4424 * tree-cfg.c (pass_warn_function_return::execute):
4425 Compare warn_return_type for greater than zero.
4427 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
4431 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
4432 to return true on Xeon and not on Xeon Phi.
4433 (ix86_check_avx256_register): Changed to ...
4434 (ix86_check_avx_upper_register): ... this. Add extra check for
4435 VALID_AVX512F_REG_OR_XI_MODE.
4436 (ix86_avx_u128_mode_needed): Changed
4437 ix86_check_avx256_register to ix86_check_avx_upper_register.
4438 (ix86_check_avx256_stores): Changed to ...
4439 (ix86_check_avx_upper_stores): ... this. Changed
4440 ix86_check_avx256_register to ix86_check_avx_upper_register.
4441 (ix86_avx_u128_mode_after): Changed
4442 avx_reg256_found to avx_upper_reg_found. Changed
4443 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
4444 (ix86_avx_u128_mode_entry): Changed
4445 ix86_check_avx256_register to ix86_check_avx_upper_register.
4446 (ix86_avx_u128_mode_exit): Ditto.
4447 * config/i386/i386.h: (host_detect_local_cpu): New define.
4449 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
4451 * config/arm/xgene1.md (xgene1): Split into automatons
4452 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
4453 (xgene1_f_load): Adjust reservations and/or types.
4454 (xgene1_f_store): Likewise.
4455 (xgene1_load_pair): Likewise.
4456 (xgene1_store_pair): Likewise.
4457 (xgene1_fp_load1): Likewise.
4458 (xgene1_load1): Likewise.
4459 (xgene1_store1): Likewise.
4460 (xgene1_move): Likewise.
4461 (xgene1_alu): Likewise.
4462 (xgene1_simd): Likewise.
4463 (xgene1_bfm): Likewise.
4464 (xgene1_neon_load1): Likewise.
4465 (xgene1_neon_store1): Likewise.
4466 (xgene1_neon_logic): Likewise.
4467 (xgene1_neon_st1): Likewise.
4468 (xgene1_neon_ld1r): Likewise.
4469 (xgene1_alu_cond): Added.
4470 (xgene1_shift_reg): Likwise.
4471 (xgene1_bfx): Likewise.
4472 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
4474 2017-11-15 Jakub Jelinek <jakub@redhat.com>
4477 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
4479 (can_widen_mult_without_libcall): New function.
4480 (expand_mul_overflow): If only checking unsigned mul overflow,
4481 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
4482 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
4483 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
4484 (expand_DIVMOD): Formatting fix.
4485 * expmed.h (expand_mult): Add NO_LIBCALL argument.
4486 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
4487 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
4489 PR tree-optimization/82977
4490 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
4491 constructed temporary to strlen_to_stridx.put.
4493 2017-11-15 Martin Liska <mliska@suse.cz>
4495 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
4496 * configure: Regenerate.
4498 2017-11-15 Martin Liska <mliska@suse.cz>
4501 * config/sh/sh-mem.cc: Use proper probability for
4504 2017-11-14 Jeff Law <law@redhat.com>
4506 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
4507 the red zone for stack_clash_protection_final_dynamic_probe targets
4508 when the total dynamic stack size is zero bytes.
4510 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
4511 blocks is post order.
4513 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
4515 * dumpfile.h (TDF_COMPARE_DEBUG): New.
4516 * final.c (rest_of_clean_state): Set it for the
4517 -fcompare-debug dump.
4518 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
4519 class when TDF_COMPARE_DEBUG is set.
4521 * dwarf2out.c (gen_producer_string): Discard
4524 2017-11-15 Joseph Myers <joseph@codesourcery.com>
4527 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
4528 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
4529 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
4530 (__TGMATH_REAL_2_3): Remove macros.
4531 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
4532 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
4534 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
4536 (remquo): Define using __TGMATH_REAL_3.
4538 2017-11-14 Jeff Law <law@redhat.com>
4540 * vr-values.c: New file with contents extracted from tree-vrp.c.
4541 * Makefile.in (OBJS): Add vr-values.o
4542 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
4543 (set_value_range, set_and_canonicalize_value_range): Likewise.
4544 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
4545 (value_range_constant_singleton, symbolic_range_p): Likewise.
4546 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
4547 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
4548 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
4549 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
4550 (find_case_label_range, find_case_label_index): Likewise.
4551 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
4552 (range_int_cst_singleton_p, value_inside_range): Likewise.
4553 (get_single_symbol): Likewise.
4554 (switch_update): Move structure definition here.
4555 (to_remove_edges, to_update_switch_stmts): Provide externs.
4556 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
4557 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
4558 (vrp_val_is_min, set_value_range): Likewise.
4559 (set_and_canonicalize_value_range, copy_value_range): Likewise.
4560 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
4561 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
4562 (range_is_nonnull, range_int_cst_p): Likewwise.
4563 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
4564 (get_single_symbol, operand_less_p): Likewise
4565 (compare_values_warnv, compare_values): Likewise.
4566 (value_inside_range, value_range_constant_singleton): Likewise.
4567 (zero_nonzero_bitgs_from_vr): Likewise.
4568 (extract_range_from_binary_expr_1): Likewise.
4569 (overflow_comparison_p): Likewise.
4570 (to_remove_edges, to_update_switch_stmts): Likewise.
4571 (find_case_label-index, find_case_label_range): Likewise.
4572 (switch_update, set_value_range_to_nonnegative): Remove.
4573 (set_value_range_to_truthvalue): Likewise.
4574 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
4575 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
4576 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
4577 (find_case_label_ranges, test_for_singularity): Likewise.
4578 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
4579 (x_vr_values): Move to its remaining use site.
4581 2017-11-10 Jeff Law <law@redhat.com>
4583 * vr-values.h (VR_INITIALIZER): Move #define here.
4584 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
4585 * Makefile.in (OBJS): Add tree-evrp.o
4586 * tree-vrp.h (assert_info): Move structure definition here.
4587 (set_value_range_to_varying): Prototype.
4588 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
4589 (infer_value_range, register_edge_assert_for): Likewise.
4590 (stmt_interesting_for_vrp): Likewise.
4591 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
4592 (set_value_range_to_varying): No longer static.
4593 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
4594 (infer_value_range, register_edge_assert_for): Likewise.
4596 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
4598 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
4599 generate the XXBRD instruction.
4601 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
4602 (rs6000_builtin_type_compatible): Treat _Float128 and long double
4603 as being compatible if -mabi=ieeelongdouble.
4604 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
4605 to setup float128 built-ins with hardware support.
4606 (BU_FLOAT128_HW_2): Likewise.
4607 (BU_FLOAT128_HW_3): Likewise.
4608 (BU_FLOAT128_HW_VSX_1): Likewise.
4609 (BU_FLOAT128_HW_VSX_2): Likewise.
4610 (scalar_extract_expq): Change float128 built-in functions to
4611 accommodate having both KFmode and TFmode functions. Use the
4612 KFmode variant as the default.
4613 (scalar_extract_sigq): Likewise.
4614 (scalar_test_neg_qp): Likewise.
4615 (scalar_insert_exp_q): Likewise.
4616 (scalar_insert_exp_qp): Likewise.
4617 (scalar_test_data_class_qp): Likewise.
4618 (sqrtf128_round_to_odd): Delete processing the round to odd
4619 built-in functions as special built-in functions, and define them
4620 as float128 built-ins. Use the KFmode variant as the default.
4621 (truncf128_round_to_odd): Likewise.
4622 (addf128_round_to_odd): Likewise.
4623 (subf128_round_to_odd): Likewise.
4624 (mulf128_round_to_odd): Likewise.
4625 (divf128_round_to_odd): Likewise.
4626 (fmaf128_round_to_odd): Likewise.
4627 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
4628 support for KFmode and TFmode xststdcqp calls.
4629 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
4630 point, switch the built-in handlers for the get/set float128
4631 exponent, get float128 mantissa, float128 test built-ins, and the
4632 float128 round to odd built-in functions. Eliminate creating the
4633 float128 round to odd built-in functions as special built-ins.
4634 (rs6000_init_builtins): Eliminate special creation of the float128
4635 round to odd built-in functions.
4636 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
4637 function insns to support both TFmode and KFmode variants.
4638 (xsxsigqp_<mode>): Likewise.
4639 (xsiexpqpf_<mode>): Likewise.
4640 (xsiexpqp_<mode>): Likewise.
4641 (xststdcqp_<mode>): Likewise.
4642 (xststdcnegqp_<mode>): Likewise.
4643 (xststdcqp_<mode>): Likewise.
4645 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4647 * tree-ssa-threadupdate.c (compute_path_counts): Remove
4648 unused path_in_freq_ptr parameter.
4649 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
4651 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4653 * ipa-inline.c (edge_badness): Dump sreal frequency.
4654 (compute_inlined_call_time): Match natural implementaiton ...
4655 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
4656 forgotten division by CGRAPH_FREQ_BASE.
4658 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4660 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
4661 Solaris 11. Update comment.
4662 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
4664 * config/sol2.h (STARTFILE_SPEC): Likewise.
4665 * configure: Regenerate.
4667 2017-11-14 Carl Love <cel@us.ibm.com>
4669 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
4670 le_ and be_ prefixes to swap* variables. Remove
4671 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
4674 2017-11-14 Jason Merrill <jason@redhat.com>
4676 Support GTY((cache)) on hash_map.
4677 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
4678 (ggc_cache_remove): Override it instead of ggc_mx.
4679 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
4680 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
4681 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
4682 (simple_cache_map_traits): Override maybe_mx.
4683 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
4684 (hash_map): Friend gt_cleare_cache.
4685 (gt_cleare_cache): New.
4686 * tree.h (tree_cache_traits): New hash_map traits class.
4687 (tree_cache_map): New typedef.
4689 2017-11-14 Richard Biener <rguenther@suse.de>
4691 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
4692 paramter and handling.
4693 (cleanup_control_flow_bb): Likewise.
4694 (cleanup_control_flow_pre): New helper performing a DFS walk
4695 to call cleanup_control_flow_bb in PRE order.
4696 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
4697 via cleanup_control_flow_pre.
4699 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
4701 * config/aarch64/aarch64-simd.md
4702 (aarch64_simd_bsl<mode>_internal): Remove DImode.
4703 (*aarch64_simd_bsl<mode>_alt): Likewise.
4704 (aarch64_simd_bsldi_internal): New.
4705 (aarch64_simd_bsldi_alt): Likewise.
4707 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4709 * tracer.c (better_p): Do not compare frequencies.
4710 * reg-stack.c (better_edge): Likewise.
4711 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
4714 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4716 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
4717 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
4718 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
4719 * lto-streamer-in.c (input_function): Use update_max_bb_count.
4720 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
4721 * predict.c (maybe_hot_frequency_p): Inline to ...
4722 (maybe_hot_count_p): ... here; rewrite to counts.
4723 (counts_to_freqs): Rename to ...
4724 (update_max_bb_count): ... this one.
4725 (expensive_function_p): Use counts.
4726 (estimate_bb_frequencies): Update.
4727 (rebuild_frequencies): Update.
4728 * predict.h (counts_to_freqs): Rename to ...
4729 (update_max_bb_count): ... this one.
4730 * profile.c (compute_branch_probabilities): Add debug info
4731 * tree-inline.c (expand_call_inline): Update debug info.
4732 (optimize_inline_calls): Use update_max_bb_count..
4733 (tree_function_versioning): Use update_max_bb_count..
4734 * value-prof.c (gimple_value_profile_transformations):
4735 Do not use update_max_bb_count.
4737 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4739 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
4740 always use frequencies.
4742 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4744 * bb-reorder.c: Remove frequencies from comments.
4745 (better_edge_p): Use profile counts.
4746 (find_traces): Dump profile counts.
4747 (rotate_loop): Use profile counts.
4748 (find_traces_1_round): Likewise.
4749 (connect_better_edge_p): Use counts instead of probabilities for
4751 (copy_bb_p): Drop early check for non-0 frequency.
4752 (sanitize_hot_paths): Update comments.
4754 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4756 * ipa-split.c (struct split_point): Add count.
4757 (consider_split): Do not compute incoming frequency; compute incoming
4758 count and store it to split_point.
4759 (split_function): Set count of the call to split part correctly.
4761 2017-11-13 Carl Love <cel@us.ibm.com>
4763 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
4765 2017-11-13 Tom Tromey <tom@tromey.com>
4767 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
4769 2017-11-13 Carl Love <cel@us.ibm.com>
4771 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
4772 Add support for builtins:
4773 unsigned int vec_first_{,miss}_match_{,or_eos}index,
4774 vector {un,}signed {char,int,short},
4775 vector {un,}signed {char,int,short}) arguments.
4776 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
4777 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
4778 Add BU_P9V_AV_2 expansions for the builtins.
4779 * config/rs6000/altivec.h (vec_first_match_index,
4780 vec_first_mismatch_index, vec_first_match_or_eos_index,
4781 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
4782 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
4783 new extern declaration.
4784 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
4785 * config/rs6000/vsx.md (first_match_index_<mode>,
4786 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
4787 first_mismatch_or_eos_index_<mode>): Add define expand.
4788 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
4789 * doc/extend.texi: Update the built-in documenation file for the new
4792 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
4794 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
4795 into the min/max operations for _Float<N> and _Float<N>X types.
4797 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4800 * dwarf2out.c (do_eh_frame): New static variable.
4801 (dwarf2out_begin_prologue): Set it.
4802 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
4804 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4806 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
4808 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
4810 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4813 * debug.h (dwarf2out_do_eh_frame): Declare.
4814 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
4815 (dwarf2out_do_frame): Use it.
4816 (dwarf2out_do_cfi_asm): Likewise.
4817 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
4818 (dwarf2out_assembly_start): Likewise.
4819 (dwarf2out_begin_prologue): Fix comment.
4820 * toplev.c (compile_file): Always call dwarf2out_frame_finish
4821 if the target needs either debug or unwind DWARF2 info.
4822 * lto-opts.c (lto_write_options): Do not save -fexceptions,
4823 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
4824 -ftrapping-math, -ftrapv and -fwrapv.
4826 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4828 * cgraph.c (cgraph_edge::sreal_frequency): New function.
4829 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
4830 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
4831 (estimate_edge_size_and_time): Likewise.
4832 (ipa_merge_fn_summary_after_inlining): Likewise.
4833 * ipa-inline.c (cgraph_freq_base_rec): Remove.
4834 (compute_uninlined_call_time): Use sreal_frequency.
4835 (compute_inlined_call_time): Likewise.
4836 (ipa_inline): Do not initialize cgraph_freq_base_rec.
4837 * profile-count.c: Include sreal.h.
4838 (profile_count::to_sreal_scale): New.
4839 * profile-count.h: Forward declare sreal.
4840 (profile_count::to_sreal_scale): Declare.
4842 2017-11-13 Nathan Sidwell <nathan@acm.org>
4844 * diagnostic.c (maybe_line_and_column): New.
4845 (diagnostic_get_location_text): Use it.
4846 (diagnostic_report_current_module): Likewise.
4847 (test_diagnostic_get_location_text): Add tests.
4849 2017-11-13 Luis Machado <luis.machado@linaro.org>
4851 * doc/md.texi (Specifying processor pipeline description): Fix
4852 incorrect latency for the div instruction example.
4854 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4856 PR tree-optimization/78821
4857 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
4858 that bit_not_p is the same.
4859 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
4860 (split_group): Count precisely bit_not_p bits in each statement.
4861 (invert_op): New function.
4862 (imm_store_chain_info::output_merged_store): Use invert_op to
4863 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
4864 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
4866 2017-11-13 Martin Liska <mliska@suse.cz>
4868 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
4869 (struct source_info): Likewise.
4870 (add_branch_counts): Likewise.
4871 (add_line_counts): Likewise.
4872 (function_summary): Likewise.
4873 (output_intermediate_line): Likewise.
4874 (generate_results): Likewise.
4876 2017-11-13 Martin Liska <mliska@suse.cz>
4878 * gcov.c (struct block_info): Remove typedef for block_t.
4879 (struct line_info): Likewise.
4880 (line_info::has_block): Likewise.
4881 (EXIT_BLOCK): Likewise.
4882 (unblock): Likewise.
4883 (circuit): Likewise.
4884 (get_cycles_count): Likewise.
4885 (process_file): Likewise.
4886 (read_graph_file): Likewise.
4887 (solve_flow_graph): Likewise.
4888 (find_exception_blocks): Likewise.
4889 (add_line_counts): Likewise.
4890 (accumulate_line_info): Likewise.
4891 (output_line_details): Likewise.
4893 2017-11-13 Martin Liska <mliska@suse.cz>
4895 * gcov.c (struct arc_info): Remove typedef for arc_t.
4896 (struct line_info): Likewise.
4897 (add_branch_counts): Likewise.
4898 (output_branch_count): Likewise.
4899 (function_info::~function_info): Likewise.
4900 (circuit): Likewise.
4901 (output_intermediate_line): Likewise.
4902 (read_graph_file): Likewise.
4903 (solve_flow_graph): Likewise.
4904 (find_exception_blocks): Likewise.
4905 (add_line_counts): Likewise.
4906 (accumulate_line_info): Likewise.
4907 (output_line_details): Likewise.
4908 (output_function_details): Likewise.
4910 2017-11-13 Martin Liska <mliska@suse.cz>
4912 * gcov.c (struct function_info): Remove typedef for function_t.
4913 (struct source_info): Likewise.
4914 (source_info::get_functions_at_location): Likewise.
4915 (solve_flow_graph): Likewise.
4916 (find_exception_blocks): Likewise.
4917 (add_line_counts): Likewise.
4918 (output_intermediate_file): Likewise.
4919 (process_file): Likewise.
4920 (generate_results): Likewise.
4921 (release_structures): Likewise.
4922 (read_graph_file): Likewise.
4923 (read_count_file): Likewise.
4924 (accumulate_line_counts): Likewise.
4925 (output_lines): Likewise.
4927 2017-11-13 Martin Liska <mliska@suse.cz>
4929 * gcov.c (function_info::function_info): Remove num_counts
4930 and add vector<gcov_type>.
4931 (function_info::~function_info): Use the vector.
4932 (process_file): Likewise.
4933 (read_graph_file): Likewise.
4934 (read_count_file): Likewise.
4935 (solve_flow_graph): Likewise.
4937 2017-11-13 Martin Liska <mliska@suse.cz>
4939 * gcov.c (function_info::is_artificial): New function.
4940 (process_file): Erase all artificial early.
4941 (generate_results): Skip as all artificial are already
4944 2017-11-13 Martin Liska <mliska@suse.cz>
4946 * gcov.c (read_graph_file): Store to global vector of functions.
4947 (read_count_file): Iterate the vector.
4948 (process_file): Likewise.
4949 (generate_results): Likewise.
4950 (release_structures): Likewise.
4952 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4954 PR tree-optimization/82954
4955 * gimple-ssa-store-merging.c
4956 (imm_store_chain_info::coalesce_immediate_stores): If
4957 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
4959 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
4961 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
4962 Upddate call to ENDIAN_LANE_N.
4963 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
4964 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
4965 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
4966 and use aarch64_endian_lane_rtx.
4967 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
4969 2017-11-12 Tom de Vries <tom@codesourcery.com>
4971 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
4973 2017-11-12 Tom de Vries <tom@codesourcery.com>
4975 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
4977 2017-11-12 Tom de Vries <tom@codesourcery.com>
4979 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
4981 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4982 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4983 * defaults.h (ASM_OUTPUT_LABELREF): Same.
4985 2017-11-11 Martin Sebor <msebor@redhat.com>
4988 * doc/extend.texi (attribute nonstring): Remove spurious argument.
4991 * prefic.c (translate_name): Replace strncpy with memcpy to
4992 avoid -Wstringop-truncation.
4994 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4996 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
4998 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
5000 * predict.c (maybe_hot_frequency_p): Do not use cfun.
5002 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
5004 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
5007 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
5009 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
5010 enable generating XXBRH if the value is in a vector register.
5011 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
5012 value is in a vector register.
5013 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
5014 register to register bswap64's instead of doing the GPR sequence
5015 used on previous machines.
5016 (bswapdi2_xxbrd): New insn.
5017 (bswapdi2_reg): Disallow on ISA 3.0.
5018 (register to register bswap64 splitter): Do not split the insn on
5019 ISA 3.0 systems that use XXBRD.
5021 2017-11-10 Martin Sebor <msebor@redhat.com>
5024 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
5026 (find_subframework_file): Same.
5028 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
5030 * auto-profile.c (afdo_indirect_call): Drop frequency.
5031 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
5032 (cgraph_node::create_edge): Drop frequency argument.
5033 (cgraph_node::create_indirect_edge): Drop frequency argument.
5034 (cgraph_edge::make_speculative): Drop frequency arguments.
5035 (cgraph_edge::resolve_speculation): Do not update frequencies
5036 (cgraph_edge::dump_edge_flags): Do not dump frequency.
5037 (cgraph_node::dump): Check consistency in IPA mode.
5038 (cgraph_edge::maybe_hot_p): Use IPA counter.
5039 (cgraph_edge::verify_count_and_frequency): Rename to ...
5040 (cgraph_edge::verify_count): ... this one; drop frequency checking.
5041 (cgraph_node::verify_node): Update.
5042 * cgraph.h (struct cgraph_edge): Drop frequency.
5043 (cgraph_edge::frequency): New function.
5044 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
5046 (cgraph_edge::rebuild_edges): Likewise.
5047 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
5048 (duplicate_thunk_for_node): Do not pass frequency.
5049 (cgraph_node::create_clone): Scale only counts.
5050 (cgraph_node::create_virtual_clone): Do not pass frequency.
5051 (cgraph_node::create_edge_including_clones): Do not pass frequency.
5052 (cgraph_node::create_version_clone): Do not pass frequency.
5053 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
5054 (cgraph_node::expand_thunk): Do not pass frequency.
5055 (cgraph_node::create_wrapper): Do not pass frequency.
5056 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
5058 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
5059 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
5060 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
5061 (gather_caller_stats): Use frequency function.
5062 (ipcp_cloning_candidate_p): Use frequency function.
5063 (ipcp_propagate_stage): Use frequency function.
5064 (get_info_about_necessary_edges): Use frequency function.
5065 (update_profiling_info): Update only IPA profile.
5066 (update_specialized_profile): Use frequency functoin.
5067 (perhaps_add_new_callers): Update only IPA profile.
5068 * ipa-devirt.c (ipa_devirt): Use IPA profile.
5069 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
5070 (dump_ipa_call_summary): Use frequency function.
5071 (estimate_edge_size_and_time): Use frequency function.
5072 (ipa_merge_fn_summary_after_inlining): Use frequency function.
5073 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
5074 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
5075 (update_noncloned_counts): ... ths one; scale counts only.
5076 (clone_inlined_nodes): Do not scale frequency.
5077 (inline_call): Do not pass frequency.
5078 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
5079 (compute_inlined_call_time): Use IPA profile.
5080 (want_inline_small_function_p): Use IPA profile.
5081 (want_inline_self_recursive_call_p): Use IPA profile.
5082 (edge_badness): Use IPA profile.
5083 (lookup_recursive_calls): Use IPA profile.
5084 (recursive_inlining): Do not pass frequency.
5085 (resolve_noninline_speculation): Do not update frequency.
5086 (inline_small_functions): Collect max of IPA profile.
5087 (dump_overall_stats): Dump IPA porfile.
5088 (dump_inline_stats): Dump IPA porfile.
5089 (ipa_inline): Collect IPA stats.
5090 * ipa-inline.h (clone_inlined_nodes): Update prototype.
5091 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
5092 (ipa_propagate_frequency): Use frequency function.
5093 (ipa_profile): Cleanup.
5094 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
5095 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
5096 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
5097 (input_node): Do not stream frequency.
5098 (input_edge): Do not stream frequency.
5099 (merge_profile_summaries): Scale only IPA profiles.
5100 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
5101 * predict.c (drop_profile): Do not recompute frequency.
5102 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
5103 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
5104 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
5105 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
5106 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
5107 * tree-inline.c (copy_bb): Do not scale frequency.
5108 (expand_call_inline): Do not scale frequency.
5109 (tree_function_versioning): Do not scale frequency.
5110 * ubsan.c (ubsan_create_edge): Do not pass frequency.
5112 2017-11-10 Julia Koval <julia.koval@intel.com>
5114 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
5115 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
5116 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
5117 (_mm256_maskz_gf2p8affine_epi64_epi8)
5118 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
5119 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
5120 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
5121 (__builtin_ia32_vgf2p8affineqb_v32qi)
5122 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
5123 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
5125 2017-11-10 Tamar Christina <tamar.christina@arm.com>
5129 (arm_option_override): Refactor.
5130 (arm_option_reconfigure_globals): New.
5131 (arm_options_perform_arch_sanity_checks): New.
5132 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
5134 (arm_options_perform_arch_sanity_checks): Likewise
5136 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
5138 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
5139 (power9-qpmul): New.
5140 * rs6000/rs6000.md ("type" attr): Add qmul.
5141 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
5142 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
5143 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
5145 2017-11-10 Martin Sebor <msebor@redhat.com>
5148 * builtins.c (compute_objsize): Handle arrays that
5149 compute_builtin_object_size likes to fail for. Make extern.
5150 * builtins.h (compute_objsize): Declare.
5151 (check_strncpy_sizes): New function.
5152 (expand_builtin_strncpy): Call check_strncpy_sizes.
5153 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
5154 -Wstringop-truncation.
5155 (gimple_fold_builtin_strncat): Same.
5156 * gimple.c (gimple_build_call_from_tree): Set call location.
5157 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
5158 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
5159 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
5160 (handle_builtin_strlen): Use strlen_to_stridx.
5161 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
5163 Use strlen_to_stridx.
5164 (pass_strlen::execute): Release strlen_to_stridx.
5165 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
5166 (-Wstringop-truncation): Document new option.
5168 2017-11-10 Martin Liska <mliska@suse.cz>
5170 PR gcov-profile/82702
5171 * gcov.c (main): Handle intermediate files in a different
5173 (get_gcov_intermediate_filename): New function.
5174 (output_gcov_file): Remove support of intermediate files.
5175 (generate_results): Allocate intermediate file.
5176 (release_structures): Clean-up properly fn_end.
5177 (output_intermediate_file): Start iterating with line 1.
5179 2017-11-10 Jakub Jelinek <jakub@redhat.com>
5181 PR tree-optimization/82929
5182 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
5183 ops_swapped_p non-static data member.
5184 (store_immediate_info::store_immediate_info): Clear it.
5185 (imm_store_chain_info::coalesce_immediate_stores): If swapping
5186 ops set ops_swapped_p.
5187 (count_multiple_uses): Handle ops_swapped_p.
5189 2017-11-10 Martin Liska <mliska@suse.cz>
5191 * coverage.c (coverage_init): Stream information about
5192 support of has_unexecuted_blocks.
5193 * doc/gcov.texi: Document that.
5194 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
5195 * gcov.c (read_graph_file): Likewise.
5196 (output_line_beginning): Fix a small issue with
5199 2017-11-10 Bin Cheng <bin.cheng@arm.com>
5201 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
5202 reference of trivial component.
5204 2017-11-10 Jakub Jelinek <jakub@redhat.com>
5207 * gimple-ssa-store-merging.c
5208 (pass_store_merging::terminate_all_aliasing_chains): For
5209 gimple_store_p stmts also call refs_output_dependent_p.
5211 PR rtl-optimization/82913
5212 * compare-elim.c (try_merge_compare): Punt if def_insn is not
5215 2017-11-09 Jeff Law <law@redhat.com>
5217 * vr-values.h: New file with vr_values class.
5218 * tree-vrp.c: Include vr-values.h
5219 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
5220 data objects into the vr_values class.
5221 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
5222 (get_value_range): Make it a member function within vr_values class.
5223 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
5224 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
5225 (op_with_constant_singleton_value_range): Likewise.
5226 (extract_range_for_var_from_comparison_expr): Likewise.
5227 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
5228 (extract_range_from_binary_expr): Likewise.
5229 (extract_range_from_unary_expr): Likewise.
5230 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
5231 (check_for_binary_op_overflow, extract_range_basic): Likewise.
5232 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
5233 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
5234 (compare_name_with_value, compare_names): Likewise.
5235 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
5236 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
5237 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
5238 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
5239 (extract_range_from_phi_node): Likewise.
5240 (simplify_truth_ops_using_ranges): Likewise.
5241 (simplify_div_or_mod_using_ranges): Likewise.
5242 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
5243 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
5244 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
5245 (simplify_float_conversion_using_ranges): Likewise.
5246 (simplify_internal_call_using_ranges): Likewise.
5247 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
5248 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
5249 poitner via x_vr_values for calls into gimple folder.
5250 (vrp_initialize_lattice): Make this the vr_values ctor.
5251 (vrp_free_lattice): Make this the vr_values dtor.
5252 (set_vr_value): New function.
5253 (class vrp_prop): Add vr_values data member. Add various member
5254 functions as well as member functions that delegate to vr_values.
5255 (check_array_ref): Make a member function within vrp_prop class.
5256 (search_for_addr_array, vrp_initialize): Likewise.
5257 (vrp_finalize): Likewise. Revamp to avoid direct access to
5258 vr_value, values_propagated, etc.
5259 (check_array_bounds): Extract vrp_prop class instance pointer from
5260 walk info structure. Use it to call member functions.
5261 (check_all_array_refs): Make a member function within vrp_prop class.
5262 Smuggle class instance pointer via walk info structure.
5263 (x_vr_values): New local static.
5264 (vrp_valueize): Use x_vr_values to get class instance.
5265 (vr_valueize_1): Likewise.
5266 (class vrp_folder): Add vr_values data member. Add various member
5267 functions as well as member functions that delegate to vr_values.
5268 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
5269 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
5270 class instance from vr_values. Use it to call member functions.
5271 (vrp_dom_walker): Add vr_values data member.
5272 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
5273 instance via x_vr_values.
5274 (identify_jump_threads): Accept vr_values as argument. Store
5275 it into the walker structure.
5276 (evrp_dom_walker): Add vr_values class data member. Add various
5278 (evrp_dom_walker::try_find_new_range): Use vr_values data
5279 member to access the memory allocator.
5280 (evrp_dom_walker::before_dom_children): Store vr_values class
5281 instance into the vrp_folder class.
5282 (evrp_dom_walker::push_value_range): Rework to avoid direct
5283 access to num_vr_values and vr_value.
5284 (evrp_dom_walker::pop_value_range): Likewise.
5285 (execute_early_vrp): Remove call to vrp_initialize_lattice.
5286 Use vr_values to get to dump_all_value_ranges member function.
5287 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
5288 and simplify_cond_using_ranges_2 via vrp_prop class instance.
5289 Pass vr_values class instance down to identify_jump_threads.
5290 Remove call to vrp_free_lattice.
5291 (debug_all_value_ranges): Remove.
5293 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
5294 (vrp_folder): Likewise.
5296 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
5297 Get it from the existing bitmap instead.
5298 (vrp_intersect_ranges_1): Likewise.
5300 2017-11-09 Jakub Jelinek <jakub@redhat.com>
5302 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
5304 (store_immediate_info::store_immediate_info): Add bitnotp argument,
5305 set bit_not_p to it.
5306 (imm_store_chain_info::coalesce_immediate_stores): Break group
5307 if bit_not_p is different.
5308 (count_multiple_uses, split_group,
5309 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
5310 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
5311 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
5312 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
5313 to store_immediate_info ctor.
5315 2017-11-09 Jim Wilson <jimw@sifive.com>
5317 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
5318 (scan_prog_file): Likewise.
5320 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
5322 * bb-reorder.c (max_entry_frequency): Remove.
5323 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
5324 connect_traces, push_to_next_round_p): Remove prototypes.
5325 (find_traces_1_round): Use counts only.
5326 (push_to_next_round_p): Likewise.
5327 (find_traces): Likewise.
5328 (rotate_loop): Likewise.
5329 (find_traces_1_round): Likewise.
5330 (connect_traces): Likewise.
5331 (edge_order): Likewise.
5333 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
5335 * config/arm/arm.c (output_return_instruction): Add comments to
5336 indicate requirement for cmse_nonsecure_entry return to account
5337 for the size of clearing instruction output here.
5338 (thumb_exit): Likewise.
5339 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
5340 return in hardfloat mode.
5342 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
5344 * config/rs6000/rs6000.c (machine_function): Add a bool,
5345 "toc_is_wrapped_separately".
5346 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
5347 if it wasn't explicitly set or unset, we are optimizing for speed, and
5348 doing separate shrink-wrapping.
5349 (rs6000_get_separate_components): Enable the TOC component if
5350 saving the TOC register in the prologue.
5351 (rs6000_components_for_bb): Handle the TOC component.
5352 (rs6000_emit_prologue_components): Store the TOC register where needed.
5353 (rs6000_set_handled_components): Mark TOC as handled, if handled.
5354 (rs6000_emit_prologue): Don't save the TOC if that is already done.
5356 2017-11-09 Martin Jambor <mjambor@suse.cz>
5358 * ipa-param-manipulation.c: New file.
5359 * ipa-param-manipulation.h: Likewise.
5360 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
5361 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
5362 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
5363 (ipa_parm_adjustment): Likewise.
5364 (ipa_parm_adjustment_vec): Likewise.
5365 (ipa_get_vector_of_formal_parms): Moved declaration to
5366 ipa-param-manipulation.h.
5367 (ipa_get_vector_of_formal_parm_types): Likewise.
5368 (ipa_modify_formal_parameters): Likewise.
5369 (ipa_modify_call_arguments): Likewise.
5370 (ipa_combine_adjustments): Likewise.
5371 (ipa_dump_param_adjustments): Likewise.
5372 (ipa_modify_expr): Likewise.
5373 (ipa_get_adjustment_candidate): Likewise.
5374 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
5375 ipa-param-manipulation.c.
5376 (ipa_get_vector_of_formal_parm_types): Likewise.
5377 (ipa_modify_formal_parameters): Likewise.
5378 (ipa_modify_call_arguments): Likewise.
5379 (ipa_modify_expr): Likewise.
5380 (get_ssa_base_param): Likewise.
5381 (ipa_get_adjustment_candidate): Likewise.
5382 (index_in_adjustments_multiple_times_p): Likewise.
5383 (ipa_combine_adjustments): Likewise.
5384 (ipa_dump_param_adjustments): Likewise.
5385 * tree-sra.c: Also include ipa-param-manipulation.h
5386 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
5389 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5390 Alan Hayward <alan.hayward@arm.com>
5391 David Sherwood <david.sherwood@arm.com>
5393 * doc/sourcebuild.texi (vect_masked_store): Document.
5395 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5396 Alan Hayward <alan.hayward@arm.com>
5397 David Sherwood <david.sherwood@arm.com>
5399 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
5401 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5402 Alan Hayward <alan.hayward@arm.com>
5403 David Sherwood <david.sherwood@arm.com>
5405 * doc/sourcebuild.texi (vect_variable_length): Document.
5407 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5408 Alan Hayward <alan.hayward@arm.com>
5409 David Sherwood <david.sherwood@arm.com>
5411 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
5413 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5414 Alan Hayward <alan.hayward@arm.com>
5415 David Sherwood <david.sherwood@arm.com>
5417 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
5419 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5420 Alan Hayward <alan.hayward@arm.com>
5421 David Sherwood <david.sherwood@arm.com>
5423 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
5424 previously undocumented selectors.
5425 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
5427 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5429 * doc/rtl.texi (const_vector): Say that elements can be
5430 const_wide_ints too.
5431 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
5432 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
5433 (gen_vec_duplicate): Use it instead of CONSTANT_P.
5434 * optabs.c (expand_vector_broadcast): Likewise.
5436 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5437 Alan Hayward <alan.hayward@arm.com>
5438 David Sherwood <david.sherwood@arm.com>
5440 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
5441 scaled index even if the unscaled address was invalid.
5442 Don't increase the complexity of using a scale in that case.
5444 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5445 Alan Hayward <alan.hayward@arm.com>
5446 David Sherwood <david.sherwood@arm.com>
5448 * doc/rtl.texi: Rewrite the subreg rules so that they partition
5449 the inner register into REGMODE_NATURAL_SIZE bytes rather than
5450 UNITS_PER_WORD bytes.
5451 * emit-rtl.c (validate_subreg): Divide subregs into blocks
5452 based on REGMODE_NATURAL_SIZE of the inner mode.
5453 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
5454 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
5455 * expmed.c (lowpart_bit_field_p): Divide the value up into
5456 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
5457 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
5458 whether something is likely to occupy more than one register.
5460 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
5463 * ipa-inline-transform.c (update_noncloned_frequencies): Use
5464 profile_count::adjust_for_ipa_scaling.
5465 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
5466 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
5468 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
5470 2017-11-09 Jakub Jelinek <jakub@redhat.com>
5472 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
5473 (split_group): Add total_orig and total_new arguments, estimate the
5474 number of statements related to the store group without store merging
5475 and with store merging.
5476 (imm_store_chain_info::output_merged_store): Adjust split_group
5477 callers, punt if estimated number of statements with store merging
5478 is not smaller than estimated number of statements without it.
5480 (handled_load): Remove has_single_use checks.
5481 (pass_store_merging::process_store): Likewise.
5483 2017-11-09 Richard Biener <rguenther@suse.de>
5485 PR tree-optimization/82902
5486 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
5488 2017-11-09 Martin Liska <mliska@suse.cz>
5491 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
5494 2017-11-09 Martin Liska <mliska@suse.cz>
5496 PR tree-optimization/82669
5497 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
5499 2017-11-09 Martin Liska <mliska@suse.cz>
5501 PR gcov-profile/48463
5502 * coverage.c (coverage_begin_function): Output also end locus
5503 of a function and information whether the function is
5505 * gcov-dump.c (tag_function): Parse and print the information.
5506 * gcov.c (INCLUDE_MAP): Add include.
5507 (INCLUDE_SET): Likewise.
5508 (struct line_info): Move earlier in the source file because
5509 of vector<line_info> in function_info structure.
5510 (line_info::line_info): Likewise.
5511 (line_info::has_block): Likewise.
5512 (struct source_info): Add new member index.
5513 (source_info::get_functions_at_location): New function.
5514 (function_info::group_line_p): New function.
5515 (output_intermediate_line): New function.
5516 (output_intermediate_file): Use the mentioned function.
5517 (struct function_start): New.
5518 (struct function_start_pair_hash): Likewise.
5519 (process_file): Add code that identifies group functions.
5520 Assign lines either to global or function scope.
5521 (generate_results): Skip artificial functions.
5522 (find_source): Assign index for each source file.
5523 (read_graph_file): Read new flag artificial and end_line.
5524 (add_line_counts): Assign it either to global of function scope.
5525 (accumulate_line_counts): Isolate core of the function to
5526 accumulate_line_info and call it for both function and global
5528 (accumulate_line_info): New function.
5529 (output_line_beginning): Fix GNU coding style.
5530 (print_source_line): New function.
5531 (output_line_details): Likewise.
5532 (output_function_details): Likewise.
5533 (output_lines): Iterate both source (global) scope and function
5535 (struct function_line_start_cmp): New class.
5536 * doc/gcov.texi: Reflect changes in documentation.
5538 2017-11-09 Jakub Jelinek <jakub@redhat.com>
5541 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
5542 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
5543 and similarly for not instead of neg.
5545 2017-11-08 Andi Kleen <ak@linux.intel.com>
5547 * config/i386/i386.opt: Add -mforce-indirect-call.
5548 * config/i386/predicates.md: Check for flag_force_indirect_call.
5549 * doc/invoke.texi: Document -mforce-indirect-call
5551 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
5553 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
5555 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
5556 config/riscv/riscv.c (predict.h): New include.
5557 (riscv_slow_unaligned_access_p): No longer static.
5558 (riscv_block_move_straight): Add require.
5559 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
5561 2017-11-08 Jakub Jelinek <jakub@redhat.com>
5564 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
5565 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
5566 nonimmediate_operand predicate for operand 1 instead of
5569 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5571 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
5573 * config/aarch64/constraints.md (Uml): New constraint.
5574 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
5577 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5579 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
5580 of two vec_duplicates into a vec_concat.
5582 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5584 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
5585 Simplify vec_merge of vec_duplicate and vec_concat.
5586 * config/aarch64/constraints.md (Utq): New constraint.
5587 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
5590 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5592 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
5593 Simplify vec_merge of vec_duplicate and const_vector.
5594 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
5596 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
5597 mode iterator. Update predicate on operand 1 to
5598 handle non-const_vec constants. Delete constraints.
5599 (*aarch64_combinez_be<mode>): Likewise for operand 2.
5601 2017-11-08 Jakub Jelinek <jakub@redhat.com>
5603 PR tree-optimization/78821
5604 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
5606 (store_operand_info::store_operand_info): Initialize it to false.
5607 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
5608 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
5609 store in the group, and if chain_info is non-NULL, to ignore altogether
5611 (compatible_load_p): Fail if bit_not_p does not match.
5612 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
5613 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
5614 (pass_store_merging::process_store): Adjust
5615 terminate_all_aliasing_chains calls to pass NULL in all current spots,
5616 call terminate_all_aliasing_chains newly when adding a store into
5617 a chain with non-NULL chain_info.
5619 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
5621 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
5623 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
5625 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5627 (aarch64_layout_frame): Initialise emit_frame_chain.
5628 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
5629 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
5631 2017-11-08 Martin Liska <mliska@suse.cz>
5633 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
5634 of gimple_call_internal_p.
5636 2017-11-07 Tom de Vries <tom@codesourcery.com>
5638 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
5640 2017-11-07 Tom de Vries <tom@codesourcery.com>
5642 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5645 2017-11-08 Martin Liska <mliska@suse.cz>
5648 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
5650 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
5652 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
5654 2017-11-07 Jakub Jelinek <jakub@redhat.com>
5657 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
5658 (*cmp<mode>_ccz_1): New insn with $k alternative.
5661 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
5662 RTX_COMM_COMPARE as commutative as well.
5663 (ix86_binary_operator_ok): Formatting fix.
5664 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
5665 *<code><mode>3<mask_name><round_saeonly_name>,
5666 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
5667 *<s>mul<mode>3_highpart<mask_name>,
5668 *vec_widen_umult_even_v16si<mask_name>,
5669 *vec_widen_umult_even_v8si<mask_name>,
5670 *vec_widen_umult_even_v4si<mask_name>,
5671 *vec_widen_smult_even_v16si<mask_name>,
5672 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
5673 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
5674 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
5675 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
5676 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
5677 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
5678 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
5679 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
5680 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
5681 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
5682 ix86_binary_operator_ok. Formatting fixes.
5683 (*<plusminus_insn><mode>3<mask_name><round_name>,
5684 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
5687 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
5689 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
5690 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
5691 eq and ne if TARGET_ISEL.
5692 (cmp): New code_iterator.
5693 (UNS, UNSU_, UNSIK): New code_attrs.
5694 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
5695 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
5697 ("eq<mode>3"): ... this.
5698 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
5700 ("ne<mode>3"): ... this.
5702 2017-11-07 Julia Koval <julia.koval@intel.com>
5705 * common/config/i386/i386-common.c
5706 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
5707 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
5708 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
5709 * config/i386/i386.opt: Ditto.
5710 * config/i386/i386.c (ix86_target_string): Ditto.
5711 (ix86_option_override_internal): Ditto.
5712 (ix86_init_mpx_builtins): Move MPX to args2.
5713 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
5714 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
5715 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
5716 __builtin_ia32_vgf2p8affineinvqb_v32qi,
5717 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
5718 __builtin_ia32_vgf2p8affineinvqb_v16qi,
5719 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
5721 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
5724 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
5725 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
5726 (zero-extendsidi peephole2): Remove peephole.
5728 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
5731 * stor-layout.c: Include attribs.h.
5732 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
5733 explicit lookup of "aligned" attribute.
5735 2017-11-07 Andrew Waterman <andrew@sifive.com>
5737 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
5738 (riscv_expand_block_move): Likewise.
5739 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
5741 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
5742 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
5743 * config/riscv/riscv.c (riscv_block_move_straight): New function.
5744 (riscv_adjust_block_mem): Likewise.
5745 (riscv_block_move_loop): Likewise.
5746 (riscv_expand_block_move): Likewise.
5747 * config/riscv/riscv.md (movmemsi): New pattern.
5749 2017-11-07 Michael Clark <michaeljclark@mac.com>
5751 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
5752 (MUSL_DYNAMIC_LINKER): Likewise.
5754 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
5756 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
5757 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
5760 2017-11-07 Richard Biener <rguenther@suse.de>
5762 * match.pd: Fix build.
5764 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
5765 Jackson Woodruff <jackson.woodruff@arm.com>
5767 PR tree-optimization/71026
5768 * match.pd: Canonicalize negate in division.
5770 2017-11-07 Sudakshina Das <sudi.das@arm.com>
5773 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
5775 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5777 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
5778 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
5780 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5782 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
5783 non-scalar integral types.
5784 * match.pd (negate_expr_p): Handle MINUS_EXPR.
5785 (-(A-B), -(~A)): New transformations.
5787 2017-11-07 Tom de Vries <tom@codesourcery.com>
5789 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
5790 semicolon after "do {} while (0)".
5791 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5792 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5793 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5794 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5795 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5797 2017-11-07 Tom de Vries <tom@codesourcery.com>
5799 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
5800 after "do {} while (0)".
5801 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5802 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5803 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5804 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5805 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5807 2017-11-07 Tom de Vries <tom@codesourcery.com>
5810 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
5812 (arm_rtx_costs_internal): Add missing semicolon after
5813 HANDLE_NARROW_SHIFT_ARITH call.
5815 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5817 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
5818 disable isel if it was not set explicitly.
5820 2017-11-06 James Bowman <james.bowman@ftdichip.com>
5822 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
5823 (add_type_attribute) likewise.
5825 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
5827 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
5828 of struct ix86_frame.
5829 (ix86_initial_elimination_offset): Likewise.
5830 (ix86_expand_split_stack_prologue): Likewise.
5832 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
5834 * tree-vrp.h (enum value_range_type): Update stale comment.
5836 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5837 Alan Hayward <alan.hayward@arm.com>
5838 David Sherwood <david.sherwood@arm.com>
5840 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
5841 (aarch64_expand_vec_perm_const): Take the number of units too.
5842 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
5843 (aarch64_expand_vec_perm_const): Likewise.
5844 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
5845 (vec_perm<mode>): Update accordingly.
5847 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5848 Alan Hayward <alan.hayward@arm.com>
5849 David Sherwood <david.sherwood@arm.com>
5851 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
5852 Take the number of units too.
5853 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
5854 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
5855 but check for a vector mode before rather than after the call.
5856 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
5857 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
5858 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
5859 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
5860 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
5861 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
5862 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
5863 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
5864 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
5865 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
5866 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
5867 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
5868 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
5869 (aarch64_sqdmull2_n<mode>): Update accordingly.
5871 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5872 Alan Hayward <alan.hayward@arm.com>
5873 David Sherwood <david.sherwood@arm.com>
5875 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
5876 the number of units too.
5877 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
5878 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
5879 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
5880 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
5881 (vec_store_lanesxi<mode>): Update accordingly.
5883 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5884 Alan Hayward <alan.hayward@arm.com>
5885 David Sherwood <david.sherwood@arm.com>
5887 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
5888 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
5889 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
5890 of units rather than the mode.
5891 * config/aarch64/iterators.md (nunits): New mode attribute.
5892 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
5893 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
5894 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
5895 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
5896 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5897 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
5898 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
5899 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
5900 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
5901 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
5902 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5903 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
5904 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
5905 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
5906 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
5907 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
5908 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
5909 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
5910 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
5911 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
5912 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
5913 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
5914 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
5915 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
5916 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
5917 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
5918 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
5919 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
5920 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
5921 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
5922 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
5923 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
5924 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
5925 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
5926 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
5927 (aarch64_simd_vec_setv2di): Likewise.
5929 2017-11-06 Carl Love <cel@us.ibm.com>
5931 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
5933 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
5934 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
5936 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
5938 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
5939 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
5940 Add power 8 macro expansions.
5941 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
5942 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
5943 power 8 instructions. (VSX_XXBR): Add iterator.
5945 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
5947 * config/arm/arm.md (predicable_short_it): Change default to "no",
5948 improve documentation, remove uses that are identical to the default.
5949 (enabled_for_depr_it): Rename to enabled_for_short_it.
5950 * gcc/config/arm/arm-fixed.md (predicable_short_it):
5951 Remove default uses.
5952 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
5953 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
5954 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
5955 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
5957 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
5960 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
5961 float128 helper macros, which are no longer used after deleting
5962 the old 'q' built-in functions, and moving the round to odd
5963 built-in functions to being special built-in functions.
5964 (BU_FLOAT128_2): Likewise.
5965 (BU_FLOAT128_1_HW): Likewise.
5966 (BU_FLOAT128_2_HW): Likewise.
5967 (BU_FLOAT128_3_HW): Likewise.
5968 (FABSQ): Delete old 'q' built-in functions.
5969 (COPYSIGNQ): Likewise.
5970 (SQRTF128_ODD): Move round to odd built-in functions to be
5971 special built-in functions, so that we can handle
5972 -mabi=ieeelongdouble.
5973 (TRUNCF128_ODD): Likewise.
5974 (ADDF128_ODD): Likewise.
5975 (SUBF128_ODD): Likewise.
5976 (MULF128_ODD): Likewise.
5977 (DIVF128_ODD): Likewise.
5978 (FMAF128_ODD): Likewise.
5979 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
5980 built-in names to 'f128'.
5981 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
5982 old 'q' built-in functions, as the machine independent code for
5983 'f128' built-in functions handles this.
5984 (rs6000_expand_builtin): Add expansion for float128 round to odd
5985 functions, keying off on -mabi=ieeelongdouble of whether to use
5986 the KFmode or TFmode variant.
5987 (rs6000_init_builtins): Initialize the _Float128 round to odd
5989 * doc/extend.texi (PowerPC Built-in Functions): Document the old
5990 _Float128 'q' built-in functions are now mapped into the new
5991 'f128' built-in functions.
5993 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
5995 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
5996 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
5998 2017-11-06 David Malcolm <dmalcolm@redhat.com>
6001 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
6002 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
6003 * toplev.c: Include "ipa-fnsummary.h".
6004 (toplev::finalize): Call ipa_fnsummary_c_finalize.
6006 2017-11-06 Jakub Jelinek <jakub@redhat.com>
6008 PR tree-optimization/82838
6009 * gimple-ssa-store-merging.c
6010 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
6011 on a separate gimple_seq which is then appended to seq.
6013 2017-11-06 Jeff Law <law@redhat.com>
6016 * config/i386/i386.c (PROBE_INTERVAL): Remove.
6017 (get_probe_interval): New functions.
6018 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
6019 (ix86_adjust_stack_and_probe): Likewise.
6020 (output_adjust_stack_and_probe): Likewise.
6021 (ix86_emit_probe_stack_range): Likewise.
6022 (ix86_expand_prologue): Likewise.
6024 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6026 PR tree-optimization/82816
6027 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
6028 if the modes of the two types are the same.
6029 (convert_plusminus_to_widen): Likewise.
6031 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6033 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
6035 (usadv16qi): New define_expand.
6036 (usadv8hi): New define_expand.
6038 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
6041 * ipa-inline-transform.c (update_noncloned_frequencies): Always
6043 (inline_transform): Likewise.
6044 * predict.c (counts_to_freqs): Remove useless conditional.
6045 * profile-count.h (profile_count::apply_scale): Move sanity check.
6046 * tree-inline.c (copy_bb): Always scale.
6047 (copy_cfg_body): Likewise.
6049 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
6052 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
6054 (*cmp_ite0): Add enabled_for_depr_it attribute.
6055 (*cmp_ite1): Likewise.
6057 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
6059 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
6062 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6064 * tree-vrp.c (vrp_int_const_binop): Return true on success and
6065 return the value by pointer.
6066 (extract_range_from_multiplicative_op_1): Update accordingly.
6067 Return as soon as an operation fails.
6069 2017-11-05 Tom de Vries <tom@codesourcery.com>
6072 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
6073 (DEF_SANITIZER_BUILTIN): ... here.
6074 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
6075 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
6077 2017-11-05 Tom de Vries <tom@codesourcery.com>
6080 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
6082 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
6083 ASM_OUTPUT_BEFORE_CASE_LABEL call.
6084 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
6086 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
6087 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
6088 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
6090 2017-11-05 Tom de Vries <tom@codesourcery.com>
6093 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
6096 2017-11-04 Michael Clark <michaeljclark@mac.com>
6098 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
6099 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
6101 (*addsi3_extended): Likewise.
6102 (*addsi3_extended2): Likewise.
6103 (<optab>si3): Likewise.
6104 (<optab>di3): Likewise.
6105 (<optab><mode>3): Likewise.
6106 (<*optabe>si3_internal): Likewise.
6107 (zero_extendqi<SUPERQI:mode>2): Likewise.
6108 (*add<mode>hi3): Likewise.
6109 (*xor<mode>hi3): Likewise.
6110 (<optab>di3): Likewise.
6111 (*<optab>si3_extend): Likewise.
6112 (*sge<u>_<X:mode><GPR:mode>): Likewise.
6113 (*slt<u>_<X:mode><GPR:mode>): Likewise.
6114 (*sle<u>_<X:mode><GPR:mode>): Likewise.
6116 2017-11-04 Andrew Waterman <andrew@sifive.com>
6118 * config/riscv/riscv.c (riscv_option_override): Conditionally set
6119 TARGET_STRICT_ALIGN based upon -mtune argument.
6121 2017-11-04 Andrew Waterman <andrew@sifive.com>
6123 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
6125 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
6127 * config/i386/i386.c (choose_basereg): Use optional scratch
6128 register and add assertion.
6129 (x86_emit_outlined_ms2sysv_save): Use scratch register when
6130 needed, and don't allocate stack.
6131 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
6132 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
6133 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
6135 2017-11-03 Jeff Law <law@redhat.com>
6137 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
6138 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
6139 to probe at the start of a noreturn function.
6141 2017-11-03 Jakub Jelinek <jakub@redhat.com>
6143 PR tree-optimization/78821
6144 * gimple-ssa-store-merging.c: Update the file comment.
6145 (MAX_STORE_ALIAS_CHECKS): Define.
6146 (struct store_operand_info): New type.
6147 (store_operand_info::store_operand_info): New constructor.
6148 (struct store_immediate_info): Add rhs_code and ops data members.
6149 (store_immediate_info::store_immediate_info): Add rhscode, op0r
6150 and op1r arguments to the ctor, initialize corresponding data members.
6151 (struct merged_store_group): Add load_align_base and load_align
6153 (merged_store_group::merged_store_group): Initialize them.
6154 (merged_store_group::do_merge): Update them.
6155 (merged_store_group::apply_stores): Pick the constant for
6156 encode_tree_to_bitpos from one of the two operands, or skip
6157 encode_tree_to_bitpos if neither operand is a constant.
6158 (class pass_store_merging): Add process_store method decl. Remove
6159 bool argument from terminate_all_aliasing_chains method decl.
6160 (pass_store_merging::terminate_all_aliasing_chains): Remove
6161 var_offset_p argument and corresponding handling.
6162 (stmts_may_clobber_ref_p): New function.
6163 (compatible_load_p): New function.
6164 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
6165 if there is overlap and rhs_code is not INTEGER_CST. For
6166 non-overlapping stores terminate group if rhs is not mergeable.
6167 (get_alias_type_for_stmts): Change first argument from
6168 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
6169 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
6170 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
6172 (get_location_for_stmts): Change first argument from
6173 auto_vec<gimple *> & to vec<gimple *> &.
6174 (struct split_store): Remove orig_stmts data member, add orig_stores.
6175 (split_store::split_store): Create orig_stores rather than orig_stmts.
6176 (find_constituent_stmts): Renamed to ...
6177 (find_constituent_stores): ... this. Change second argument from
6178 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
6179 to info structures rather than the statements.
6180 (split_group): Rename ALLOW_UNALIGNED argument to
6181 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
6182 it. Adjust find_constituent_stores caller.
6183 (imm_store_chain_info::output_merged_store): Handle rhs_code other
6184 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
6185 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
6186 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
6187 (mem_valid_for_store_merging): New function.
6188 (handled_load): New function.
6189 (pass_store_merging::process_store): New method.
6190 (pass_store_merging::execute): Use process_store method. Adjust
6191 terminate_all_aliasing_chains caller.
6193 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6195 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
6196 Return true for more constants, symbols and label references.
6197 (aarch64_valid_floating_const): Remove unused function.
6199 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6202 * config/aarch64/aarch64.c (aarch64_layout_frame):
6203 Undo forcing of LR at bottom of frame.
6205 2017-11-03 Jeff Law <law@redhat.com>
6208 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
6209 for int_registers_saved.
6211 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
6212 extracted from tree-ssa-dom.c.
6213 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
6214 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
6215 (record_equivalences_from_incoming_edge): Add additional argument
6216 to single_pred_edge_ignoring_loop_edges call.
6217 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
6218 (uncprop_dom_walker::before_dom_children): Add additional argument
6219 to single_pred_edge_ignoring_loop_edges call.
6220 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
6221 single_pred_edge_ignoring_loop_edges rather than open coding.
6222 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
6224 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
6226 * match.pd (-(-A)): Rewrite.
6228 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
6230 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
6231 (rs6000_emit_int_cmove): New declaration.
6232 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
6233 (rs6000_emit_sISEL): Delete.
6234 (rs6000_emit_int_cmove): Make non-static.
6235 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
6236 instead of rs6000_emit_sISEL.
6238 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
6240 * asan.c (create_cond_insert_point): Maintain profile.
6241 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
6243 * basic-block.h (struct basic_block_def): Remove frequency.
6244 (EDGE_FREQUENCY): Use to_frequency
6245 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
6247 (find_traces): Update to use to_frequency.
6248 (find_traces_1_round): Likewise; use only IPA counts.
6249 (bb_to_key): Likewise.
6250 (connect_traces): Use IPA counts only.
6251 (copy_bb_p): Update to use to_frequency.
6252 (fix_up_crossing_landing_pad): Likewise.
6253 (sanitize_hot_paths): Likewise.
6254 * bt-load.c (basic_block_freq): Likewise.
6255 * cfg.c (init_flow): Set count_max to uninitialized.
6256 (check_bb_profile): Remove frequencies; check counts.
6257 (dump_bb_info): Do not dump frequencies.
6258 (update_bb_profile_for_threading): Update counts only.
6259 (scale_bbs_frequencies_int): Likewise.
6260 (MAX_SAFE_MULTIPLIER): Remove.
6261 (scale_bbs_frequencies_gcov_type): Update counts only.
6262 (scale_bbs_frequencies_profile_count): Update counts only.
6263 (scale_bbs_frequencies): Update counts only.
6264 * cfg.h (struct control_flow_graph): Add count-max.
6265 (update_bb_profile_for_threading): Update prototype.
6266 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
6267 (find_many_sub_basic_blocks): Likewise.
6268 * cfgcleanup.c (try_forward_edges): Likewise.
6269 (try_crossjump_to_edge): Likewise.
6270 * cfgexpand.c (expand_gimple_cond): Likewise.
6271 (expand_gimple_tailcall): Likewise.
6272 (construct_init_block): Likewise.
6273 (construct_exit_block): Likewise.
6274 * cfghooks.c (verify_flow_info): Check consistency of counts.
6275 (dump_bb_for_graph): Do not dump frequencies.
6276 (split_block_1): Do not update frequencies.
6277 (split_edge): Do not update frequencies.
6278 (make_forwarder_block): Do not update frequencies.
6279 (duplicate_block): Do not update frequencies.
6280 (account_profile_record): Do not update frequencies.
6281 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
6282 for global heuristics.
6283 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
6284 (expected_loop_iterations_unbounded): Use counts only.
6285 * cfgloopmanip.c (scale_loop_profile): Simplify.
6286 (create_empty_loop_on_edge): Simplify
6288 (duplicate_loop_to_header_edge): Simplify
6289 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
6290 (update_br_prob_note): Take care of removing note when profile
6292 (relink_block_chain): Do not dump frequency.
6293 (rtl_account_profile_record): Use to_frequency.
6294 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
6295 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
6296 (cgraph_update_edges_for_call_stmt_node): Likewise.
6297 (cgraph_edge::verify_count_and_frequency): Update.
6298 (cgraph_node::verify_node): Temporarily disable frequency verification.
6299 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
6300 to_cgraph_frequency.
6301 (cgraph_edge::rebuild_edges): Convert to ipa counts.
6302 * cgraphunit.c (init_lowered_empty_function): Do not initialize
6304 (cgraph_node::expand_thunk): Update profile.
6305 * except.c (dw2_build_landing_pads): Do not update frequency.
6306 * final.c (compute_alignments): Use to_frequency.
6307 (dump_basic_block_info): Do not dump frequency.
6308 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
6309 (dump_gimple_bb_header): Do not dump frequency.
6310 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
6312 * gimple-streamer-in.c (input_bb): Do not stream frequency.
6313 * gimple-streamer-out.c (output_bb): Do not stream frequency.
6314 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
6315 (init_before_recovery): Do not update frequency.
6316 (sched_create_recovery_edges): Do not update frequency.
6317 * hsa-gen.c (convert_switch_statements): Do not update frequency.
6318 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
6319 (ipa_cp_c_finalize): Set max_count to uninitialized.
6320 * ipa-fnsummary.c (get_minimal_bb): Use counts.
6321 (param_change_prob): Use counts.
6322 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
6324 * ipa-split.c (consider_split): Use to_frequency.
6325 (split_function): Use to_frequency.
6326 * ira-build.c (loop_compare_func): Likewise.
6327 (mark_loops_for_removal): Likewise.
6328 (mark_all_loops_for_removal): Likewise.
6329 * loop-doloop.c (doloop_modify): Do not update frequency.
6330 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
6332 * lto-streamer-in.c (input_function): Update count_max.
6333 * omp-expand.c (expand_omp_taskreg): Update count_max.
6334 * omp-simd-clone.c (simd_clone_adjust): Update profile.
6335 * predict.c (maybe_hot_frequency_p): Use to_frequency.
6336 (maybe_hot_count_p): Use ipa counts only.
6337 (maybe_hot_bb_p): Simplify.
6338 (maybe_hot_edge_p): Simplify.
6339 (probably_never_executed): Do not take frequency argument.
6340 (probably_never_executed_bb_p): Do not pass frequency.
6341 (probably_never_executed_edge_p): Likewise.
6342 (combine_predictions_for_bb): Check that profile is nonzero.
6343 (propagate_freq): Do not set frequency.
6344 (drop_profile): Simplify.
6345 (counts_to_freqs): Simplify.
6346 (expensive_function_p): Use to_frequency.
6347 (propagate_unlikely_bbs_forward): Simplify.
6348 (determine_unlikely_bbs): Simplify.
6349 (estimate_bb_frequencies): Add hack to silence graphite issues.
6350 (compute_function_frequency): Use ipa counts.
6351 (pass_profile::execute): Update.
6352 (rebuild_frequencies): Use counts only.
6353 (force_edge_cold): Use counts only.
6354 * profile-count.c (profile_count::dump): Dump new count types.
6355 (profile_count::differs_from_p): Check compatiblity.
6356 (profile_count::to_frequency): New function.
6357 (profile_count::to_cgraph_frequency): New function.
6358 * profile-count.h (struct function): Declare.
6359 (enum profile_quality): Add profile_guessed_local and
6360 profile_guessed_global0.
6361 (class profile_proability): Decrease number of bits to 29;
6362 update from_reg_br_prob_note and to_reg_br_prob_note.
6363 (class profile_count: Update comment; decrease number of bits
6364 to 61. Check compatibility.
6365 (profile_count::compatible_p): New private member function.
6366 (profile_count::ipa_p): New member function.
6367 (profile_count::operator<): Handle global zero correctly.
6368 (profile_count::operator>): Handle global zero correctly.
6369 (profile_count::operator<=): Handle global zero correctly.
6370 (profile_count::operator>=): Handle global zero correctly.
6371 (profile_count::nonzero_p): New member function.
6372 (profile_count::force_nonzero): New member function.
6373 (profile_count::max): New member function.
6374 (profile_count::apply_scale): Handle IPA scalling.
6375 (profile_count::guessed_local): New member function.
6376 (profile_count::global0): New member function.
6377 (profile_count::ipa): New member function.
6378 (profile_count::to_frequency): Declare.
6379 (profile_count::to_cgraph_frequency): Declare.
6380 * profile.c (OVERLAP_BASE): Delete.
6381 (compute_frequency_overlap): Delete.
6382 (compute_branch_probabilities): Do not use compute_frequency_overlap.
6383 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
6384 * sched-ebb.c (rank): Use counts only.
6385 * shrink-wrap.c (handle_simple_exit): Use counts only.
6386 (try_shrink_wrapping): Use counts only.
6387 (place_prologue_for_one_component): Use counts only.
6388 * tracer.c (find_best_predecessor): Use to_frequency.
6389 (find_trace): Use to_frequency.
6390 (tail_duplicate): Use to_frequency.
6391 * trans-mem.c (expand_transaction): Do not update frequency.
6392 * tree-call-cdce.c: Do not update frequency.
6393 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
6394 (gimple_merge_blocks): Likewise.
6395 (gimple_split_edge): Likewise.
6396 (gimple_duplicate_sese_region): Likewise.
6397 (gimple_duplicate_sese_tail): Likewise.
6398 (move_sese_region_to_fn): Likewise.
6399 (gimple_account_profile_record): Likewise.
6400 (insert_cond_bb): Likewise.
6401 * tree-complex.c (expand_complex_div_wide): Likewise.
6402 * tree-eh.c (lower_resx): Update profile.
6403 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
6405 (initialize_cfun): Do not initialize frequencies
6406 (freqs_to_counts): Delete.
6407 (copy_cfg_body): Ignore count parameter.
6408 (copy_body): Update.
6409 (expand_call_inline): Update count_max.
6410 (optimize_inline_calls): Update count_max.
6411 (tree_function_versioning): Update count_max.
6412 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
6413 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
6415 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
6416 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
6417 (try_peel_loop): Likewise.
6418 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
6420 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
6421 (tree_transform_and_unroll_loop): Do not use frequencies
6422 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
6423 Use reliable prediction only.
6424 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
6425 * tree-ssa-sink.c (select_best_block): Use to_frequency.
6426 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
6427 probability scaling.
6428 * tree-ssa-threadupdate.c (create_block_for_threading): Do
6429 not update frequency
6430 (any_remaining_duplicated_blocks): Likewise.
6431 (update_profile): Likewise.
6432 (estimated_freqs_path): Delete.
6433 (freqs_to_counts_path): Delete.
6434 (clear_counts_path): Delete.
6435 (ssa_fix_duplicate_block_edges): Likewise.
6436 (duplicate_thread_path): Likewise.
6437 * tree-switch-conversion.c (gen_inbound_check): Use counts.
6438 * tree-tailcall.c (decrease_profile): Do not update frequency.
6439 (eliminate_tail_call): Likewise.
6440 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
6441 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
6442 (optimize_mask_stores): Likewise.
6443 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
6444 * ubsan.c (ubsan_expand_null_ifn): Update profile.
6445 (ubsan_expand_ptr_ifn): Update profile.
6446 * value-prof.c (gimple_ic): Simplify.
6447 * value-prof.h (gimple_ic): Update prototype.
6448 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
6449 * ipa-inline.c (compute_uninlined_call_time): Be sure that
6451 (want_inline_self_recursive_call_p): Likewise.
6452 (resolve_noninline_speculation): Only cummulate defined counts.
6453 (inline_small_functions): Use nonzero_p.
6454 (ipa_inline): Do not access freed node.
6456 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6458 * config/aarch64/aarch64.c (aarch64_override_options_internal):
6459 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
6461 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
6463 * config/riscv/riscv.c (riscv_legitimize_move): Handle
6464 non-legitimate address.
6466 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
6468 * config/rs6000/rs6000.md (*lt0_disi): Delete.
6469 (*lt0_<mode>di, *lt0_<mode>si): New.
6471 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
6473 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
6474 TARGET_PAIRED_FLOAT.
6476 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
6477 Jim Wilson <jim.wilson@linaro.org>
6479 * config/aarch64/aarch64-cores.def (saphira): New CPU.
6480 * config/aarch64/aarch64-tune.md: Regenerated.
6481 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
6482 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
6484 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
6486 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
6487 (arc_expand_prologue): Restore blink for millicode.
6488 * config/arc/linux.h (LINK_EH_SPEC): Defined.
6490 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
6493 * config/i386/i386.c (ix86_vector_duplicate_value): Use
6494 gen_vec_duplicate after forcing the scalar into a register.
6496 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
6498 * combine (try_combine): Print the insns input to try_combine to the
6501 2017-11-02 Steve Ellcey <sellcey@cavium.com>
6504 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
6505 Remove second argument from aarch64_process_target_attr call.
6506 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
6508 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
6510 (aarch64_handle_attr_arch): Remove second argument.
6511 (aarch64_handle_attr_cpu): Ditto.
6512 (aarch64_handle_attr_tune): Ditto.
6513 (aarch64_handle_attr_isa_flags): Ditto.
6514 (aarch64_process_one_target_attr): Ditto.
6515 (aarch64_process_target_attr): Ditto.
6516 (aarch64_option_valid_attribute_p): Remove second argument.
6517 on aarch64_process_target_attr call.
6519 2017-11-02 David Malcolm <dmalcolm@redhat.com>
6521 * diagnostic.c: Include "selftest-diagnostic.h".
6522 (selftest::assert_location_text): New function.
6523 (selftest::test_diagnostic_get_location_text): New function.
6524 (selftest::diagnostic_c_tests): Call it.
6526 2017-11-02 David Malcolm <dmalcolm@redhat.com>
6528 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
6529 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
6530 (class selftest::test_diagnostic_context): Move to...
6531 * selftest-diagnostic.c: New file.
6532 * selftest-diagnostic.h: New file.
6534 2017-11-02 James Bowman <james.bowman@ftdichip.com>
6536 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
6537 offset range for FT32B.
6538 * config/ft32/ft32.h: option "mcompress" enables relaxation.
6539 * config/ft32/ft32.md: Add TARGET_NOPM.
6540 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
6541 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
6543 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
6545 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
6547 2017-11-02 Jeff Law <law@redhat.com>
6549 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
6550 virtual keyword on FINAL OVERRIDE members.
6552 * tree-ssa-propagate.h (ssa_propagation_engine): Group
6553 virtuals together. Add virtual destructor.
6554 (substitute_and_fold_engine): Similarly.
6556 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
6558 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
6560 2017-11-02 Richard Biener <rguenther@suse.de>
6562 PR tree-optimization/82795
6563 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
6565 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6567 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
6569 (gcc_GAS_CHECK_FEATURE): Remove.
6570 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
6571 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
6572 * configure: Regenerate.
6574 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
6576 * config/arc/arc.c (hwloop_optimize): Account for empty
6579 2017-11-02 Richard Biener <rguenther@suse.de>
6582 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
6583 Truncate ARRAY_REF index and element size.
6585 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
6587 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
6589 2017-11-01 Jeff Law <law@redhat.com>
6591 * tree-ssa-ccp.c (ccp_folder): New class derived from
6592 substitute_and_fold_engine.
6593 (ccp_folder::get_value): New member function.
6594 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
6595 (ccp_fold_stmt): Remove prototype.
6596 (ccp_finalize): Call substitute_and_fold from the ccp_class.
6597 * tree-ssa-copy.c (copy_folder): New class derived from
6598 substitute_and_fold_engine.
6599 (copy_folder::get_value): Renamed from get_value.
6600 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
6601 * tree-vrp.c (vrp_folder): New class derived from
6602 substitute_and_fold_engine.
6603 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
6604 (vrp_folder::get_value): New member function.
6605 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
6606 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
6607 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
6608 provide a class interface to folder/substitute routines.
6609 (ssa_prop_fold_stmt_fn): Remove typedef.
6610 (ssa_prop_get_value_fn): Likewise.
6611 (subsitute_and_fold): Remove prototype.
6612 (replace_uses_in): Likewise.
6613 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
6614 Renamed from replace_uses_in. Call the virtual member function
6615 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
6616 (substitute_and_fold_dom_walker): Remove initialization of
6617 data member entries for calbacks. Add substitute_and_fold_engine
6618 member and initialize it.
6619 (substitute_and_fold_dom_walker::before_dom_children0: Use the
6620 member functions for get_value, replace_phi_args_in c
6621 replace_uses_in, and fold_stmt calls.
6622 (substitute_and_fold_engine::substitute_and_fold): Renamed from
6623 substitute_and_fold. Remove assert. Update ctor call.
6625 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
6626 (ssa_prop_visit_phi_fn): Likewise.
6627 (class ssa_propagation_engine): New class to provide an interface
6629 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
6631 (ssa_prop_visit_phi): Likewise.
6632 (ssa_propagation_engine::simulate_stmt): Moved into class.
6633 Call visit_phi/visit_stmt from the class rather than via
6634 file scoped static variables.
6635 (ssa_propagation_engine::simulate_block): Moved into class.
6636 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
6637 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
6638 set file scoped statics for the visit_stmt/visit_phi callbacks.
6639 * tree-complex.c (complex_propagate): New class derived from
6640 ssa_propagation_engine.
6641 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
6642 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
6643 (tree_lower_complex): Call ssa_propagate via the complex_propagate
6645 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
6646 ssa_propagation_engine.
6647 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
6648 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
6649 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
6650 * tree-ssa-copy.c (copy_prop): New class derived from
6651 ssa_propagation_engine.
6652 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
6653 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
6654 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
6655 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
6656 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
6657 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
6658 (execute_vrp): Call ssa_propagate from the vrp_prop class.
6660 2017-11-01 Jakub Jelinek <jakub@redhat.com>
6662 PR rtl-optimization/82778
6663 PR rtl-optimization/82597
6664 * compare-elim.c (struct comparison): Add in_a_setter field.
6665 (find_comparison_dom_walker::before_dom_children): Remove killed
6666 bitmap and df_simulate_find_defs call, instead walk the defs.
6667 Compute last_setter and initialize in_a_setter. Merge definitions
6668 with first initialization for a few variables.
6669 (try_validate_parallel): Use insn_invalid_p instead of
6670 recog_memoized. Return insn rather than just the pattern.
6671 (try_merge_compare): Fix up comment. Don't uselessly test if
6672 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
6674 (execute_compare_elim_after_reload): Remove df_chain_add_problem
6677 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6678 Alan Hayward <alan.hayward@arm.com>
6679 David Sherwood <david.sherwood@arm.com>
6681 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
6682 aarch64_hard_regno_nregs to get the number of registers
6685 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6686 Alan Hayward <alan.hayward@arm.com>
6687 David Sherwood <david.sherwood@arm.com>
6689 * config/aarch64/constraints.md (Upl): Rename to...
6691 * config/aarch64/aarch64.md
6692 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
6695 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6696 Alan Hayward <alan.hayward@arm.com>
6697 David Sherwood <david.sherwood@arm.com>
6699 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
6700 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
6703 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6704 Alan Hayward <alan.hayward@arm.com>
6705 David Sherwood <david.sherwood@arm.com>
6707 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
6708 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
6709 (aarch64_evpc_dup): Generate rtl direcly, rather than using
6711 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
6712 of a single element.
6713 * config/aarch64/iterators.md: Add a comment above the permute
6714 unspecs to say that they are generated directly by
6715 aarch64_expand_vec_perm_const.
6716 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
6718 2017-11-01 Nathan Sidwell <nathan@acm.org>
6720 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
6722 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
6724 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
6725 and medany code models, and describe what they do.
6727 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6729 Revert accidental duplicate:
6731 * combine.c (can_change_dest_mode): Reject changes in
6732 REGMODE_NATURAL_SIZE.
6734 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
6736 PR rtl-optimization/64682
6737 PR rtl-optimization/69567
6738 PR rtl-optimization/69737
6739 PR rtl-optimization/82683
6740 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
6741 register mentioned in the note, drop the note, unless it came from I3,
6742 in which case it should go to I3 again.
6744 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6746 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
6747 and return false if not.
6748 (clear_bytes_written_by, live_bytes_read): Update accordingly.
6750 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6752 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
6753 range is known to be empty.
6755 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6756 Alan Hayward <alan.hayward@arm.com>
6757 David Sherwood <david.sherwood@arm.com>
6759 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
6760 and CONST_VECTOR_NUNITS instead of computing the number of units from
6761 the byte sizes of the vector and element.
6762 (simplify_binary_operation_1): Likewise.
6763 (simplify_const_binary_operation): Likewise.
6764 (simplify_ternary_operation): Likewise.
6766 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6767 Alan Hayward <alan.hayward@arm.com>
6768 David Sherwood <david.sherwood@arm.com>
6770 * var-tracking.c (INT_MEM_OFFSET): Replace with...
6771 (int_mem_offset): ...this new function.
6772 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
6773 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
6774 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
6777 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6778 Alan Hayward <alan.hayward@arm.com>
6779 David Sherwood <david.sherwood@arm.com>
6781 * lower-subreg.c (interesting_mode_p): New function.
6782 (compute_costs, find_decomposable_subregs, decompose_register)
6783 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
6784 (resolve_clobber, dump_choices): Use it.
6786 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6787 Alan Hayward <alan.hayward@arm.com>
6788 David Sherwood <david.sherwood@arm.com>
6790 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
6792 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6793 Alan Hayward <alan.hayward@arm.com>
6794 David Sherwood <david.sherwood@arm.com>
6796 * alias.c (find_base_value, find_base_term): Only process integer
6797 truncations. Check the precision rather than the size.
6799 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6800 Alan Hayward <alan.hayward@arm.com>
6801 David Sherwood <david.sherwood@arm.com>
6803 * machmode.h (is_narrower_int_mode): New function
6804 * optabs.c (expand_float, expand_fix): Use it.
6805 * dwarf2out.c (rotate_loc_descriptor): Likewise.
6807 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6808 Alan Hayward <alan.hayward@arm.com>
6809 David Sherwood <david.sherwood@arm.com>
6811 * rtl.h (narrower_subreg_mode): New function.
6812 * ira-color.c (update_costs_from_allocno): Use it.
6814 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6815 Alan Hayward <alan.hayward@arm.com>
6816 David Sherwood <david.sherwood@arm.com>
6818 * optabs-query.h (convert_optab_p): New function, split out from...
6819 (convert_optab_handler): ...here.
6820 (widening_optab_handler): Delete.
6821 (find_widening_optab_handler): Remove permit_non_widening parameter.
6822 (find_widening_optab_handler_and_mode): Likewise. Provide an
6823 override that operates on mode class wrappers.
6824 * optabs-query.c (widening_optab_handler): Delete.
6825 (find_widening_optab_handler_and_mode): Remove permit_non_widening
6826 parameter. Assert that the two modes are the same class and that
6827 the "from" mode is narrower than the "to" mode. Use
6828 convert_optab_handler instead of widening_optab_handler.
6829 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
6830 instead of widening_optab_handler.
6831 * expr.c (expand_expr_real_2): Update calls to
6832 find_widening_optab_handler.
6833 * optabs.c (expand_widen_pattern_expr): Likewise.
6834 (expand_binop_directly): Take the insn_code as a parameter.
6835 (expand_binop): Only call find_widening_optab_handler for
6836 conversion optabs; use optab_handler otherwise. Update calls
6837 to find_widening_optab_handler and expand_binop_directly.
6838 Use convert_optab_handler instead of widening_optab_handler.
6839 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
6840 find_widening_optab_handler and use scalar_mode rather than
6842 (convert_plusminus_to_widen): Likewise.
6844 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6845 Alan Hayward <alan.hayward@arm.com>
6846 David Sherwood <david.sherwood@arm.com>
6848 * machmode.h (fixed_size_mode): New class.
6849 * rtl.h (get_pool_mode): Return fixed_size_mode.
6850 * gengtype.c (main): Add fixed_size_mode.
6851 * target.def (get_raw_result_mode): Return a fixed_size_mode.
6852 (get_raw_arg_mode): Likewise.
6853 * doc/tm.texi: Regenerate.
6854 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
6855 * targhooks.c (default_get_reg_raw_mode): Likewise.
6856 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
6857 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
6858 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
6859 (msp430_get_raw_result_mode): Likewise.
6860 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
6861 * dbxout.c (dbxout_parms): Require fixed-size modes.
6862 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
6863 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
6864 * omp-low.c (lower_oacc_reductions): Likewise.
6865 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
6866 (simplify_subreg): Update accordingly.
6867 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
6868 (force_const_mem): Update accordingly. Return NULL_RTX for modes
6869 that aren't fixed-size.
6870 (get_pool_mode): Return a fixed_size_mode.
6871 (output_constant_pool_2): Take a fixed_size_mode.
6873 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6874 Alan Hayward <alan.hayward@arm.com>
6875 David Sherwood <david.sherwood@arm.com>
6877 * doc/rtl.texi (vec_series): Document.
6878 (const): Say that the operand can be a vec_series.
6879 * rtl.def (VEC_SERIES): New rtx code.
6880 * rtl.h (const_vec_series_p_1): Declare.
6881 (const_vec_series_p): New function.
6882 * emit-rtl.h (gen_const_vec_series): Declare.
6883 (gen_vec_series): Likewise.
6884 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
6885 (gen_vec_series): Likewise.
6886 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
6887 * simplify-rtx.c (simplify_unary_operation): Handle negations
6889 (simplify_binary_operation_series): New function.
6890 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
6891 (test_vector_ops_series): New function.
6892 (test_vector_ops): Call it.
6893 * config/powerpcspe/altivec.md (altivec_lvsl): Use
6894 gen_const_vec_series.
6895 (altivec_lvsr): Likewise.
6896 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
6898 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6899 Alan Hayward <alan.hayward@arm.com>
6900 David Sherwood <david.sherwood@arm.com>
6902 * doc/rtl.texi (const): Update description of address constants.
6903 Say that vector constants are allowed too.
6904 * common.md (E, F): Use CONSTANT_P instead of checking for
6906 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
6907 checking for CONST_VECTOR.
6908 * expmed.c (make_tree): Use build_vector_from_val for a CONST
6910 * expr.c (expand_expr_real_2): Check for vector modes instead
6911 of checking for CONST_VECTOR.
6912 * rtl.h (const_vec_p): New function.
6913 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
6914 (unwrap_const_vec_duplicate): Handle them here too.
6916 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6917 David Malcolm <dmalcolm@redhat.com>
6918 Alan Hayward <alan.hayward@arm.com>
6919 David Sherwood <david.sherwood@arm.com>
6921 * rtl.h (vec_duplicate_p): New function.
6922 * selftest-rtl.c (assert_rtx_eq_at): New function.
6923 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
6924 (assert_rtx_eq_at): Declare.
6925 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
6926 * selftest-run-tests.c (selftest::run_tests): Call it.
6927 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
6928 (simplify_unary_operation_1): Recursively handle vector duplicates.
6929 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
6931 (simplify_subreg): Handle subregs of vector duplicates.
6932 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
6933 (selftest::simplify_rtx_c_tests): New functions.
6935 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6936 Alan Hayward <alan.hayward@arm.com>
6937 David Sherwood <david.sherwood@arm.com>
6939 * emit-rtl.h (gen_const_vec_duplicate): Declare.
6940 (gen_vec_duplicate): Likewise.
6941 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
6943 (gen_const_vector): ...here.
6944 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
6945 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
6946 whose elements are all equal.
6947 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
6948 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
6949 (simplify_relational_operation): Likewise.
6950 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
6952 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
6953 (aarch64_expand_vector_init): Likewise.
6954 * config/arm/arm.c (neon_vdup_constant): Likewise.
6955 (neon_expand_vector_init): Likewise.
6956 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
6957 (arm_block_set_unaligned_vect): Likewise.
6958 (arm_block_set_aligned_vect): Likewise.
6959 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
6960 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
6961 (expand_vec_perm_even_odd_pack): Likewise.
6962 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
6963 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
6964 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
6965 gen_const_vec_duplicate.
6966 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
6967 * config/mips/mips.c (mips_gen_const_int_vector): Use
6968 gen_const_vec_duplicate.
6969 (mips_expand_vector_init): Use CONST0_RTX.
6970 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
6971 (define_split): Use gen_const_vec_duplicate.
6972 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
6973 (define_split): Use gen_const_vec_duplicate.
6974 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
6975 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
6976 * config/spu/spu.c (spu_const): Likewise.
6978 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6979 Alan Hayward <alan.hayward@arm.com>
6980 David Sherwood <david.sherwood@arm.com>
6982 * combine.c (can_change_dest_mode): Reject changes in
6983 REGMODE_NATURAL_SIZE.
6985 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
6987 * configure.ac (--enable-libssp): New.
6988 (gcc_cv_libc_provides_ssp): Check for explicit setting before
6989 trying to determine target-specific default. Adjust indentation.
6990 * configure: Regenerated.
6991 * doc/install.texi (Configuration): Expand --disable-libssp
6994 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
6996 config/i386/i386.c (ix86_expand_epilogue): Correct stack
6999 2017-10-31 Martin Jambor <mjambor@suse.cz>
7002 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
7004 2017-10-31 David Malcolm <dmalcolm@redhat.com>
7006 * auto-profile.c (autofdo_source_profile::read): Use
7007 UNKNOWN_LOCATION rather than 0.
7008 * diagnostic-core.h (warning_at_rich_loc): Rename to...
7009 (warning_at): ...this overload.
7010 (warning_at_rich_loc_n): Rename to...
7011 (warning_n): ...this overload.
7012 (error_at_rich_loc): Rename to...
7013 (error_at): ...this overload.
7014 (pedwarn_at_rich_loc): Rename to...
7015 (pedwarn): ...this overload.
7016 (permerror_at_rich_loc): Rename to...
7017 (permerror): ...this overload.
7018 (inform_at_rich_loc): Rename to...
7019 (inform): ...this overload.
7020 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
7021 (diagnostic_n_impl_richloc): Rename to...
7022 (diagnostic_n_impl): ...this rich_location *-based decl.
7023 (inform_at_rich_loc): Rename to...
7024 (inform): ...this, and add an assertion.
7025 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
7026 (warning_at_rich_loc): Rename to...
7027 (warning_at): ...this, and add an assertion.
7028 (warning_at_rich_loc_n): Rename to...
7029 (warning_n): ...this, and add an assertion.
7030 (warning_n): Update location_t-based implementation for removal of
7031 location_t-based diagnostic_n_impl.
7032 (pedwarn_at_rich_loc): Rename to...
7033 (pedwarn): ...this, and add an assertion.
7034 (permerror_at_rich_loc): Rename to...
7035 (permerror): ...this, and add an assertion.
7036 (error_n): Update for removal of location_t-based diagnostic_n_impl.
7037 (error_at_rich_loc): Rename to...
7038 (error_at): ...this, and add an assertion.
7039 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
7040 (driver::do_spec_on_infiles): Likewise.
7041 * substring-locations.c (format_warning_va): Update for renaming
7042 of inform_at_rich_loc.
7044 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
7046 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
7047 _Float<N>X built-in functions so that the variant without the
7048 "__builtin_" prefix is only enabled for the GNU C and Objective C
7049 languages when they are in non-strict ANSI/ISO mode.
7050 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
7051 * target.def (floatn_builtin_p): Add a target hook to control
7052 whether _Float<N> and _Float<N>X built-in functions without the
7053 "__builtin_" prefix are enabled, and return true for C and
7054 Objective C in the default hook. Include langhooks.h in
7056 * targhooks.h (default_floatn_builtin_p): Likewise.
7057 * targhooks.c (default_floatn_builtin_p): Likewise.
7058 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
7059 floatn_builtin_p target hook.
7060 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
7062 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
7063 Eric Botcazou <ebotcazou@adacore.com>
7065 PR rtl-optimization/81803
7066 * lra-constraints.c (curr_insn_transform): Also reload the whole
7067 register for a strict subreg no wider than a word if this is for
7068 a WORD_REGISTER_OPERATIONS target.
7070 2017-10-31 Jason Merrill <jason@redhat.com>
7072 * gdbinit.in: Skip over inlines from timevar.h.
7074 2017-10-31 Martin Liska <mliska@suse.cz>
7076 * doc/gcov.texi: Document new option.
7077 * gcov.c (print_usage): Likewise print it.
7078 (process_args): Support the argument.
7079 (format_count): New function.
7080 (format_gcov): Use the function.
7082 2017-10-31 Martin Liska <mliska@suse.cz>
7084 * gcov.c (struct name_map): do not use typedef.
7085 Define operator== and operator<.
7086 (name_search): Remove.
7087 (name_sort): Remove.
7088 (main): Do not allocate names.
7089 (process_file): Add vertical space.
7090 (generate_results): Use std::find.
7091 (release_structures): Do not release memory.
7092 (find_source): Use std::find.
7094 2017-10-31 Martin Liska <mliska@suse.cz>
7096 * gcov.c (struct line_info): Remove it's typedef.
7097 (line_info::line_info): Add proper ctor.
7098 (line_info::has_block): Do not use a typedef.
7099 (struct source_info): Do not use typedef.
7100 (circuit): Likewise.
7101 (get_cycles_count): Likewise.
7102 (output_intermediate_file): Iterate via vector iterator.
7103 (add_line_counts): Use std::vector methods.
7104 (accumulate_line_counts): Likewise.
7105 (output_lines): Likewise.
7107 2017-10-31 Martin Liska <mliska@suse.cz>
7109 * gcov.c (struct source_info): Remove typedef.
7110 (source_info::source_info): Add proper ctor.
7111 (accumulate_line_counts): Use struct, not it's typedef.
7112 (output_gcov_file): Likewise.
7113 (output_lines): Likewise.
7114 (main): Do not allocate an array.
7115 (output_intermediate_file): Use size of vector container.
7116 (process_file): Resize the vector.
7117 (generate_results): Do not preallocate, use newly added vector
7119 (release_structures): Do not release sources.
7120 (find_source): Use vector methods.
7121 (add_line_counts): Do not use typedef.
7123 2017-10-31 Martin Liska <mliska@suse.cz>
7125 * doc/gcov.texi: Document that.
7126 * gcov.c (add_line_counts): Mark lines with a non-executed
7128 (output_line_beginning): Handle such lines.
7129 (output_lines): Pass new argument.
7130 (output_intermediate_file): Print it in intermediate format.
7132 2017-10-31 Martin Liska <mliska@suse.cz>
7134 * color-macros.h: New file.
7135 * diagnostic-color.c: Factor out color related to macros to
7137 * doc/gcov.texi: Document -k option.
7138 * gcov.c (INCLUDE_STRING): Include string.h.
7139 (print_usage): Add -k option.
7140 (process_args): Parse it.
7141 (pad_count_string): New function.
7142 (output_line_beginning): Likewise.
7143 (DEFAULT_LINE_START): New macro.
7144 (output_lines): Support color output.
7146 2017-10-31 Martin Liska <mliska@suse.cz>
7148 PR gcov-profile/82633
7149 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
7150 their interaction with GCOV infrastructure.
7151 * configure.ac: Add -fkeep-{inline,static}-functions to
7153 * configure: Regenerate.
7155 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
7158 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
7160 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
7163 * config/rs6000/rs6000.md (allocate_stack): Force update interval
7164 into a register if it does not fit into an immediate offset field.
7166 2017-10-31 Olivier Hainque <hainque@adacore.com>
7168 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
7170 2017-10-31 Julia Koval <julia.koval@intel.com>
7172 * config.gcc: Add gfniintrin.h.
7173 * config/i386/gfniintrin.h: New.
7174 * config/i386/i386-builtin-types.def
7175 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7176 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7177 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7178 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7179 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7180 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
7181 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
7182 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
7183 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
7184 V64QI_FTYPE_V64QI_V64QI_INT): New types.
7185 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
7186 * config/i386/immintrin.h: Include gfniintrin.h.
7187 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
7189 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
7191 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
7193 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
7195 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
7196 (arm_ashldi3_1bit): Remove pattern.
7197 (ashrdi3): Remove shift by 1 expansion.
7198 (arm_ashrdi3_1bit): Remove pattern.
7199 (lshrdi3): Remove shift by 1 expansion.
7200 (arm_lshrdi3_1bit): Remove pattern.
7201 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
7202 cost of ashldi3 by 1.
7203 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
7204 (<shift>di3_neon): Likewise.
7206 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7208 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
7209 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
7210 and (*aarch64_simd_mov<VQ:mode>).
7211 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
7212 pattern alternative.
7213 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
7214 attributes to match pattern alternative.
7216 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
7218 * config.gcc (powerpc*-*-*): Add emmintrin.h.
7219 * config/rs6000/emmintrin.h: New file.
7220 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
7222 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
7224 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
7225 * (movdi_vfp_cortexa8): Remove pattern.
7227 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7229 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
7231 (Specific, alpha*-dec-osf5.1): Remove.
7232 (Specific, mips-sgi-irix5): Remove.
7233 (Specific, mips-sgi-irix6): Remove.
7235 2017-10-30 Jakub Jelinek <jakub@redhat.com>
7238 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
7239 arguments to clear_bit_region_be.
7241 2017-10-30 Jim Wilson <wilson@tuliptree.org>
7243 * gimplify.c: Include memmodel.h.
7245 2017-10-30 Martin Jambor <mjambor@suse.cz>
7247 * omp-grid.c (grid_attempt_target_gridification): Also insert a
7248 condition whether loop should be executed at all.
7250 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
7252 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7253 gimple folding of vec_madd() intrinsics.
7254 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
7255 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
7256 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
7258 2017-10-30 Richard Biener <rguenther@suse.de>
7260 PR tree-optimization/82762
7262 2017-10-23 Richard Biener <rguenther@suse.de>
7264 PR tree-optimization/82129
7266 2017-08-01 Richard Biener <rguenther@suse.de>
7268 PR tree-optimization/81181
7269 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7270 (compute_antic): ... end of iteration here.
7272 2017-10-30 Joseph Myers <joseph@codesourcery.com>
7274 * doc/invoke.texi (C Dialect Options): Document -std=c17,
7275 -std=iso9899:2017 and -std=gnu17.
7276 * doc/standards.texi (C Language): Document C17 support.
7277 * doc/cpp.texi (Overview): Mention -std=c17.
7278 (Standard Predefined Macros): Document C11 and C17 values of
7279 __STDC_VERSION__. Do not refer to C99 support as incomplete.
7280 * doc/extend.texi (Inline): Do not list individual options for
7281 standards newer than C99.
7282 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
7284 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
7287 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
7289 * asan.c (asan_finish_file): Align asan globals array by shadow
7292 2017-10-30 Jakub Jelinek <jakub@redhat.com>
7295 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
7296 (struct store_immediate_info): Add bitregion_start and bitregion_end
7298 (store_immediate_info::store_immediate_info): Add brs and bre
7299 arguments and initialize bitregion_{start,end} from those.
7300 (struct merged_store_group): Add bitregion_start, bitregion_end,
7301 align_base and mask fields. Drop unnecessary struct keyword from
7302 struct store_immediate_info. Add do_merge method.
7303 (clear_bit_region_be): Use memset instead of loop storing zeros.
7304 (merged_store_group::do_merge): New method.
7305 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
7306 stores as long as the surrounding bitregions have no gaps.
7307 (merged_store_group::merge_overlapping): Use do_merge.
7308 (merged_store_group::apply_stores): Test that bitregion_{start,end}
7309 is byte aligned, rather than requiring that start and width are
7310 byte aligned. Drop unnecessary struct keyword from
7311 struct store_immediate_info. Allocate and populate also mask array.
7312 Make start of the arrays relative to bitregion_start rather than
7313 start and size them according to bitregion_{end,start} difference.
7314 (struct imm_store_chain_info): Drop unnecessary struct keyword from
7315 struct store_immediate_info.
7316 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
7317 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
7318 struct keyword from struct store_immediate_info.
7319 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
7320 between stores as long as the surrounding bitregions have no gaps.
7322 (struct split_store): Add orig non-static data member.
7323 (split_store::split_store): Initialize orig to false.
7324 (find_constituent_stmts): Return store_immediate_info *, non-NULL
7325 if there is exactly a single original stmt. Change stmts argument
7326 to pointer from reference, if NULL, don't push anything to it. Add
7327 first argument, use it to optimize skipping over orig stmts that
7328 are known to be before bitpos already. Simplify.
7329 (split_group): Return unsigned int count how many stores are or
7330 would be needed rather than a bool. Add allow_unaligned argument.
7331 Change split_stores argument from reference to pointer, if NULL,
7332 only do a dry run computing how many stores would be produced.
7333 Rewritten algorithm to use both alignment and misalign if
7334 !allow_unaligned and handle bitfield stores with gaps.
7335 (imm_store_chain_info::output_merged_store): Set start_byte_pos
7336 from bitregion_start instead of start. Compute allow_unaligned
7337 here, if true, do 2 split_group dry runs to compute which one
7338 produces fewer stores and prefer aligned if equal. Punt if
7339 new count is bigger or equal than original before emitting any
7340 statements, rather than during that. Remove no longer needed
7341 new_ssa_names tracking. Replace num_stmts with
7342 split_stores.length (). Use 32-bit stack allocated entries
7343 in split_stores auto_vec. Try to reuse original store lhs/rhs1
7344 if possible. Handle bitfields with gaps.
7345 (pass_store_merging::execute): Ignore bitsize == 0 stores.
7346 Compute bitregion_{start,end} for the stores and construct
7347 store_immediate_info with that. Formatting fixes.
7349 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
7352 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
7353 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
7355 2017-10-29 Jim Wilson <wilson@tuliptree.org>
7357 * gimplify.c: Include tm_p.h.
7359 * common.opt (gcoff): Re-add as ignored option.
7360 (gcoff1, gcoff2, gcoff3): Likewise.
7362 * Makefile.in (OBJS): Delete sdbout.o.
7363 (GTFILES): Delete $(srcdir)/sdbout.c.
7364 * debug.h: Delete sdb_debug_hooks.
7365 * final.c: Delete sdbout.h include.
7366 (final_scan_insn): Delete SDB_DEBUG check.
7367 (rest_of_clean_state): Likewise.
7368 * output.h: Delete sdb_begin_function_line.
7371 * toplev.c: Delete sdbout.h include.
7372 (process_options): Delete SDB_DEBUG check.
7373 * tree-core.h (tree_type_common): Delete pointer field of
7375 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
7376 TYPE_SYMTAB_POINTER.
7377 * tree.h (TYPE_SYMTAB_POINTER): Delete.
7378 (TYPE_SYMTAB_IS_POINTER): Delete.
7379 (TYPE_SYMTAB_IS_DIE): Renumber.
7380 * xcoffout.c: Refer to former sdbout.c file.
7381 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
7383 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
7384 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
7385 (Debugging Options): Delete -gcoff.
7386 (-gstabs): Delete SDB reference.
7388 (-gcoff@var{level}): Delete.
7389 * doc/passes.texi (Debugging information output): Delete SDB and
7390 sdbout.c references.
7391 * doc/tm.texi: Regenerate.
7392 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
7393 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
7395 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
7396 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
7397 (SDB_DEBUGGING_INFO): Delete.
7398 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
7399 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
7400 * target.def (output_source_filename): Delete COFF reference.
7402 * common.opt (gcoff): Delete.
7403 (gxcoff+): Update Negative chain.
7404 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
7406 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
7407 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
7408 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
7410 (expand_function_start): Change sdb reference to past tense.
7411 (expand_function_end): Change sdb reference to past tense.
7412 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
7413 * opts.c (debug_type_names): Delete coff entry.
7414 (common_handle_option): Delete OPT_gcoff case.
7415 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
7417 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
7418 * config/cris/cris.h: Delete SDB reference in comment.
7419 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
7420 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
7421 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
7422 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
7424 (ix86_expand_prologue): Likewise.
7425 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
7426 * config/ia64/ia64.h: Likewise.
7427 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
7428 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
7430 * config/mmix/mmix.h: Likewise.
7431 * config/nds32/nds32.c: Likewise.
7432 * config/stormy/storym16.h: Likewise.
7433 * config/visium/visium.h: Likewise.
7434 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
7436 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
7438 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
7439 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
7440 FRAME_POINTER_REGNUM point at high end of local var area.
7442 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
7444 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
7445 Move comment around. Do not reset best_edge for a copiable
7446 destination if the copy would cause a partition change.
7447 (better_edge_p): Remove redundant check.
7449 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
7451 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
7453 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
7455 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
7456 for math functions that have _Float<N> and _Float<N>X variants.
7457 (mathfn_built_in_2): Add support for math functions that have
7458 _Float<N> and _Float<N>X variants.
7459 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
7460 (expand_builtin_mathfn_ternary): Add support for fma with
7461 _Float<N> and _Float<N>X variants.
7462 (expand_builtin): Likewise.
7463 (fold_builtin_3): Likewise.
7464 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
7465 create math function _Float<N> and _Float<N>X variants as external
7467 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
7468 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
7469 the __builtin_ prefix and if not strict ansi, without the prefix.
7470 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
7471 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
7472 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
7473 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
7474 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
7475 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
7476 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
7477 function signatures for fma _Float<N> and _Float<N>X variants.
7478 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
7479 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
7480 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
7481 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
7482 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
7483 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
7484 * gencfn-macros.c (print_case_cfn): Add support for math functions
7485 that have _Float<N> and _Float<N>X variants.
7486 (print_define_operator_list): Likewise.
7487 (fltfn_suffixes): Likewise.
7489 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
7490 for math functions that have _Float<N> and _Float<N>X variants.
7491 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
7492 and _Float<N>X variants.
7493 (COPYSIGN): Likewise.
7496 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
7497 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
7499 (integer_valued_read_call_p): Likewise.
7500 * fold-const-call.c (fold_const_call_ss): Likewise.
7501 (fold_const_call_sss): Add support for copysign, fmin, and fmax
7502 _Float<N> and _Float<N>X variants.
7503 (fold_const_call_ssss): Add support for fma _Float<N> and
7504 _Float<N>X variants.
7505 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
7506 support for copysign and fma _Float<N> and _Float<N>X variants.
7507 (backprop::process_builtin_call_use): Likewise.
7508 * tree-call-cdce.c (can_test_argument_range); Add support for
7509 sqrt _Float<N> and _Float<N>X variants.
7510 (edom_only_function): Likewise.
7511 (get_no_error_domain): Likewise.
7512 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
7513 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
7514 copysign _Float<N> and _Float<N>X variants.
7515 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
7516 handled by machine independent code.
7517 (FMAF128): Likewise.
7518 * doc/cpp.texi (Common Predefined Macros): Document defining
7519 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
7520 fma _Float<N> and _Float<N>X variants.
7522 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
7525 * config/i386/i386-modes.def (CCFPU): Remove definition.
7526 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
7527 (ix86_cc_modes_compatible): Ditto.
7528 (ix86_expand_carry_flag_compare): Ditto.
7529 (ix86_expand_int_movcc): Ditto.
7530 (ix86_expand_int_addcc): Ditto.
7531 (ix86_reverse_condition): Ditto.
7532 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
7533 Return true/false for unordered/ordered fp comparisons.
7534 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
7535 (ix86_prepare_fp_compare_args): Update for rename.
7536 (ix86_expand_fp_compare): Update for rename. Generate unordered
7537 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
7538 (ix86_expand_sse_compare_and_jump): Ditto.
7539 * config/i386/predicates.md (fcmov_comparison_operator):
7540 Remove CCFPU mode handling.
7541 (ix86_comparison_operator): Ditto.
7542 (ix86_carry_flag_operator): Ditto.
7543 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
7544 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
7545 (*cmpu<mode>_cc_i387): Ditto.
7546 (FPCMP): Remove mode iterator.
7547 (unord): Remove mode attribute.
7548 (unord_subst): New define_subst transformation
7549 (unord): New define_subst attribute.
7551 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
7552 (*cmpi<unord>xf_i387): Ditto.
7553 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
7554 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
7555 using unord_subst transformation.
7556 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
7557 (round_saeonly): Also handle CCFP mode.
7558 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
7559 Remove UNSPEC_SAHF unspec handling.
7561 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
7563 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
7565 2017-10-27 Jeff Law <law@redhat.com>
7567 * gimple-ssa-sprintf.c: Include domwalk.h.
7568 (class sprintf_dom_walker): New class, derived from dom_walker.
7569 (sprintf_dom_walker::before_dom_children): New function.
7570 (struct call_info): Moved into sprintf_dom_walker class
7571 (compute_formath_length, handle_gimple_call): Likewise.
7572 (sprintf_length::execute): Call the dominator walker rather
7573 than walking the statements.
7575 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
7576 gimple statement locations.
7577 (check_array_bounds): Corresponding changes. Get the statement's
7578 location directly from wi->stmt.
7580 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
7583 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
7585 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
7587 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
7588 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
7590 2017-10-27 Jakub Jelinek <jakub@redhat.com>
7593 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
7594 * config/i386/i386.c (maybe_get_pool_constant): Removed.
7595 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
7596 maybe_get_pool_constant.
7597 * config/i386/predicates.md (zero_extended_scalar_load_operand):
7600 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7602 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
7603 2.26 caveat. Update gas and gld versions.
7604 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
7607 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7609 * cgraph.h (set_malloc_flag): Declare.
7610 * cgraph.c (set_malloc_flag_1): New function.
7611 (set_malloc_flag): Likewise.
7612 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
7613 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
7615 (read_ipa_call_summary): Add support for reading is_return_callee.
7616 (write_ipa_call_summary): Stream is_return_callee.
7617 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
7618 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
7619 ipa-prop.h, ipa-fnsummary.h.
7620 (pure_const_names): Change to static.
7621 (malloc_state_e): Define.
7622 (malloc_state_names): Define.
7623 (funct_state_d): Add field malloc_state.
7624 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
7625 (check_retval_uses): New function.
7626 (malloc_candidate_p): Likewise.
7627 (analyze_function): Add support for malloc attribute.
7628 (pure_const_write_summary): Stream malloc_state.
7629 (pure_const_read_summary): Add support for reading malloc_state.
7630 (dump_malloc_lattice): New function.
7631 (propagate_malloc): New function.
7632 (warn_function_malloc): New function.
7633 (ipa_pure_const::execute): Call propagate_malloc and
7634 ipa_free_fn_summary.
7635 (pass_local_pure_const::execute): Add support for malloc attribute.
7636 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
7637 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
7639 2017-10-27 Martin Liska <mliska@suse.cz>
7641 PR gcov-profile/82457
7642 * doc/invoke.texi: Document that one needs a non-strict ISO mode
7643 for fork-like functions to be properly instrumented.
7645 2017-10-27 Richard Biener <rguenther@suse.de>
7648 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
7649 info when we redirected EH.
7651 2017-10-26 Michael Collison <michael.collison@arm.com>
7653 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
7655 (<optab>_trunchf<GPI:mode>2: New pattern.
7656 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
7657 * config/aarch64/iterators.md (wv): New mode attribute.
7658 (vf, VF): New mode attributes.
7659 (vgp, VGP): New mode attributes.
7660 (s): Update attribute with SImode and DImode prefixes.
7662 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7664 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
7665 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
7666 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
7667 (nios2_option_overide): Initialize it. Don't allow R0-relative
7668 addressing with PIC.
7669 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
7670 (nios2_symbolic_constant_p): Likewise.
7671 (nios2_legitimate_address_p): Likewise.
7672 (nios2_r0rel_section_name_p): New.
7673 (nios2_symbol_ref_in_r0rel_data_p): New.
7674 (nios2_emit_move_sequence): Handle r0rel_constant_p.
7675 (r0rel_constant_p): New.
7676 (nios2_print_operand_address): Handle r0rel_constant_p.
7677 (nios2_cdx_narrow_form_p): Likewise.
7678 * config/nios2/nios2.opt (mr0rel-sec=): New option.
7679 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
7680 (Nios II Options): Document -mr0rel-sec.
7682 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7684 * config/nios2/nios2.c: Include xregex.h.
7685 (nios2_gprel_sec_regex): New.
7686 (nios2_option_overide): Initialize it. Don't allow GP-relative
7687 addressing with PIC.
7688 (nios2_small_section_name_p): Check for regex match.
7689 * config/nios2/nios2.opt (mgprel-sec=): New option.
7690 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
7691 (Nios II Options): Document -mgprel-sec.
7693 2017-10-26 Jim Wilson <wilson@tuliptree.org>
7695 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
7697 2017-10-26 Tom de Vries <tom@codesourcery.com>
7699 PR tree-optimization/82707
7700 * gimple.c (gimple_copy): Fix unsharing of
7701 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
7703 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
7705 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
7706 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
7707 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
7708 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
7709 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
7710 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
7711 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
7712 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
7713 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
7714 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
7715 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
7716 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
7717 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
7718 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
7719 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
7720 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
7721 _mm512_mask_cmpunord_ps_mask): New intrinsics.
7723 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7725 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
7727 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
7728 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
7729 warning to rs6000.c. Remove the Undocumented flag, since it has
7731 (-mabi=ibmlongdouble): Likewise.
7732 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
7733 already set, set the default format for long double.
7734 (rs6000_debug_reg_global): Print whether long double is IBM or
7736 (rs6000_option_override_internal): Rework setting long double
7737 format. Only warn if the user is changing the long double default
7738 and they did not use -Wno-psabi.
7739 * doc/invoke.texi (PowerPC options): Update the documentation for
7740 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
7742 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7743 Alan Hayward <alan.hayward@arm.com>
7744 David Sherwood <david.sherwood@arm.com>
7746 * rtl.h (wider_subreg_mode): New function.
7747 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
7748 rather than an unsigned int *.
7749 * ira-color.c (regno_max_ref_width): Replace with...
7750 (regno_max_ref_mode): ...this new variable.
7751 (coalesced_pseudo_reg_slot_compare): Update accordingly.
7752 Use wider_subreg_mode.
7753 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
7754 rather than an unsigned int *.
7755 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
7756 (process_alt_operands): Likewise.
7757 (invariant_p): Likewise.
7758 * lra-spills.c (assign_mem_slot): Likewise.
7759 (add_pseudo_to_slot): Likewise.
7760 * lra.c (collect_non_operand_hard_regs): Likewise.
7761 (add_regs_to_insn_regno_info): Likewise.
7762 * reload1.c (regno_max_ref_width): Replace with...
7763 (regno_max_ref_mode): ...this new variable.
7764 (reload): Update accordingly. Update call to
7765 ira_sort_regnos_for_alter_reg.
7766 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
7767 (init_eliminable_invariants): Update to use regno_max_ref_mode.
7768 (scan_paradoxical_subregs): Likewise.
7770 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7772 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
7773 (aarch64_frame): Add emit_frame_chain boolean.
7774 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7775 Move eh_return case to aarch64_layout_frame.
7776 (aarch64_layout_frame): Initialize emit_frame_chain.
7777 (aarch64_expand_prologue): Use emit_frame_chain.
7779 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7781 * config/aarch64/aarch64.c (aarch64_layout_frame):
7782 Ensure LR is always stored at the bottom of the callee-saves.
7783 Remove rarely used frame layout which saves callee-saves at top of
7784 frame, so the store of LR can be used as a valid probe in all cases.
7786 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7788 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
7789 Improve unaligned TImode/TFmode base/offset split.
7791 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7792 Alan Hayward <alan.hayward@arm.com>
7793 David Sherwood <david.sherwood@arm.com>
7795 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
7796 * combine.c (find_single_use_1): Likewise.
7797 (expand_field_assignment): Likewise.
7798 (move_deaths): Likewise.
7799 * lra-constraints.c (simplify_operand_subreg): Likewise.
7800 (curr_insn_transform): Likewise.
7801 * lra.c (collect_non_operand_hard_regs): Likewise.
7802 (add_regs_to_insn_regno_info): Likewise.
7803 * rtlanal.c (reg_referenced_p): Likewise.
7804 (covers_regno_no_parallel_p): Likewise.
7806 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7808 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
7809 Don't print any bits outside the precision of the value.
7810 * wide-int.cc (test_printing): Add some new tests.
7812 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7814 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
7815 supports -xbrace_comment option.
7816 * configure: Regenerate.
7817 * config.in: Regenerate.
7818 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
7819 (ASM_CPU_SPEC): Use it.
7821 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7823 * target.def (static_rtx_alignment): New hook.
7824 * targhooks.h (default_static_rtx_alignment): Declare.
7825 * targhooks.c (default_static_rtx_alignment): New function.
7826 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
7827 * doc/tm.texi: Regenerate.
7828 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
7829 instead of targetm.constant_alignment. Remove call to
7831 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7832 (cris_preferred_mininum_alignment): New function, split out from...
7833 (cris_constant_alignment): ...here.
7834 (cris_static_rtx_alignment): New function.
7835 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
7837 (ix86_constant_alignment): ...here.
7838 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7839 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7840 (mmix_static_rtx_alignment): New function.
7841 * config/spu/spu.c (spu_static_rtx_alignment): New function.
7842 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7844 2017-10-26 Tamar Christina <tamar.christina@arm.com>
7847 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
7848 Add flag_trapping_math and flag_fp_int_builtin_inexact.
7850 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
7852 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
7853 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
7854 mark as a sign-extending load.
7855 (local_pic_load_u): Define.
7857 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
7860 * fold-const.c (operand_equal_for_comparison_p): Also return true
7861 if ARG0 is a simple variant of ARG1 with narrower precision.
7862 (fold_ternary_loc): Always pass unstripped operands to the predicate.
7864 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
7866 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
7868 * i386.h (processor_costs): Add gather_static, gather_per_elt,
7869 scatter_static, scatter_per_elt.
7870 * x86-tune-costs.h: Add new cost entries.
7872 2017-10-25 Richard Biener <rguenther@suse.de>
7874 * tree-ssa-sccvn.h (vn_eliminate): Declare.
7875 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
7876 class pass_fre): Move to ...
7877 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
7878 class pass_fre): ... here and adjust for statistics.
7880 2017-10-25 Jakub Jelinek <jakub@redhat.com>
7883 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
7884 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
7885 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
7886 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
7889 2017-10-25 Richard Biener <rguenther@suse.de>
7891 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
7892 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
7893 eliminate_push_avail, eliminate_insert): Move inside...
7894 (class eliminate_dom_walker): ... this class in preparation
7896 (fini_eliminate): Remove by merging with ...
7897 (eliminate): ... this function. Adjust for class changes.
7898 (pass_pre::execute): Remove fini_eliminate call.
7899 (pass_fre::execute): Likewise.
7901 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7904 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
7905 (VPERMI2, VPERMI2I): New mode iterators.
7906 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
7907 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
7909 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
7910 mode iterator. Remove 3 old define_insn patterns.
7911 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
7912 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
7913 VPERMI2 mode iterator, remove the other two expanders.
7914 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
7915 to use VPERMI2 mode iterator, add another alternative for vpermi2*
7916 instructions, remove the other two patterns.
7917 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
7918 mode iterator, remove the other two patterns.
7919 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
7920 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
7921 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
7922 and adjust argument order accordingly.
7923 (ix86_expand_vec_perm): Adjust caller.
7924 (expand_vec_perm_1): Likewise.
7925 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
7926 (expand_vec_perm_vpermt2_vpshub2): ... this.
7927 (ix86_expand_vec_perm_const_1): Adjust caller.
7928 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
7931 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
7932 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
7933 (vec_shl_<mode>): Remove unused expander.
7934 (avx512bw_<shift_insn><mode>3): New define_insn.
7935 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
7936 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
7938 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
7941 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
7944 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7946 PR rtl-optimization/82396
7947 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
7948 (autopref_multipass_init): Simplify initialization.
7949 (autopref_rank_data): Simplify sort order.
7950 * gcc/sched-int.h (autopref_multipass_data_): Remove
7951 multi_mem_insn_p, min_offset and max_offset.
7953 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7956 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7957 Check special value of flag_omit_frame_pointer.
7958 (aarch64_can_eliminate): Likewise.
7959 (aarch64_override_options_after_change_1): Simplify handling of
7960 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
7962 2017-10-24 Richard Biener <rguenther@suse.de>
7964 PR tree-optimization/82697
7965 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
7966 zero for conditional load and unconditional store.
7968 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7970 * doc/install.texi: Document bootstrap-cet.
7972 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7975 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
7976 ENDBR instruction at function entrance if function is only
7979 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7982 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
7983 patterns to better describe from which operation the CF is computed.
7984 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
7985 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
7986 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
7987 is 0, use _0 suffixed expanders instead of emitting a comparison
7990 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
7992 * config/i386/i386.md(*movsf_internal, *movdf_internal):
7993 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
7995 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
7998 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
7999 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
8000 * loop-iv.c (iv_get_reaching_def): Likewise.
8001 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
8002 variable is promoted and the partition contains undefined values.
8004 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8006 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
8008 (nios2_address_cost): Define.
8009 (nios2_legitimize_address): Recognize (exp + constant) directly.
8010 (TARGET_ADDRESS_COST): Define.
8012 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8014 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
8015 (nios2_symbolic_memory_operand_p): Declare.
8016 (nios2_split_large_constant): Declare.
8017 (nios2_split_symbolic_memory_operand): Declare.
8018 * config/nios2/nios2.c: Adjust includes.
8019 (nios2_symbolic_constant_allowed): New.
8020 (nios2_symbolic_constant_p): New.
8021 (nios2_plus_symbolic_constant_p): New.
8022 (nios2_valid_addr_expr_p): Recognize addresses involving
8024 (nios2_legitimate_address_p): Likewise, also LO_SUM.
8025 (nios2_symbolic_memory_operand_p): New.
8026 (nios2_large_constant_p): New.
8027 (nios2_split_large_constant): New.
8028 (nios2_split_plus_large_constant): New.
8029 (nios2_split_symbolic_memory_operand): New.
8030 (nios2_legitimize_address): Code refactoring. Handle addresses
8031 involving symbolic constants.
8032 (nios2_emit_move_sequence): Likewise.
8033 (nios2_print_operand): Improve error output.
8034 (nios2_print_operand_address): Handle LO_SUM.
8035 (nios2_cdx_narrow_form_p): Likewise.
8036 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
8037 operands involving symbolic constants.
8038 (movhi_internal, movsi_internal): Likewise.
8039 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
8040 (extendhisi2, extendqi<mode>2): Likewise.
8042 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8044 * tree-pass.h (PROP_rtl_split_insns): Define.
8045 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
8047 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8049 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
8051 2017-10-23 Jakub Jelinek <jakub@redhat.com>
8054 * target.def (const_not_ok_for_debug_p): Default to
8055 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
8056 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
8057 * targhooks.c (default_const_not_ok_for_debug_p): New function.
8058 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
8059 which targetm.const_not_ok_for_debug_p returned true.
8060 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
8062 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
8064 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
8065 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
8066 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
8068 (ix86_const_not_ok_for_debug_p): New function.
8069 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
8070 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
8072 2017-10-23 David Malcolm <dmalcolm@redhat.com>
8075 * system.h: Conditionally include "unique-ptr.h" if
8076 INCLUDE_UNIQUE_PTR is defined.
8077 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
8078 of defining INCLUDE_UNIQUE_PTR before including "system.h".
8080 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
8082 * config/rl78/rl78.md: New define_expand "subdi3".
8084 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
8087 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
8088 DF_REF_INSN if DF_REF_INSN_INFO is false.
8090 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
8092 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
8093 xmm_move instead of sse_move.
8094 (sse_store_index): New function.
8095 (ix86_register_move_cost): Be more sensible about mismatch stall;
8096 model AVX moves correctly; make difference between sse->integer and
8098 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
8099 moves; make difference between SSE and AVX.
8100 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
8101 and zmm_move. Increase size of sse load and store tables;
8102 add unaligned load and store tables; add ssemmx_to_integer.
8103 * x86-tune-costs.h: Update all entries according to real
8104 move latencies from Agner Fog's manual and chip documentation.
8106 2017-10-23 Jakub Jelinek <jakub@redhat.com>
8109 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
8110 * config/i386/constraints.md (Wf): New constraint.
8111 * config/i386/i386.md (UNSPEC_SBB): New unspec.
8112 (cmp<dwi>_doubleword): Removed.
8113 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
8114 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
8115 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
8116 expand with cmp<dwi>_doubleword. For LTU and GEU use
8117 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
8119 * common.opt (gcolumn-info): Enable by default.
8120 * doc/invoke.texi (gcolumn-info): Document new default.
8122 2017-10-23 Richard Biener <rguenther@suse.de>
8124 PR tree-optimization/82672
8125 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
8126 Fold the stmt if we propagated into it.
8128 2017-10-23 Richard Biener <rguenther@suse.de>
8130 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
8131 (bitmap_remove_expr_from_set): ... this. All callers call this
8132 for non-constant values.
8133 (bitmap_set_subtract): Rename to...
8134 (bitmap_set_subtract_expressions): ... this. Adjust and
8136 (bitmap_set_contains_value): Remove superfluous check.
8137 (bitmap_set_replace_value): Inline into single caller ...
8138 (bitmap_value_replace_in_set): ... here and simplify.
8139 (dependent_clean): Merge into ...
8140 (clean): ... this using an overload. Adjust.
8141 (prune_clobbered_mems): Adjust.
8142 (compute_antic_aux): Likewise.
8143 (compute_partial_antic_aux): Likewise.
8145 2017-10-23 Richard Biener <rguenther@suse.de>
8147 PR tree-optimization/82129
8149 2017-08-01 Richard Biener <rguenther@suse.de>
8151 PR tree-optimization/81181
8152 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8153 (compute_antic): ... end of iteration here.
8155 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8157 * target.def (starting_frame_offset): New hook.
8158 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
8159 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
8160 * doc/tm.texi.in: Regenerate.
8161 * hooks.h (hook_hwi_void_0): Declare.
8162 * hooks.c (hook_hwi_void_0): New function.
8163 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
8164 STARTING_FRAME_OFFSET.
8165 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
8166 * reload1.c (reload): Likewise.
8167 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
8168 instead of STARTING_FRAME_OFFSET.
8169 * function.c (try_fit_stack_local): Likewise.
8170 (assign_stack_local_1): Likewise
8171 (instantiate_virtual_regs): Likewise.
8172 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
8173 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
8174 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
8175 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
8176 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
8177 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
8178 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
8179 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
8180 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
8181 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
8182 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
8183 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
8184 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
8185 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
8186 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
8187 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
8188 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
8189 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
8190 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
8191 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
8192 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
8193 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
8194 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
8195 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
8196 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
8197 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
8198 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
8199 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
8200 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
8201 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
8202 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
8203 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
8204 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
8205 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
8206 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
8207 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
8208 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
8209 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
8210 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
8211 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
8212 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
8213 * config/avr/avr.c (avr_starting_frame_offset): Make static and
8214 return a HOST_WIDE_INT.
8215 (avr_builtin_setjmp_frame_value): Use it instead of
8216 STARTING_FRAME_OFFSET.
8217 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8218 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
8219 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
8221 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8222 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
8223 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
8224 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8225 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
8226 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
8227 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8228 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
8229 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
8230 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8231 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
8232 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
8234 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8235 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
8236 * config/mips/mips.c (mips_compute_frame_info): Refer to
8237 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
8238 (mips_starting_frame_offset): New function.
8239 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8240 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
8241 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
8242 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
8243 and return a HOST_WIDE_INT.
8244 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8245 (mmix_initial_elimination_offset): Refer to
8246 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
8247 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
8248 * config/pa/pa.c (pa_starting_frame_offset): New function.
8249 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
8250 (pa_expand_prologue): Likewise.
8251 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8252 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
8253 !FRAME_GROWS_DOWNWARD handling to...
8254 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8255 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
8256 !FRAME_GROWS_DOWNWARD handling to...
8257 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8258 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
8259 !FRAME_GROWS_DOWNWARD handling to...
8260 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8261 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
8263 (rs6000_starting_frame_offset): New function.
8264 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
8265 !FRAME_GROWS_DOWNWARD handling to...
8266 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8267 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
8268 !FRAME_GROWS_DOWNWARD handling to...
8269 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8270 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
8271 !FRAME_GROWS_DOWNWARD handling to...
8272 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8273 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
8274 (rs6000_starting_frame_offset): New function.
8275 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
8276 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
8277 * config/vax/vax.c (vax_starting_frame_offset): New function.
8278 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
8279 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8280 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
8281 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
8282 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8283 * system.h (STARTING_FRAME_OFFSET): Poison.
8285 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8287 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
8288 SCALAR_TYPE_MODE instead of TYPE_MODE.
8290 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8291 Alan Hayward <alan.hayward@arm.com>
8292 David Sherwood <david.sherwood@arm.com>
8294 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
8296 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8297 Alan Hayward <alan.hayward@arm.com>
8298 David Sherwood <david.sherwood@arm.com>
8300 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
8302 2017-10-23 Richard Biener <rguenther@suse.de>
8304 PR tree-optimization/82129
8305 * tree-ssa-pre.c (bitmap_set_and): Remove.
8306 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
8307 canonicalizing expressions in the set to those with lowest
8308 ID rather than taking that from the first edge.
8310 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8312 * combine.c (rtx_equal_for_field_assignment_p): Use
8313 byte_lowpart_offset.
8315 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8316 Alan Hayward <alan.hayward@arm.com>
8317 David Sherwood <david.sherwood@arm.com>
8319 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
8320 to a SUBREG_PROMOTED_VAR.
8322 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8323 Alan Hayward <alan.hayward@arm.com>
8324 David Sherwood <david.sherwood@arm.com>
8326 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
8327 (expand_debug_source_expr): Likewise.
8328 * combine.c (combine_simplify_rtx): Likewise.
8329 * cse.c (fold_rtx): Likewise.
8330 * optabs.c (expand_float): Likewise.
8331 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8332 (simplify_binary_operation_1): Likewise.
8334 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8335 Alan Hayward <alan.hayward@arm.com>
8336 David Sherwood <david.sherwood@arm.com>
8338 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
8339 (record_promoted_value): Likewise.
8340 * expr.c (expand_expr_real_2): Likewise.
8341 * ree.c (update_reg_equal_equiv_notes): Likewise.
8342 (combine_set_extension): Likewise.
8343 * rtlanal.c (low_bitmask_len): Likewise.
8344 * simplify-rtx.c (neg_const_int): Likewise.
8345 (simplify_binary_operation_1): Likewise.
8347 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8348 Alan Hayward <alan.hayward@arm.com>
8349 David Sherwood <david.sherwood@arm.com>
8351 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
8352 * regcprop.c (maybe_mode_change): Likewise.
8353 * reload1.c (alter_reg): Likewise.
8355 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8357 * inchash.h (inchash::hash::add_wide_int): New function.
8358 * lto-streamer-out.c (hash_tree): Use it.
8360 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8362 * inchash.h (inchash::hash::add_wide_int): Rename to...
8363 (inchash::hash::add_hwi): ...this.
8364 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
8365 (polymorphic_call_target_hasher::hash): Likewise.
8366 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
8367 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
8368 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
8369 * lto-streamer-out.c (hash_tree): Likewise.
8370 * optc-save-gen.awk: Likewise.
8371 * tree.c (add_expr): Likewise.
8373 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
8376 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
8377 for ordered inequality comparisons even with TARGET_IEEE_FP.
8379 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
8382 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
8383 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
8384 Expand with cmp<dwi>_doubleword.
8386 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8388 * extend.texi: Add x86 specific to 'nocf_check' attribute.
8389 List CET intrinsics.
8390 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
8391 specific to -fcf-protection option.
8393 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8395 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
8396 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
8397 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
8398 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
8399 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
8400 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
8401 (extra_objs): Add cet.o for Linux/x86 targets.
8402 (tmake_file): Add i386/t-cet for Linux/x86 targets.
8403 * config/i386/cet.c: New file.
8404 * config/i386/cetintrin.h: Likewise.
8405 * config/i386/t-cet: Likewise.
8406 * config/i386/cpuid.h (bit_SHSTK): New.
8407 (bit_IBT): Likewise.
8408 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
8409 pass IBT and SHSTK bits.
8410 * config/i386/i386-builtin-types.def
8411 (VOID_FTYPE_UNSIGNED_PVOID): New.
8412 (VOID_FTYPE_UINT64_PVOID): Likewise.
8413 * config/i386/i386-builtin.def: Add CET intrinsics.
8414 * config/i386/i386-c.c (ix86_target_macros_internal): Add
8415 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
8416 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
8417 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
8419 * config/i386/i386.c (rest_of_insert_endbranch): New.
8420 (pass_data_insert_endbranch): Likewise.
8421 (pass_insert_endbranch): Likewise.
8422 (make_pass_insert_endbranch): Likewise.
8423 (ix86_notrack_prefixed_insn_p): Likewise.
8424 (ix86_target_string): Add -mibt, -mshstk flags.
8425 (ix86_option_override_internal): Add flag_cf_protection
8427 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
8428 (ix86_print_operand): Add 'notrack' prefix output.
8429 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
8430 (ix86_expand_builtin): Expand CET intrinsics.
8431 (x86_output_mi_thunk): Add 'endbranch' instruction.
8432 * config/i386/i386.h (TARGET_IBT): New.
8433 (TARGET_IBT_P): Likewise.
8434 (TARGET_SHSTK): Likewise.
8435 (TARGET_SHSTK_P): Likewise.
8436 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
8437 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
8438 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
8439 (builtin_setjmp_setup): New pattern.
8440 (builtin_longjmp): Likewise.
8441 (rdssp<mode>): Likewise.
8442 (incssp<mode>): Likewise.
8443 (saveprevssp): Likewise.
8444 (rstorssp): Likewise.
8445 (wrss<mode>): Likewise.
8446 (wruss<mode>): Likewise.
8447 (setssbsy): Likewise.
8448 (clrssbsy): Likewise.
8449 (nop_endbr): Likewise.
8450 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
8452 * config/i386/immintrin.h: Include <cetintrin.h>.
8453 * config/i386/linux-common.h
8454 (file_end_indicate_exec_stack_and_cet): New prototype.
8455 (TARGET_ASM_FILE_END): New.
8457 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
8459 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
8460 latencies instead of having separate table; make difference between
8461 integer and float costs.
8462 * i386.h (processor_costs): Remove scalar_stmt_cost,
8463 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
8464 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
8466 * x86-tune-costs.h: Remove entries which has been removed in
8467 procesor_costs from all tables; make cond_taken_branch_cost
8468 and cond_not_taken_branch_cost COST_N_INSNS based.
8470 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
8472 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
8474 2017-10-20 Jakub Jelinek <jakub@redhat.com>
8476 * config/i386/i386.md (isa): Remove fma_avx512f.
8477 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
8478 <avx512>_fmadd_<mode>_mask3<round_name>,
8479 <avx512>_fmsub_<mode>_mask<round_name>,
8480 <avx512>_fmsub_<mode>_mask3<round_name>,
8481 <avx512>_fnmadd_<mode>_mask<round_name>,
8482 <avx512>_fnmadd_<mode>_mask3<round_name>,
8483 <avx512>_fnmsub_<mode>_mask<round_name>,
8484 <avx512>_fnmsub_<mode>_mask3<round_name>,
8485 <avx512>_fmaddsub_<mode>_mask<round_name>,
8486 <avx512>_fmaddsub_<mode>_mask3<round_name>,
8487 <avx512>_fmsubadd_<mode>_mask<round_name>,
8488 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
8489 (*vec_widen_umult_even_v16si<mask_name>,
8490 *vec_widen_smult_even_v16si<mask_name>): Likewise.
8491 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
8493 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8495 * extend.texi: Add 'nocf_check' documentation.
8496 * gimple.texi: Add second parameter to
8497 gimple_build_call_from_tree.
8498 * invoke.texi: Add -fcf-protection documentation.
8499 * rtl.texi: Add REG_CALL_NOTRACK documenation.
8501 2017-10-20 Richard Biener <rguenther@suse.de>
8503 PR tree-optimization/82473
8504 * tree-vect-loop.c (vectorizable_reduction): Properly get at
8505 the largest input type.
8507 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8509 * c-attribs.c (handle_nocf_check_attribute): New function.
8510 (c_common_attribute_table): Add 'nocf_check' handling.
8511 * gimple-parser.c: Add second argument NULL to
8512 gimple_build_call_from_tree.
8513 * attrib.c (comp_type_attributes): Check nocf_check attribute.
8514 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
8516 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
8517 * common.opt: Add fcf-protection flag.
8518 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
8519 * flag-types.h: Add enum cf_protection_level.
8520 * gimple.c (gimple_build_call_from_tree): Add second parameter.
8521 Add 'nocf_check' attribute propagation to gimple call.
8522 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
8523 (gimple_build_call_from_tree): Update prototype.
8524 (gimple_call_nocf_check_p): New function.
8525 (gimple_call_set_nocf_check): Likewise.
8526 * gimplify.c: Add second argument to gimple_build_call_from_tree.
8527 * ipa-icf.c: Add nocf_check attribute in statement hash.
8528 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
8529 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
8530 * toplev.c (process_options): Add flag_cf_protection handling.
8532 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8534 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
8536 2017-10-20 Richard Biener <rguenther@suse.de>
8538 PR tree-optimization/82603
8539 * tree-if-conv.c (predicate_mem_writes): Make sure to only
8540 remove false predicated stores.
8542 2017-10-20 Richard Biener <rguenther@suse.de>
8544 * graphite-isl-ast-to-gimple.c
8545 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
8546 Remove return value and simplify, dump copied stmt after lhs
8548 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
8549 Reduce dump verbosity.
8550 (gsi_insert_earliest): Likewise.
8551 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
8552 * graphite.c (print_global_statistics): Adjust dumping.
8553 (print_graphite_scop_statistics): Likewise.
8554 (print_graphite_statistics): Do not dump loops here.
8555 (graphite_transform_loops): But here.
8557 2017-10-20 Nicolas Roche <roche@adacore.com>
8559 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
8560 * configure: Regenerate.
8562 2017-10-20 Jakub Jelinek <jakub@redhat.com>
8565 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
8566 functions when optimizing replace GIMPLE_RETURN stmts with
8567 calls to __builtin_unreachable ().
8570 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
8571 for -fsanitize=thread link of executables.
8572 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
8573 link of executables.
8576 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
8578 (<shift_insn><mode>3<mask_name>): Change the last of the 3
8579 define_insns for logical vector shifts to use VI248_AVX512BW
8580 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
8581 condition, useless isa and prefix attributes. Change the first
8582 2 of these define_insns to ...
8583 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
8584 define_insn for avx512vl.
8585 (<shift_insn><mode>3): ... and this, new define_insn without
8586 masking for non-avx512vl.
8589 * config/i386/sse.md (*andnot<mode>3,
8590 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
8591 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
8592 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
8593 not applied use empty suffix even for TARGET_AVX512VL.
8594 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
8595 is applied, supply evex,evex or evex,evex,evex instead of just
8598 2017-10-20 Julia Koval <julia.koval@intel.com>
8600 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
8601 (OPTION_MASK_ISA_GFNI_UNSET): New.
8602 (ix86_handle_option): Handle OPT_mgfni.
8603 * config/i386/cpuid.h (bit_GFNI): New.
8604 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
8605 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
8606 * config/i386/i386.c (ix86_target_string): Add -mgfni.
8607 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
8608 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
8609 * config/i386/i386.opt: Add mgfni.
8611 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
8613 * config/msp430/msp430.c (msp430_option_override): Disable
8614 -fdelete-null-pointer-checks.
8615 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
8617 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8619 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
8620 of x87 and SSE instructions.
8622 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8624 * asan.c (create_cond_insert_point): Do not update edge count.
8625 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
8626 (afdo_propagate_circuit): Likewise.
8627 (afdo_calculate_branch_prob): Likewise.
8628 (afdo_annotate_cfg): Likewise.
8629 * basic-block.h (struct edge_def): Remove count.
8630 (edge_def::count): New accessor.
8631 * bb-reorder.c (rotate_loop): Update.
8632 (find_traces_1_round): Update.
8633 (connect_traces): Update.
8634 (sanitize_hot_paths): Update.
8635 * cfg.c (unchecked_make_edge): Update.
8636 (make_single_succ_edge): Update.
8637 (check_bb_profile): Update.
8638 (dump_edge_info): Update.
8639 (update_bb_profile_for_threading): Update.
8640 (scale_bbs_frequencies_int): Update.
8641 (scale_bbs_frequencies_gcov_type): Update.
8642 (scale_bbs_frequencies_profile_count): Update.
8643 (scale_bbs_frequencies): Update.
8644 * cfganal.c (connect_infinite_loops_to_exit): Update.
8645 * cfgbuild.c (compute_outgoing_frequencies): Update.
8646 (find_many_sub_basic_blocks): Update.
8647 * cfgcleanup.c (try_forward_edges): Update.
8648 (try_crossjump_to_edge): Update
8649 * cfgexpand.c (expand_gimple_cond): Update
8650 (expand_gimple_tailcall): Update
8651 (construct_exit_block): Update
8652 * cfghooks.c (verify_flow_info): Update
8653 (redirect_edge_succ_nodup): Update
8654 (split_edge): Update
8655 (make_forwarder_block): Update
8656 (duplicate_block): Update
8657 (account_profile_record): Update
8658 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
8659 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
8660 * cfgloopmanip.c (scale_loop_profile): Update.
8662 (lv_adjust_loop_entry_edge): Update.
8663 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
8664 (force_nonfallthru_and_redirect): Update.
8665 (purge_dead_edges): Update.
8666 (rtl_flow_call_edges_add): Update.
8667 * cgraphunit.c (init_lowered_empty_function): Update.
8668 (cgraph_node::expand_thunk): Update.
8669 * gimple-pretty-print.c (dump_probability): Update.
8670 (dump_edge_probability): Update.
8671 * gimple-ssa-isolate-paths.c (isolate_path): Update.
8672 * haifa-sched.c (sched_create_recovery_edges): Update.
8673 * hsa-gen.c (convert_switch_statements): Update.
8674 * ifcvt.c (dead_or_predicable): Update.
8675 * ipa-inline-transform.c (inline_transform): Update.
8676 * ipa-split.c (split_function): Update.
8677 * ipa-utils.c (ipa_merge_profiles): Update.
8678 * loop-doloop.c (add_test): Update.
8679 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
8680 * lto-streamer-in.c (input_cfg): Update.
8681 (input_function): Update.
8682 * lto-streamer-out.c (output_cfg): Update.
8683 * modulo-sched.c (sms_schedule): Update.
8684 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
8685 * predict.c (maybe_hot_edge_p): Update.
8686 (unlikely_executed_edge_p): Update.
8687 (probably_never_executed_edge_p): Update.
8688 (dump_prediction): Update.
8689 (drop_profile): Update.
8690 (propagate_unlikely_bbs_forward): Update.
8691 (determine_unlikely_bbs): Update.
8692 (force_edge_cold): Update.
8693 * profile.c (compute_branch_probabilities): Update.
8694 * reg-stack.c (better_edge): Update.
8695 * shrink-wrap.c (handle_simple_exit): Update.
8696 * tracer.c (better_p): Update.
8697 * trans-mem.c (expand_transaction): Update.
8698 (split_bb_make_tm_edge): Update.
8699 * tree-call-cdce.c: Update.
8700 * tree-cfg.c (gimple_find_sub_bbs): Update.
8701 (gimple_split_edge): Update.
8702 (gimple_duplicate_sese_region): Update.
8703 (gimple_duplicate_sese_tail): Update.
8704 (gimple_flow_call_edges_add): Update.
8705 (insert_cond_bb): Update.
8706 (execute_fixup_cfg): Update.
8707 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
8708 * tree-complex.c (expand_complex_div_wide): Update.
8709 * tree-eh.c (lower_resx): Update.
8710 (unsplit_eh): Update.
8711 (cleanup_empty_eh_move_lp): Update.
8712 * tree-inline.c (copy_edges_for_bb): Update.
8713 (freqs_to_counts): Update.
8714 (copy_cfg_body): Update.
8715 * tree-ssa-dce.c (remove_dead_stmt): Update.
8716 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
8717 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
8718 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
8719 (unloop_loops): Update.
8720 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8721 * tree-ssa-loop-split.c (connect_loops): Update.
8722 (split_loop): Update.
8723 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
8724 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
8725 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
8726 * tree-ssa-reassoc.c (branch_fixup): Update.
8727 * tree-ssa-tail-merge.c (replace_block_by): Update.
8728 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
8729 (compute_path_counts): Update.
8730 (update_profile): Update.
8731 (recompute_probabilities): Update.
8732 (update_joiner_offpath_counts): Update.
8733 (estimated_freqs_path): Update.
8734 (freqs_to_counts_path): Update.
8735 (clear_counts_path): Update.
8736 (ssa_fix_duplicate_block_edges): Update.
8737 (duplicate_thread_path): Update.
8738 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
8739 (case_bit_test_cmp): Update.
8740 (collect_switch_conv_info): Update.
8741 (gen_inbound_check): Update.
8742 (do_jump_if_equal): Update.
8743 (emit_cmp_and_jump_insns): Update.
8744 * tree-tailcall.c (decrease_profile): Update.
8745 (eliminate_tail_call): Update.
8746 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
8747 (vect_do_peeling): Update.
8748 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8749 * ubsan.c (ubsan_expand_null_ifn): Update.
8750 (ubsan_expand_ptr_ifn): Update.
8751 * value-prof.c (gimple_divmod_fixed_value): Update.
8752 (gimple_mod_pow2): Update.
8753 (gimple_mod_subtract): Update.
8754 (gimple_ic): Update.
8755 (gimple_stringop_fixed_value): Update.
8757 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8760 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
8762 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
8764 PR rtl-optimization/82395
8765 * ira-color.c (allocno_priority_compare_func): Fix comparison step
8766 based on non_spilled_static_chain_regno_p.
8768 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8770 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
8771 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
8772 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
8774 2017-10-19 Martin Sebor <msebor@redhat.com>
8776 PR tree-optimization/82596
8777 * tree.c (array_at_struct_end_p): Handle STRING_CST.
8779 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8781 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
8782 (get_mem_refs_of_builtin_call): Likewise.
8783 * builtins.c (expand_builtin_apply): Adjust call to
8784 allocate_dynamic_stack_space.
8785 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
8786 the third argument to allocate_dynamic_stack_space, otherwise -1.
8787 (expand_builtin): Deal with all alloca variants.
8788 (is_inexpensive_builtin): Likewise.
8789 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
8790 * calls.c (special_function_p): Deal with all alloca variants.
8791 (initialize_argument_information): Adjust call to
8792 allocate_dynamic_stack_space.
8793 (expand_call): Likewise.
8794 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
8795 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
8796 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
8797 use it for the stack usage computation.
8798 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
8799 * function.c (gimplify_parameters): Call build_alloca_call_expr.
8800 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
8801 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
8802 (in_loop_p): Remove first argument and useless check.
8803 (pass_walloca::execute): Remove useless test and adjust call to above.
8804 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
8805 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
8806 (gimplify_call_expr): Deal with all alloca variants.
8807 * hsa-gen.c (gen_hsa_alloca): Likewise.
8808 (gen_hsa_insns_for_call): Likewise.
8809 * ipa-pure-const.c (special_builtin_state): Likewise.
8810 * tree-chkp.c (chkp_build_returned_bound): Likewise.
8811 * tree-object-size.c (alloc_object_size): Likewise.
8812 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
8813 (call_may_clobber_ref_p_1): Likewise.
8814 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
8815 (ccp_fold_stmt): Likewise.
8816 (optimize_stack_restore): Likewise.
8817 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
8818 (mark_all_reaching_defs_necessary_1): Likewise.
8819 (propagate_necessity): Likewise.
8820 (eliminate_unnecessary_stmts): Likewise.
8821 * tree.c (build_common_builtin_nodes): Build
8822 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
8823 (build_alloca_call_expr): New function.
8824 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
8825 (CASE_BUILT_IN_ALLOCA): Likewise.
8826 (build_alloca_call_expr): Declare.
8827 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
8829 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8832 * dwarf2out.c (new_die_raw): New static inline function.
8833 (new_die): Use it to create the DIE.
8834 (add_AT_external_die_ref): Likewise.
8835 (clone_die): Likewise.
8836 (clone_as_declaration): Likewise.
8837 (dwarf2out_vms_debug_main_pointer): Likewise.
8838 (base_type_die): Likewise. Remove early return for corner cases.
8839 Do not call add_pubtype on the DIE here.
8840 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
8841 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
8842 typedefs for base types with DW_AT_endianity. Make sure a DIE with
8843 native order exists for base types, attach the DIE manually and call
8844 add_pubtype on it. Do not equate a reverse order DIE to the type.
8846 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
8848 * config/arm/arm.c (align_ok_ldrd_strd): New function.
8849 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
8851 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
8853 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8855 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
8856 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
8857 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
8858 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
8859 * opts.c (sanitizer_opts): Add builtin.
8860 * ubsan.c (instrument_builtin): New function.
8861 (pass_ubsan::execute): Call it.
8862 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
8863 * doc/invoke.texi: Document -fsanitize=builtin.
8865 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
8866 builtins, store max (log2 (align), 0) into uchar field instead of
8867 align into uptr field.
8868 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
8869 store uchar 0 field instead of uptr 0 field.
8870 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
8871 instead of passing one address of struct with 2 locations pass
8872 two addresses of structs with 1 location each.
8873 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
8874 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
8875 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
8876 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
8877 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
8878 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
8879 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
8880 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
8882 2017-10-19 Martin Liska <mliska@suse.cz>
8885 * file-find.c (remove_prefix): Remove.
8886 * file-find.h (remove_prefix): Likewise.
8887 * gcc-ar.c: Remove smartness of lookup.
8889 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
8891 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
8892 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
8893 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
8895 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8898 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
8899 (setcc + and to xor + setcc): New peephole2.
8901 2017-10-19 Tom de Vries <tom@codesourcery.com>
8903 * doc/sourcebuild.texi (Test Directives, Variants of
8904 dg-require-support): Add dg-require-stack-size.
8906 2017-10-19 Martin Liska <mliska@suse.cz>
8909 * gimplify.c (gimplify_decl_expr): Do not instrument variables
8910 that have a large alignment.
8911 (gimplify_target_expr): Likewise.
8913 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
8915 PR rtl-optimization/82602
8916 * ira.c (rtx_moveable_p): Return false for volatile asm.
8918 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
8921 * config/i386/i386-modes.def (CCGZ): New CC mode.
8922 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
8923 * config/i386/predicates.md (ix86_comparison_operator):
8925 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
8926 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
8927 with double-word subtraction.
8928 (put_condition_code): Handle CCGZmode.
8930 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
8932 * wide-int.cc (debug (const wide_int &)): New.
8933 (debug (const wide_int *)): New.
8934 (debug (const widest_int &)): New.
8935 (debug (const widest_int *)): New.
8937 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
8940 * lra-constraints.c (curr_insn_transform): Use non-input operand
8941 instead of output one for matched reload.
8943 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8945 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
8946 (tree-ssa-loop-ivopts.h): New header file.
8947 (struct builtin_info): New fields.
8948 (classify_builtin_1): Compute and record base and offset parts for
8949 memset builtin partition by calling strip_offset.
8950 (offset_cmp, fuse_memset_builtins): New functions.
8951 (finalize_partitions): Fuse adjacent memset partitions by calling
8953 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
8954 Expose the interface.
8955 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
8957 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8959 PR tree-optimization/82574
8960 * tree-loop-distribution.c (find_single_drs): New parameter. Check
8961 that data reference must be executed exactly once per iteration
8962 against the outermost loop in nest.
8963 (classify_partition): Update call to above function.
8965 2017-10-18 Richard Biener <rguenther@suse.de>
8967 PR tree-optimization/82591
8968 * graphite.c (graphite_transform_loops): Move code gen message
8970 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8971 Here. Handle scop_to_isl_ast failing.
8972 (scop_to_isl_ast): Limit the number of ISL operations.
8974 2017-10-18 Richard Biener <rguenther@suse.de>
8976 * graphite-isl-ast-to-gimple.c
8977 (translate_isl_ast_to_gimple::set_rename): Simplify.
8978 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
8979 (graphite_copy_stmts_from_block): ... here.
8980 (copy_bb_and_scalar_dependences): Simplify.
8981 (add_parameters_to_ivs_params): Canonicalize.
8982 (generate_entry_out_of_ssa_copies): Simplify.
8983 * graphite-sese-to-poly.c (extract_affine_name): Simplify
8984 by passing in ISL dimension.
8985 (parameter_index_in_region_1): Rename to ...
8986 (parameter_index_in_region): ... this.
8987 (extract_affine): Adjust assert, pass down parameter index.
8988 (add_param_constraints): Use range-info when available.
8989 (build_scop_context): Adjust.
8990 * sese.c (new_sese_info): Adjust.
8991 (free_sese_info): Likewise.
8992 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
8993 Remove unused typedefs.
8994 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
8996 2017-10-18 Martin Liska <mliska@suse.cz>
8998 * combine.c (simplify_compare_const): Add gcc_fallthrough.
9000 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
9002 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
9003 (s390_sched_init): Do not reset s390_sched_state if we entered the
9004 current basic block via a fallthru edge and all others are unlikely.
9006 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
9008 * config/s390/s390.c (NUM_SIDES): New variable.
9009 (LONGRUNNING_THRESHOLD): New variable.
9010 (LATENCY_FACTOR): New variable.
9011 (s390_sched_score): Decrease score for long-running instructions on
9013 (s390_sched_variable_issue): Perform bookkeeping for long-running
9016 2017-10-18 Richard Biener <rguenther@suse.de>
9018 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
9019 Simplify with removal of the parameter rename map.
9020 (set_rename): Likewise.
9021 (should_copy_to_new_region): Likewise.
9022 (graphite_copy_stmts_from_block): Likewise.
9023 (copy_bb_and_scalar_dependences): Remove initialization of
9024 unused copied_bb_map.
9026 (copy_internal_parameters): Likewise.
9027 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
9028 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
9029 Use INTEGRAL_TYPE_P.
9030 (parameter_index_in_region_1): Rename to ...
9031 (assign_parameter_index_in_region): ... this. Assert we have
9032 a parameter we handle.
9033 (scan_tree_for_params): Adjust.
9034 * sese.h (parameter_rename_map_t): Remove.
9035 (struct sese_info_t): Remove unused parameter_rename_map and
9036 copied_bb_map members.
9037 * sese.c (new_sese_info): Adjust.
9038 (free_sese_info): Likewise.
9040 2017-10-18 Martin Liska <mliska@suse.cz>
9043 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
9044 on an abnormal edge.
9046 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
9048 * doc/invoke.texi (ffunction-sections and fdata-sections):
9051 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
9053 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
9054 the use statement can throw internally.
9056 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
9058 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
9059 any RTX present on the RHS of a SET.
9060 * compare-elim.c (try_eliminate_compare): Restore comment.
9062 2017-10-17 Jakub Jelinek <jakub@redhat.com>
9064 * langhooks.h (struct lang_hooks): Document that tree_size langhook
9065 may be also called on tcc_type nodes.
9066 * langhooks.c (lhd_tree_size): Likewise.
9068 2017-10-17 David Malcolm <dmalcolm@redhat.com>
9070 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
9071 format_warning_at_substring.
9072 (maybe_warn): Convert source_range * param to a location_t. Pass
9073 UNKNOWN_LOCATION rather than NULL to fmtwarn.
9074 (format_directive): Remove code to extract source_ranges and
9075 source_range * in favor of just a location_t.
9076 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
9078 * substring-locations.c (format_warning_va): Convert
9079 source_range * param to a location_t.
9080 (format_warning_at_substring): Likewise.
9081 * substring-locations.h (format_warning_va): Likewise.
9082 (format_warning_at_substring): Likewise.
9084 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
9086 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
9088 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
9089 and scatter/gather ops.
9091 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
9092 vec_gather_load and vec_scatter_store.
9093 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
9094 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
9095 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
9096 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
9097 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
9098 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
9100 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
9102 * reg-stack.c (compare_for_stack_reg): Add bool argument.
9103 Detect FTST instruction and handle its register pops. Only pop
9104 second operand if can_pop_second_op is true.
9105 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
9106 set can_pop_second_op to false in the compare_for_stack_reg call.
9108 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
9109 output_fp_compare for stack register operands.
9110 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
9111 instructions here. Do not emit stack register pops here. Assert
9112 that FCOMPP pops next to top stack register. Rewrite function.
9114 2017-10-17 Nathan Sidwell <nathan@acm.org>
9117 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
9118 use DECL_ASSEMBLER_NAME_RAW.
9121 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
9124 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
9125 Wilco Dijkstra <wilco.dijkstra@arm.com>
9127 * builtins.c (expand_builtin_update_setjmp_buf): Add a
9128 converstion to Pmode from the buf_addr.
9130 2017-10-17 Richard Biener <rguenther@suse.de>
9132 * graphite-dependences.c (scop_get_reads_and_writes): Change
9133 output parameters to references.
9135 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
9137 PR 71026/tree-optimization
9138 * fold-const.c (distribute_real_division): Removed.
9139 (fold_binary_loc): Remove calls to distribute_real_divison.
9141 2017-10-17 Richard Biener <rguenther@suse.de>
9143 * graphite-scop-detection.c
9144 (scop_detection::stmt_has_simple_data_refs_p): Always use
9145 the full nest as region.
9146 (try_generate_gimple_bb): Likewise.
9147 * sese.c (scalar_evolution_in_region): Simplify now that
9148 SCEV can handle instantiation in regions.
9149 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
9150 in the non-loop part of a function if requested.
9152 2017-10-17 Richard Biener <rguenther@suse.de>
9154 PR tree-optimization/82563
9155 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
9157 (graphite_regenerate_ast_isl): Call it.
9158 * graphite-scop-detection.c (build_scops): Remove entry edge split.
9160 2017-10-17 Jakub Jelinek <jakub@redhat.com>
9162 PR tree-optimization/82549
9163 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
9164 Formatting fixes. Instead of calling make_bit_field_ref with negative
9167 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
9169 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
9170 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
9171 _mm_maskz_reduce_ss): New.
9172 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
9173 __builtin_ia32_reducess_mask): Ditto..
9174 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
9175 * config/i386/sse.md (reduces<mode>): Renamed to ...
9176 (reduces<mode><mask_scalar_name>): ... this.
9177 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
9179 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
9180 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
9182 2017-10-16 David Malcolm <dmalcolm@redhat.com>
9184 * Makefile.in (OBJS): Add unique-ptr-tests.o.
9185 * selftest-run-tests.c (selftest::run_tests): Call
9186 selftest::unique_ptr_tests_cc_tests.
9187 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
9188 * unique-ptr-tests.cc: New file.
9190 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
9193 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9195 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9196 (make_hard_regno_born, make_hard_regno_dead): Update
9197 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
9199 2017-10-16 Jeff Law <law@redhat.com>
9201 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
9203 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
9205 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
9207 2017-10-16 Olivier Hainque <hainque@adacore.com>
9209 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
9210 with_cpu if we were configured for an e500v2 target cpu name.
9212 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
9214 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
9215 * doc/invoke.texi: Document +nodsp as a valid extension for
9218 2017-10-16 Martin Liska <mliska@suse.cz>
9220 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
9221 (test_set_range): Likewise.
9222 (test_range_functions): Rename to ...
9223 (test_bit_in_range): ... this.
9224 (sbitmap_c_tests): Add new test.
9226 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9228 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
9230 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
9231 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
9233 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9235 * config/aarch64/aarch64-builtins.c
9236 (aarch64_types_quadopu_lane_qualifiers): New.
9237 (TYPES_QUADOPU_LANE): New.
9238 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
9239 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
9240 (aarch64_<sur>dot_laneq<vsi2qi>): New.
9241 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
9242 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
9243 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
9244 (Vdottype, DOTPROD): New.
9245 (sur): Add SDOT and UDOT.
9247 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9249 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
9250 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
9251 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
9253 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
9254 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
9255 Enable TARGET_DOTPROD.
9256 (cortex-a75.cortex-a55): Likewise.
9257 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
9259 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9261 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
9262 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
9264 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
9266 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
9267 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
9268 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
9269 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
9270 * config/arm/types.md (neon_dot, neon_dot_q): New.
9271 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
9273 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9275 * config/arm/arm.h (TARGET_DOTPROD): New.
9276 * config/arm/arm.c (arm_arch_dotprod): New.
9277 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
9278 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
9279 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
9280 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
9281 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
9282 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
9283 * doc/invoke.texi (armv8.2-a): Document dotprod
9285 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
9287 * i386.c (ix86_vec_cost): New function.
9288 (ix86_rtx_costs): Handle vector operations better.
9289 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
9290 * x86-tune-costs.h: Add new costs to all tables.
9292 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
9294 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
9296 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
9297 divsd, sqrtss and sqrtsd
9298 * x86-tune-costs.h: Add new entries to all costs.
9299 (znver1_cost): Fix to match real instruction latencies.
9301 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9302 Michael Collison <michael.collison@arm.com>
9304 * compare-elim.c: Include emit-rtl.h.
9305 (can_merge_compare_into_arith): New function.
9306 (try_validate_parallel): Likewise.
9307 (try_merge_compare): Likewise.
9308 (try_eliminate_compare): Call the above when no previous clobber
9310 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
9313 2017-10-14 Jakub Jelinek <jakub@redhat.com>
9317 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
9318 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
9322 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
9323 to be any operand_equal_p operands. For & (B - 1) require
9324 B to be power of 2. Recognize
9325 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
9327 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
9330 * optabs.c (expand_memory_blockage): Fix call of
9331 targetm.have_memory_blockage.
9333 2017-10-14 Jakub Jelinek <jakub@redhat.com>
9336 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
9337 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
9338 objects to extra_objs instead of overwriting it.
9340 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
9342 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
9343 Use any_fp_register_operand as operand[3] predicate. Simplify
9344 equality test for operands[2] and operands[4] memory location.
9345 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
9346 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
9347 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
9348 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
9349 any_fp_register_operand as operand[1] predicate. Simplify
9350 equality test for operands[0] and operands[3] memory location.
9351 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
9352 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
9353 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
9355 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
9357 * target-insns.def: Add memory_blockage.
9358 * optabs.c (expand_memory_blockage): New function.
9359 (expand_asm_memory_barrier): Rename ...
9360 (expand_asm_memory_blockage): ... to this.
9361 (expand_mem_thread_fence): Call expand_memory_blockage
9362 instead of expand_asm_memory_barrier.
9363 (expand_mem_singnal_fence): Ditto.
9364 (expand_atomic_load): Ditto.
9365 (expand_atomic_store): Ditto.
9366 * doc/md.texi (Standard Pattern Names For Generation):
9367 Document memory_blockage instruction pattern.
9369 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
9371 * config/rl78/rl78.c (rl78_emit_libcall): New function.
9372 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
9373 * config/rl78/rl78.md: New define_expand "adddi3".
9375 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
9377 * cfghooks.c (verify_flow_info): Disable check that all probabilities
9380 2017-10-13 Jeff Law <law@redhat.com>
9382 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
9384 2017-10-13 Jakub Jelinek <jakub@redhat.com>
9387 * internal-fn.c (expand_mul_overflow): If both operands have
9388 the same highpart of -1 or 0 and the topmost bit of lowpart
9389 is different, overflow is if res <= 0 rather than res < 0.
9391 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
9393 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
9394 TARGET_P9_VECTOR code for unaligned_load case.
9396 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
9398 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
9400 2017-10-13 Nathan Sidwell <nathan@acm.org>
9402 * tree-core.h (tree_contains_struct): Make bool.
9403 * tree.c (tree_contains_struct): Likewise.
9404 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
9405 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
9406 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
9407 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
9408 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
9410 2017-10-13 Richard Biener <rguenther@suse.de>
9412 * graphite-isl-ast-to-gimple.c
9413 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
9414 parameters and dominance check.
9415 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
9416 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
9417 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
9418 Do not update SSA form here or do intermediate IL verification.
9419 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
9420 (graphite_initialize): Remove check on the number of loops in
9421 the function and inline into graphite_transform_loops.
9422 (graphite_finalize): Inline into graphite_transform_loops.
9423 (graphite_transform_loops): Perform SSA update and IL verification
9425 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
9427 2017-10-13 Richard Biener <rguenther@suse.de>
9429 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
9430 graphite_expression_type_precision): Avoid global constructor
9432 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
9433 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
9434 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
9435 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
9436 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
9437 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
9439 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
9442 * config/i386/i386.h (ix86_red_zone_size): New.
9443 * config/i386/i386.md (push peephole2s): Replace
9444 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
9446 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9447 Alan Hayward <alan.hayward@arm.com>
9448 David Sherwood <david.sherwood@arm.com>
9450 * combine.c (can_change_dest_mode): Reject changes in
9451 REGMODE_NATURAL_SIZE.
9453 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9454 Alan Hayward <alan.hayward@arm.com>
9455 David Sherwood <david.sherwood@arm.com>
9457 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
9458 (expand_debug_source_expr): Likewise.
9459 * combine.c (combine_simplify_rtx): Likewise.
9460 * cse.c (fold_rtx): Likewise.
9461 * fwprop.c (canonicalize_address): Likewise.
9462 * targhooks.c (default_shift_truncation_mask): Likewise.
9464 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9465 Alan Hayward <alan.hayward@arm.com>
9466 David Sherwood <david.sherwood@arm.com>
9468 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
9469 (widened_mode): Likewise.
9470 (expand_unop): Likewise.
9471 * ree.c (transform_ifelse): Likewise.
9472 (merge_def_and_ext): Likewise.
9473 (combine_reaching_defs): Likewise.
9474 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9476 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9477 Alan Hayward <alan.hayward@arm.com>
9478 David Sherwood <david.sherwood@arm.com>
9480 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
9481 * combine.c (gen_lowpart_for_combine): Likewise.
9482 * dwarf2out.c (rtl_for_decl_location): Likewise.
9483 * final.c (alter_subreg): Likewise.
9484 * rtlhooks.c (gen_lowpart_general): Likewise.
9485 (gen_lowpart_if_possible): Likewise.
9487 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9488 Alan Hayward <alan.hayward@arm.com>
9489 David Sherwood <david.sherwood@arm.com>
9491 * calls.c (expand_call): Use subreg_lowpart_offset.
9492 * cse.c (cse_insn): Likewise.
9493 * regcprop.c (copy_value): Likewise.
9494 (copyprop_hardreg_forward_1): Likewise.
9496 2017-10-13 Jakub Jelinek <jakub@redhat.com>
9499 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
9500 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
9501 =Q constraints to +Q and into insn condition add check
9502 that operands[0] and operands[1] are equal.
9503 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
9504 =Q constraints to +Q and into insn condition add check
9505 that operands[0] is equal to either operands[1] or operands[2].
9508 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
9509 instead of handling MINUS_EXPR twice (once for each argument),
9510 canonicalize operand order and handle just once, use rtype where
9511 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
9514 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
9515 any values of __C while still being pattern recognizable as a simple
9518 2017-10-13 Richard Biener <rguenther@suse.de>
9520 PR tree-optimization/82451
9522 2017-10-02 Richard Biener <rguenther@suse.de>
9524 PR tree-optimization/82355
9525 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
9526 a mapping for the enclosing loop but avoid generating one for
9528 (copy_bb_and_scalar_dependences): Remove premature codegen
9529 error on PHIs in blocks duplicated into multiple places.
9530 * graphite-scop-detection.c
9531 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
9532 in the region use it as loop and nest to analyze the DR in.
9533 (try_generate_gimple_bb): Likewise.
9534 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
9535 (add_loop_constraints): For blocks in a loop not in the region
9536 create a dimension with a single iteration.
9537 * sese.h (gbb_loop_at_index): Remove assert.
9539 * cfgloop.c (loop_preheader_edge): For the loop tree root
9540 return the single successor of the entry block.
9541 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
9542 Reset the SCEV hashtable and niters.
9543 * graphite-scop-detection.c
9544 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
9545 assert that we only have POLYNOMIAL_CHREC that vary in loops
9546 contained in the region.
9547 (scop_detection::graphite_can_represent_expr): Adjust.
9548 (scop_detection::stmt_has_simple_data_refs_p): For loops
9549 not in the region set loop to NULL. The nest is now the
9550 entry edge to the region.
9551 (try_generate_gimple_bb): Likewise.
9552 * sese.c (scalar_evolution_in_region): Adjust for
9553 instantiate_scev change.
9554 * tree-data-ref.h (graphite_find_data_references_in_stmt):
9555 Make nest parameter the edge into the region.
9556 (create_data_ref): Likewise.
9557 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
9558 entry edge into a region and adjust instantiate_scev calls.
9559 (create_data_ref): Likewise.
9560 (graphite_find_data_references_in_stmt): Likewise.
9561 (find_data_references_in_stmt): Pass the loop preheader edge
9562 from the nest argument.
9563 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
9564 parameter the edge into the region.
9565 (instantiate_parameters): Use the loop preheader edge as entry.
9566 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
9568 (get_instantiated_value_entry): Make instantiate_below parameter
9569 the edge into the region.
9570 (instantiate_scev_name): Likewise. Adjust dominance checks,
9571 when we cannot use loop-based instantiation instantiate by
9572 walking use-def chains.
9573 (instantiate_scev_poly): Adjust.
9574 (instantiate_scev_binary): Likewise.
9575 (instantiate_scev_convert): Likewise.
9576 (instantiate_scev_not): Likewise.
9577 (instantiate_array_ref): Remove.
9578 (instantiate_scev_3): Likewise.
9579 (instantiate_scev_2): Likewise.
9580 (instantiate_scev_1): Likewise.
9581 (instantiate_scev_r): Do not blindly handle N-operand trees.
9582 Do not instantiate array-refs. Handle all constants and invariants.
9583 (instantiate_scev): Make instantiate_below parameter
9584 the edge into the region.
9585 (resolve_mixers): Use the loop preheader edge for the region
9586 parameter to instantiate_scev_r.
9587 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
9589 2017-10-13 Richard Biener <rguenther@suse.de>
9591 PR tree-optimization/82525
9592 * graphite-isl-ast-to-gimple.c
9593 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
9595 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
9596 Fail code generation when we cannot represent the isl integer.
9597 (binary_op_to_tree): Elide modulo operations that are no-ops
9598 in the type we code generate. Remove now superfluous code
9601 2017-10-13 Richard Biener <rguenther@suse.de>
9603 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
9604 (scop_detection::harmful_loop_in_region): Remove premature
9605 IV type restriction.
9606 (scop_detection::graphite_can_represent_scev): We can handle
9607 pointer IVs just fine.
9609 2017-10-13 Alan Modra <amodra@gmail.com>
9611 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
9612 "Clobbers and Scratch Registers". Add paragraph on
9613 alternative to clobbers for scratch registers and OpenBLAS
9616 2017-10-13 Alan Modra <amodra@gmail.com>
9618 * doc/extend.texi (Clobbers): Correct vax example. Delete old
9619 example of a memory input for a string of known length. Move
9620 commentary out of table. Add a number of new examples
9621 covering array memory inputs.
9623 2017-10-12 Martin Liska <mliska@suse.cz>
9625 PR tree-optimization/82493
9626 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
9627 (test_range_functions): New function.
9628 (sbitmap_c_tests): Likewise.
9629 * selftest-run-tests.c (selftest::run_tests): Run new tests.
9630 * selftest.h (sbitmap_c_tests): New function.
9632 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
9634 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
9636 * config/rs6000/amo.h: Fix spacing issue.
9638 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9641 * config/i386/i386.md (*ashl<mode>3_mask_1,
9642 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
9643 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
9646 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
9648 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
9649 (profile_probability): Set max_probability
9650 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
9652 * profile-count.c (profile_probability::differs_from_p): Do not
9653 rely on max_probaiblity == 10000
9655 2017-10-12 Jeff Law <law@redhat.com>
9657 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
9660 2017-10-12 Martin Sebor <msebor@redhat.com>
9664 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
9665 (handle_alias_pairs): Call it.
9666 * common.opt (-Wattribute-alias): New option.
9667 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
9668 * doc/invoke.texi (-Wattribute-alias): Document.
9670 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
9673 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9675 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9677 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9678 (make_hard_regno_born, make_hard_regno_dead): Update
9679 bb_killed_pseudos and bb_gen_pseudos.
9681 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
9683 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
9685 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
9687 * config/alpha/alpha.c (alpha_split_conditional_move):
9688 Use std::swap instead of manually swapping.
9689 (alpha_stdarg_optimize_hook): Ditto.
9690 (alpha_canonicalize_comparison): Ditto.
9692 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9694 * tree-loop-distribution.c (struct builtin_info): New struct.
9695 (struct partition): Refactor fields into struct builtin_info.
9696 (partition_free): Free struct builtin_info.
9697 (build_size_arg_loc, build_addr_arg_loc): Delete.
9698 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
9699 information from struct builtin_info.
9700 (find_single_drs): New function refactored from classify_partition.
9701 Also moved builtin validity checks to this function.
9702 (compute_access_range, alloc_builtin): New functions.
9703 (classify_builtin_st, classify_builtin_ldst): New functions.
9704 (classify_partition): Refactor code into functions find_single_drs,
9705 classify_builtin_st and classify_builtin_ldst.
9706 (distribute_loop): Don't do runtime alias check when distributing
9708 (find_seed_stmts_for_distribution): New function.
9709 (pass_loop_distribution::execute): Refactor code finding seed
9710 stmts into above function. Support distribution for the innermost
9711 two-level loop nest. Adjust dump information.
9713 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9715 * tree-loop-distribution.c: Adjust the general comment.
9716 (NUM_PARTITION_THRESHOLD): New macro.
9717 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
9718 (classify_partition): Skip builtin pattern of loop nest's inner loop.
9719 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
9720 in call to build_partition_graph.
9721 (finalize_partitions): New parameter. Make loop distribution more
9722 conservative by fusing more partitions.
9723 (distribute_loop): Don't do runtime alias check in case of loop nest
9725 (find_seed_stmts_for_distribution): New function.
9726 (prepare_perfect_loop_nest): New function.
9727 (pass_loop_distribution::execute): Refactor code finding seed stmts
9728 and loop nest into above functions. Support loop nest distribution.
9729 Adjust dump information accordingly.
9731 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9733 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
9734 and set PTYPE_SEQUENTIAL for merged partition.
9736 2017-10-12 Richard Biener <rguenther@suse.de>
9738 PR tree-optimization/69728
9740 2017-09-19 Richard Biener <rguenther@suse.de>
9742 PR tree-optimization/69728
9743 * graphite-sese-to-poly.c (schedule_error): New global.
9744 (add_loop_schedule): Handle empty domain by failing the
9746 (build_original_schedule): Handle schedule_error.
9748 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
9749 domain by returning an unchanged schedule.
9751 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9753 * genrecog.c (validate_pattern): For VEC_SELECT verify that
9754 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
9756 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
9758 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
9761 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9764 * expr.c (store_field): Don't optimize away bitsize == 0 store
9765 from CALL_EXPR with addressable return type.
9767 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9769 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
9770 * config/rs6000/rs6000.md (sel): Delete mode attribute.
9771 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
9772 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
9773 TARGET_ISEL instead of TARGET_ISEL<sel>.
9775 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
9777 * config/rs6000/rs6000.c
9778 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
9780 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9782 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
9784 (reg_or_cint_operand): Fix comment.
9785 (reg_or_zero_operand): New predicate.
9786 * config/rs6000/rs6000-protos.h (output_isel): Delete.
9787 * config/rs6000/rs6000.c (output_isel): Delete.
9788 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
9789 instead of reg_or_cint_operand. Output instruction directly (not via
9791 (isel_unsigned_<mode>): Ditto.
9792 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
9793 gpc_reg_operand. Add an instruction alternative for this. Output
9794 instruction directly.
9795 (*isel_reversed_unsigned_<mode>): Ditto.
9797 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9799 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
9800 (TARGET_CANONICALIZE_COMPARISON): Define.
9802 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
9805 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
9806 Check whether the dest is REG before adding REG_EQUIV note.
9808 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9811 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9813 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9814 (make_hard_regno_born, make_hard_regno_dead): Update
9815 bb_killed_pseudos and bb_gen_pseudos.
9817 2017-10-11 Nathan Sidwell <nathan@acm.org>
9819 * incpath.h (enum incpath_kind): Name enum, prefix values.
9820 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
9821 * incpath.c (heads, tails): Use INC_MAX.
9822 (add_env_var_paths, add_standard_paths): Use incpath_kind.
9823 (merge_include_chains, split_quote_chain,
9824 register_include_chains): Update incpath_kind names.
9825 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
9826 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
9828 (add_framework_path, darwin_register_objc_includes): Likewise.
9829 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
9831 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9833 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
9834 Do not use float_operator operator predicate.
9835 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
9836 * config/i386/predicates.md (float_operator): Remove predicate.
9838 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9840 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
9841 (*jccxf_i387): Ditto.
9842 (*jcc<mode>_i387): Ditto.
9843 (*jccu<mode>_i387): Ditto.
9844 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
9845 (*jcc_*_i387 splitters): Remove.
9846 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
9847 * config/i386/i386.c (ix86_split_fp_branch): Remove.
9848 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
9851 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9853 * profile-count.h (slow_safe_scale_64bit): New function.
9854 (safe_scale_64bit): New inline.
9855 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
9856 * profile-count.c: Include wide-int.h
9857 (slow_safe_scale_64bit): New.
9859 2017-10-11 Nathan Sidwell <nathan@acm.org>
9861 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
9862 HAS_DECL_ASSEMBLER_NAME_P.
9863 * gimple-expr.c (gimple_decl_printable_name: Check
9864 HAS_DECL_ASSEMBLER_NAME_P too.
9865 * ipa-utils.h (type_in_anonymous_namespace_p): Check
9866 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
9867 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
9868 * passes.c (rest_of_decl_compilation): Check
9869 HAS_DECL_ASSEMBLER_NAME_P too.
9870 * recog.c (verify_changes): Likewise.
9871 * tree-pretty-print.c (dump_decl_name): Likewise.
9872 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
9874 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
9875 (SET_DECL_ASSEMBLER_NAME): Use it.
9876 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
9877 (COPY_DECL_ASSEMBLER_NAME): Likewise.
9878 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
9880 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9882 * config.gcc (i386, x86_64): Add extra objects.
9883 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
9884 (ix86_min_insn_size): Declare.
9885 (ix86_issue_rate): Declare.
9886 (ix86_adjust_cost): Declare.
9887 (ia32_multipass_dfa_lookahead): Declare.
9888 (ix86_macro_fusion_p): Declare.
9889 (ix86_macro_fusion_pair_p): Declare.
9890 (ix86_bd_has_dispatch): Declare.
9891 (ix86_bd_do_dispatch): Declare.
9892 (ix86_core2i7_init_hooks): Declare.
9893 (ix86_atom_sched_reorder): Declare.
9894 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
9895 (COSTS_N_BYTES): Move to x86-tune-costs.h.
9896 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
9897 (rip_relative_addr_p): Rename to ...
9898 (ix86_rip_relative_addr_p): ... this one; export.
9899 (memory_address_length): Update.
9900 (ix86_issue_rate): Move to x86-tune-sched.c.
9901 (ix86_flags_dependent): Move to x86-tune-sched.c.
9902 (ix86_agi_dependent): Move to x86-tune-sched.c.
9903 (exact_dependency_1): Move to x86-tune-sched.c.
9904 (exact_store_load_dependency): Move to x86-tune-sched.c.
9905 (ix86_adjust_cost): Move to x86-tune-sched.c.
9906 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
9907 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
9908 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
9909 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
9910 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
9911 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
9912 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
9913 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
9914 (min_insn_size): Rename to ...
9915 (ix86_min_insn_size): ... this one; export.
9916 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
9917 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
9918 (core2i7_first_cycle_multipass_backtrack): Move to
9919 x86-tune-sched-core.c.
9920 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
9921 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
9922 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
9923 (ix86_avoid_jump_mispredicts): Update.
9924 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
9925 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
9926 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
9927 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
9928 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
9929 (MAX_INSN): Move to ix86-tune-sched-bd.c.
9930 (MAX_IMM): Move to ix86-tune-sched-bd.c.
9931 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
9932 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
9933 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
9934 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
9935 (MAX_STORE): Move to ix86-tune-sched-bd.c.
9936 (BIG): Move to ix86-tune-sched-bd.c.
9937 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
9938 (enum insn_path): Move to ix86-tune-sched-bd.c.
9939 (get_mem_group): Move to ix86-tune-sched-bd.c.
9940 (is_cmp): Move to ix86-tune-sched-bd.c.
9941 (dispatch_violation): Move to ix86-tune-sched-bd.c.
9942 (is_branch): Move to ix86-tune-sched-bd.c.
9943 (is_prefetch): Move to ix86-tune-sched-bd.c.
9944 (init_window): Move to ix86-tune-sched-bd.c.
9945 (allocate_window): Move to ix86-tune-sched-bd.c.
9946 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
9947 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
9948 (process_end_window): Move to ix86-tune-sched-bd.c.
9949 (allocate_next_window): Move to ix86-tune-sched-bd.c.
9950 (find_constant): Move to ix86-tune-sched-bd.c.
9951 (get_num_immediates): Move to ix86-tune-sched-bd.c.
9952 (has_immediate): Move to ix86-tune-sched-bd.c.
9953 (get_insn_path): Move to ix86-tune-sched-bd.c.
9954 (get_insn_group): Move to ix86-tune-sched-bd.c.
9955 (count_num_restricted): Move to ix86-tune-sched-bd.c.
9956 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
9957 (add_insn_window): Move to ix86-tune-sched-bd.c.
9958 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
9959 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
9960 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
9961 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
9962 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
9963 (do_dispatch): Move to ix86-tune-sched-bd.c.
9964 (has_dispatch): Move to ix86-tune-sched-bd.c.
9965 * config/i386/t-i386: Add new object files.
9966 * config/i386/x86-tune-costs.h: New file.
9967 * config/i386/x86-tune-sched-atom.c: New file.
9968 * config/i386/x86-tune-sched-bd.c: New file.
9969 * config/i386/x86-tune-sched-core.c: New file.
9970 * config/i386/x86-tune-sched.c: New file.
9972 2017-10-11 Liu Hao <lh_mouse@126.com>
9974 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
9975 the generic version below instead.
9976 (should_colorize): Recognize Windows consoles as terminals
9978 * pretty-print.c [__MINGW32__] (write_all): New function.
9979 [__MINGW32__] (find_esc_head): Likewise.
9980 [__MINGW32__] (find_esc_terminator): Likewise.
9981 [__MINGW32__] (eat_esc_sequence): Likewise.
9982 [__MINGW32__] (mingw_ansi_fputs): New function that handles
9984 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
9987 2017-10-11 Richard Biener <rguenther@suse.de>
9989 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
9990 Properly call analyze_scalar_evolution with the loop of the stmt.
9992 2017-10-11 Richard Biener <rguenther@suse.de>
9994 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
9995 * tree-core.h (tree_base): Add chrec_var union member.
9996 * tree.h (CHREC_VAR): Remove.
9997 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
9998 * tree-chrec.h (build_polynomial_chrec): Adjust.
9999 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
10000 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
10002 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
10004 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
10005 * match.pd: ... here.
10006 ((T) X == (T) Y): Relax condition.
10008 2017-10-11 Bin Cheng <bin.cheng@arm.com>
10010 PR tree-optimization/82472
10011 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
10013 (break_alias_scc_partitions): Update postorder number.
10015 2017-10-11 Martin Liska <mliska@suse.cz>
10018 * opts.c (parse_no_sanitize_attribute): Do not use error_value
10020 * opts.h (parse_no_sanitize_attribute): Remove last argument.
10022 2017-10-11 Martin Liska <mliska@suse.cz>
10024 * print-rtl.c (print_insn): Move declaration of idbuf
10025 to same scope as name.
10027 2017-10-11 Martin Liska <mliska@suse.cz>
10032 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
10033 volatile arguments.
10035 2017-10-11 Martin Liska <mliska@suse.cz>
10038 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
10039 volatile arguments.
10041 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
10043 * config.gcc (default_gnu_indirect_function): Default to yes for
10044 arm*-*-linux* with glibc.
10046 2017-10-11 Richard Biener <rguenther@suse.de>
10048 * tree-scalar-evolution.c (get_scalar_evolution): Handle
10049 default-defs and types we do not want to analyze.
10050 (interpret_loop_phi): Replace unreachable code with an assert.
10051 (compute_scalar_evolution_in_loop): Remove and inline ...
10052 (analyze_scalar_evolution_1): ... here, replacing condition with
10053 what makes the intent clearer. Remove handling of cases
10054 get_scalar_evolution now handles.
10056 2017-10-10 Jim Wilson <wilson@tuliptree.org>
10058 PR rtl-optimization/81434
10059 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
10060 comment for main loop. In sched_group_found if, also add checks for
10061 pass and min_cost_group.
10063 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
10065 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
10066 (rs6000_insn_cost): New function.
10067 * config/rs6000/rs6000.md (cost): New attribute.
10069 2017-10-10 Jakub Jelinek <jakub@redhat.com>
10070 H.J. Lu <hongjiu.lu@intel.com>
10074 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
10075 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
10076 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
10077 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
10078 to OPTION_MASK_ISA_AVX512VL - builtins that have both
10079 OPTION_MASK_ISA_MMX and some other bit set require both
10080 mmx and the ISAs without the mmx bit.
10081 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
10082 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
10083 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
10084 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
10085 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
10086 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
10087 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
10088 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
10089 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
10090 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
10091 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
10092 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
10093 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
10094 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
10095 Add OPTION_MASK_ISA_MMX.
10097 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
10099 * config.gcc (armv7*-*-freebsd*): New target.
10100 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
10102 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
10104 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
10105 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
10108 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
10110 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
10111 a template parameter.
10112 (WIDE_INT_REF_FOR): Update accordingly.
10113 * tree.h (wi::int_traits <const_tree>): Delete.
10114 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
10115 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
10116 (wi::tree_to_wide_ref): New typedef.
10117 (wi::to_wide): New function.
10118 * calls.c (get_size_range): Use wi::to_wide when operating on
10119 trees as wide_ints.
10120 * cgraph.c (cgraph_node::create_thunk): Likewise.
10121 * config/i386/i386.c (ix86_data_alignment): Likewise.
10122 (ix86_local_alignment): Likewise.
10123 * dbxout.c (stabstr_O): Likewise.
10124 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
10125 * expr.c (const_vector_from_tree): Likewise.
10126 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
10127 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
10128 (fold_negate_expr_1, int_const_binop_1, const_binop)
10129 (fold_convert_const_int_from_real, optimize_bit_field_compare)
10130 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
10131 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
10132 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
10133 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
10134 (fold_not_const, round_up_loc): Likewise.
10135 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
10136 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
10137 (alloca_call_type): Likewise.
10138 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
10139 * godump.c (go_output_typedef): Likewise.
10140 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
10141 * internal-fn.c (get_min_precision): Likewise.
10142 * ipa-cp.c (ipcp_store_vr_results): Likewise.
10143 * ipa-polymorphic-call.c
10144 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
10145 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
10146 (ipa_modify_call_arguments): Likewise.
10147 * match.pd: Likewise.
10148 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
10149 * print-tree.c (print_node_brief, print_node): Likewise.
10150 * stmt.c (expand_case): Likewise.
10151 * stor-layout.c (layout_type): Likewise.
10152 * tree-affine.c (tree_to_aff_combination): Likewise.
10153 * tree-cfg.c (group_case_labels_stmt): Likewise.
10154 * tree-data-ref.c (dr_analyze_indices): Likewise.
10155 (prune_runtime_alias_test_list): Likewise.
10156 * tree-dump.c (dequeue_and_dump): Likewise.
10157 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
10158 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
10159 * tree-pretty-print.c (dump_generic_node): Likewise.
10160 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
10161 (simple_iv_with_niters): Likewise.
10162 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
10163 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
10164 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
10165 * tree-ssa-loop-niter.c (split_to_var_and_offset)
10166 (refine_value_range_using_guard, number_of_iterations_ne_max)
10167 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
10168 (get_cst_init_from_scev, record_nonwrapping_iv)
10169 (scev_var_range_cant_overflow): Likewise.
10170 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
10171 * tree-ssa-pre.c (compute_avail): Likewise.
10172 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
10173 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
10174 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
10175 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
10176 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
10177 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
10178 (dump_case_nodes, try_switch_expansion): Likewise.
10179 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
10180 (vect_do_peeling): Likewise.
10181 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
10182 * tree-vect-stmts.c (vectorizable_load): Likewise.
10183 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
10184 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
10185 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
10186 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
10187 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
10188 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
10189 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
10190 (evrp_dom_walker::before_dom_children): Likewise.
10191 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
10192 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
10193 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
10194 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
10195 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
10196 (get_range_pos_neg): Likewise.
10197 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
10198 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
10199 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
10200 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
10201 * config/avr/avr.c (avr_fold_builtin): Likewise.
10202 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
10203 * config/msp430/msp430.c (msp430_attr): Likewise.
10204 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
10205 * config/powerpcspe/powerpcspe-c.c
10206 (altivec_resolve_overloaded_builtin): Likewise.
10207 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
10208 (rs6000_expand_ternop_builtin): Likewise.
10209 * config/rs6000/rs6000-c.c
10210 (altivec_resolve_overloaded_builtin): Likewise.
10211 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
10212 (rs6000_expand_ternop_builtin): Likewise.
10213 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
10215 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10217 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
10218 when copying loop nest with only one inner loop.
10220 2017-10-10 Richard Biener <rguenther@suse.de>
10222 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
10223 blocks if SCEV is active.
10224 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
10226 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
10227 (scev_initialize): Assert we are not yet initialized.
10229 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10231 * tree-loop-distribution.c (generate_loops_for_partition): Remove
10232 inner loop's exit stmt by making it always exit the loop, otherwise
10233 we would generate an infinite empty loop.
10235 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10237 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
10238 renaming variables in new preheader if it's deleted.
10240 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10242 * tree-loop-distribution.c (struct partition): Remove unused field
10243 loops of the structure.
10244 (partition_alloc, partition_free): Ditto.
10245 (build_rdg_partition_for_vertex): Ditto.
10247 2017-10-09 Jeff Law <law@redhat.com>
10249 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
10250 return type to match prototype and documentation.
10252 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10254 * config/rs6000/rs6000.c (processor_costs): Move to ...
10255 * config/rs6000/rs6000.h: ... here.
10256 (rs6000_cost): Declare.
10258 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
10260 * except.c (setjmp_fn): New global variable.
10261 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
10262 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
10263 if DONT_USE_BUILTIN_SETJMP is defined.
10265 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10267 * target.def (insn_cost): New hook.
10268 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
10269 * doc/tm.texi: Regenerate.
10270 * rtlanal.c (insn_cost): Use the new hook.
10272 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10274 * combine.c (combine_validate_cost): Compute the new insn_cost,
10275 not just pattern_cost.
10276 (try_combine): Adjust comment.
10278 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10280 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
10282 * combine.c (uid_insn_cost): Adjust comment.
10283 (combine_validate_cost): Adjust comment. Use pattern_cost instead
10285 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
10286 * dse.c (find_shift_sequence): Ditto.
10287 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
10288 (bb_valid_for_noce_process_p): Use pattern_cost.
10289 * rtl.h (insn_rtx_cost): Delete.
10290 (pattern_cost): New prototype.
10291 (insn_cost): New prototype.
10292 * rtlanal.c (insn_rtx_cost): Rename to...
10293 (pattern_cost): ... this.
10296 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
10298 * config/i386/i386.md (*jcc_2): Remove insn pattern.
10299 (*jcc<mode>_0_r_i387): Ditto.
10300 (*jccxf_r_i387): Ditto.
10301 (*jcc<mode>_r_i387): Ditto.
10302 (*jccu<mode>_r_i387): Ditto.
10303 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
10304 (*jcc): Rename from *jcc_1.
10306 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10308 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
10309 deferred rescans after the lvx/stvx recombination pre-pass.
10311 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
10313 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
10314 memory operation instruction support.
10315 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
10316 (rs6000-ibm-aix[789]*): Likewise.
10317 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
10318 Document new functions.
10320 2017-10-09 Richard Biener <rguenther@suse.de>
10322 PR tree-optimization/82397
10323 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
10324 equality only for semantically equal trees.
10326 2017-10-09 Richard Biener <rguenther@suse.de>
10328 PR tree-optimization/82449
10329 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
10330 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
10331 allow constant addresses.
10332 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
10335 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10337 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
10340 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10343 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
10346 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10349 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
10351 2017-10-09 Jakub Jelinek <jakub@redhat.com>
10354 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
10355 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
10357 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
10359 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
10360 (WI_BINARY_PREDICATE_RESULT): Likewise.
10361 (wi::binary_traits::operator_result): New type.
10362 (wi::binary_traits::predicate_result): Likewise.
10363 (generic_wide_int::operator~, unary generic_wide_int::operator-)
10364 (generic_wide_int::operator==, generic_wide_int::operator!=)
10365 (generic_wide_int::operator&, generic_wide_int::and_not)
10366 (generic_wide_int::operator|, generic_wide_int::or_not)
10367 (generic_wide_int::operator^, generic_wide_int::operator+
10368 (binary generic_wide_int::operator-, generic_wide_int::operator*):
10370 (operator~, unary operator-, operator==, operator!=, operator&)
10371 (operator|, operator^, operator+, binary operator-, operator*): New
10373 * expr.c (get_inner_reference): Use wi::bit_and_not.
10374 * fold-const.c (fold_binary_loc): Likewise.
10375 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
10376 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
10377 (bit_value_binop): Likewise.
10378 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
10379 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
10380 (extract_range_from_binary_expr_1): Likewise.
10381 (masked_increment): Likewise.
10382 (simplify_bit_ops_using_ranges): Likewise.
10384 2017-10-09 Martin Jambor <mjambor@suse.cz>
10387 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
10388 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
10389 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
10391 (hsa_fixup_mov_insn_type): New function.
10392 (hsa_op_with_type::get_in_type): Use it.
10393 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
10394 immediates in an assert.
10395 (hsa_op_with_type::extend_int_to_32bit): New method.
10396 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
10397 types. Convert to dest type if necessary.
10398 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
10399 (reg_for_gimple_ssa): Pass false as min32int to
10400 hsa_type_for_scalar_tree_type.
10401 (gen_hsa_addr): Fixup type when creating addresable temporary.
10402 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
10403 (gen_hsa_unary_operation): Extend operands and convert to dest type if
10404 necessary. Call hsa_fixup_mov_insn_type.
10405 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
10406 extend operands and convert to dest type if necessary.
10407 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
10408 to dest type if necessary.
10409 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
10410 if conversion nt necessary and size matches.
10411 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
10412 to dest type if necessary.
10413 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
10414 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
10416 (gen_hsa_clrsb): Likewise.
10417 (gen_hsa_ffs): Likewise.
10418 (gen_hsa_divmod): Extend operands and convert to dest type if
10420 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
10422 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
10424 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
10425 Remove empty default arguments. Use a brace block as output
10427 (conditional return): Ditto.
10429 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
10430 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
10432 (group_ending_nop): Ditto.
10433 (doloop_end): Ditto.
10434 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
10435 (splitters for those): Ditto.
10437 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
10439 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
10440 a conditional jump (and the compare for it) so that pc_rtx is the
10442 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
10443 for the deleted and renamed ctr<mode>_internal[234] patterns.
10444 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
10445 Delete second conditional return pattern.
10446 (ctr<mode>_internal2): Delete this second bdnz pattern.
10447 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
10448 (ctr<mode>_internal4): Delete this second bdz pattern.
10450 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
10452 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
10453 (always_initialized_rtx_for_ssa_name_p): New predicate.
10454 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
10455 (finish_out_of_ssa): Free new field of SA.
10456 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
10457 * tree-ssa-coalesce.c: Include tree-ssa.h.
10458 (get_parm_default_def_partitions): Remove extern keyword.
10459 (get_undefined_value_partitions): New function.
10460 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
10461 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
10462 uninitialized bits.
10463 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
10465 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
10467 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
10469 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
10471 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
10472 for targets that preffer 128bit.
10474 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
10476 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
10478 2017-10-08 Olivier Hainque <hainque@adacore.com>
10480 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
10481 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
10483 (thumb_set_return_address): Likewise.
10485 2017-10-08 Olivier Hainque <hainque@adacore.com>
10487 * common/config/arm/arm-common.c (arm_except_unwind_info):
10488 Handle DWARF2_UNWIND_INFO.
10490 2017-10-07 Michael Collison <michael.collison@arm.com>
10492 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
10495 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
10497 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
10498 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
10499 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
10500 defined, force the creation of a new block for a dispatch label.
10502 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
10504 * invoke.texi (Wsuggest-attribute=cold): Document.
10505 * common.opt (Wsuggest-attribute=cold): New
10506 * ipa-pure-const.c (warn_function_cold): New function.
10507 * predict.c (compute_function_frequency): Use it.
10508 * predict.h (warn_function_cold): Declare.
10510 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
10512 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
10515 2017-10-06 Martin Liska <mliska@suse.cz>
10517 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
10518 keyword for member functions.
10519 (struct sanopt_tree_couple): New struct.
10520 (struct sanopt_tree_couple_hash): New function.
10521 (struct sanopt_ctx): Add new hash_map.
10522 (has_dominating_ubsan_ptr_check): New function.
10523 (record_ubsan_ptr_check_stmt): Likewise.
10524 (maybe_optimize_ubsan_ptr_ifn): Likewise.
10525 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
10526 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
10528 2017-10-06 Sudakshina Das <sudi.das@arm.com>
10531 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
10532 aarch64_simd_valid_immediate on CONST_VECTORs.
10533 (aarch64_reg_or_bic_imm): Likewise.
10535 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
10537 PR rtl-optimization/82396
10538 * haifa-sched.c (ready_sort_real): Disable qsort checking.
10540 2017-10-06 Sebastian Pop <sebpop@gmail.com>
10542 * graphite-dependences.c (scop_get_reads): Move code to...
10543 (scop_get_must_writes): Move code to...
10544 (scop_get_may_writes): Move code to...
10545 (scop_get_reads_and_writes): ... here.
10546 (scop_get_dependences): Call scop_get_reads_and_writes.
10548 2017-10-06 Jakub Jelinek <jakub@redhat.com>
10550 PR tree-optimization/82434
10551 * fold-const.h (can_native_encode_type_p,
10552 can_native_encode_string_p): Remove.
10553 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
10554 don't encode anything, just return what would be otherwise returned.
10555 (native_encode_fixed, native_encode_complex, native_encode_vector):
10557 (native_encode_string): Likewise. Inline by hand
10558 can_native_encode_string_p.
10559 (can_native_encode_type_p): Remove.
10560 (can_native_encode_string_p): Remove.
10561 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
10562 STRING_CSTs using can_native_encode_string_p, test all
10563 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
10564 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
10565 argument from native_encode_expr.
10566 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
10567 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
10570 2017-10-06 Richard Biener <rguenther@suse.de>
10572 PR tree-optimization/82397
10573 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
10574 operand_equal_p but rely on data_ref_compare_tree for detecting
10576 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
10577 to match up with dr_group_sort_cmp.
10579 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10582 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
10584 * config/s390/s390-builtin-types.def: Regenerate.
10586 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10589 * config/s390/s390-builtin-types.def: Regenerate.
10590 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
10591 Change flag from B_VXE to B_VX.
10592 (s390_vec_min_dbl): Remove B_VXE flag.
10594 2017-10-06 Richard Biener <rguenther@suse.de>
10596 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
10597 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
10598 translate_isl_ast_to_gimple::is_valid_rename,
10599 translate_isl_ast_to_gimple::get_rename,
10600 translate_isl_ast_to_gimple::get_def_bb_for_const,
10601 translate_isl_ast_to_gimple::get_new_name,
10602 translate_isl_ast_to_gimple::collect_all_ssa_names,
10603 translate_isl_ast_to_gimple::copy_loop_phi_args,
10604 translate_isl_ast_to_gimple::collect_all_ssa_names,
10605 translate_isl_ast_to_gimple::copy_loop_phi_args,
10606 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
10607 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
10608 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
10609 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
10610 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
10611 translate_isl_ast_to_gimple::copy_cond_phi_args,
10612 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
10613 translate_isl_ast_to_gimple::edge_for_new_close_phis,
10614 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
10615 translate_isl_ast_to_gimple::rename_uses,
10616 translate_isl_ast_to_gimple::rename_all_uses): Remove.
10617 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
10618 (set_rename_for_each_def): Likewise.
10619 (graphite_copy_stmts_from_block): Handle debug stmt resetting
10620 here. Handle rewriting SCEV analyzable uses here.
10621 (copy_bb_and_scalar_dependences): Generate code for PHI
10623 (graphite_regenerate_ast_isl): Adjust.
10624 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
10625 (add_write, add_read): New functions.
10626 (build_cross_bb_scalars_def): Use it and simplify.
10627 (build_cross_bb_scalars_use): Likewise.
10628 (graphite_find_cross_bb_scalar_vars): Inline into...
10629 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
10630 simulating out-of-SSA. Compute liveout and add dependencies.
10631 (build_scops): Force an empty entry block.
10632 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
10634 (sese_build_liveouts): Declare.
10635 (sese_trivially_empty_bb_p): Likewise.
10636 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
10637 compute liveout and debug_liveout.
10638 (sese_bad_liveouts_use): Remove.
10639 (sese_reset_debug_liveouts_bb): Likewise.
10640 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
10641 (sese_build_liveouts): Build liveout and debug_liveout and store
10643 (new_sese_info): Adjust.
10644 (free_sese_info): Likewise.
10645 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
10646 do not build liveout here.
10647 (move_sese_in_condition): Adjust region entry.
10648 (scev_analyzable_p): Match up with chrec_apply requirements.
10649 (sese_trivially_empty_bb_p): New.
10650 * tree-into-ssa.c (get_reaching_def): Properly support generating
10651 default-defs for incremental rewrite of anonymous names.
10653 2017-10-06 Richard Biener <rguenther@suse.de>
10655 * graphite-sese-to-poly.c (extract_affine): For casts increasing
10656 precision do not perform modulo reduction.
10658 2017-10-06 Richard Biener <rguenther@suse.de>
10660 PR tree-optimization/82436
10661 * tree-vect-slp.c (vect_supported_load_permutation_p): More
10662 conservatively choose the vectorization factor when checking
10663 whether we can perform the required load permutation.
10664 (vect_transform_slp_perm_load): Assert when we may not fail.
10666 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
10668 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
10669 message for incompatible -msdata=* and -mcall-* options.
10671 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10673 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
10674 rate for post-reload scheduling.
10676 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10678 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
10680 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10682 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
10683 to improve monte carlo in scimark.
10685 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10687 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
10688 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
10689 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
10690 pentium4_cost, nocona_cost): Set reassociation width to 1.
10691 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
10692 width to 2 for fp operations and 1 otherwise.
10693 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
10695 (atom_cost): Set reassociation width to 2.
10696 (slm_cost, generic_cost): Set fp reassociation width
10697 to 2 and 1 otherwise.
10698 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
10699 (core_cost): Set fp reassociation width to 4 and vector to 2.
10700 (ix86_reassociation_width): Rewrite using cost table; special case
10701 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
10702 and TARGET_AVX128_OPTIMAL.
10703 * config/i386/i386.h (processor_costs): Add
10704 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
10705 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
10706 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
10707 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
10708 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
10709 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
10711 2017-10-05 Nathan Sidwell <nathan@acm.org>
10713 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
10715 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10717 * config/arm/arm.c (arm_test_fpu_data): New.
10718 (arm_run_selftests): Call arm_test_fpu_data.
10720 2017-10-04 Nathan Sidwell <nathan@acm.org>
10722 * toplev.c (toplev::main): Remove excess parens on pretty_printer
10724 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
10726 2017-10-04 Sudakshina Das <sudi.das@arm.com>
10728 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
10729 check type for aarch64_simd_valid_immediate.
10730 (aarch64_output_simd_mov_immediate): Update prototype.
10731 (aarch64_simd_valid_immediate): Update prototype.
10732 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
10733 support for ORR-immediate.
10734 (and<mode>3): modified pattern to add support for BIC-immediate.
10735 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
10736 now checks for valid immediate for BIC and ORR based on new enum
10738 (aarch64_output_simd_mov_immediate): Function now used to output
10739 BIC/ORR imm as well based on new enum argument.
10740 * config/aarch64/constraints.md (Do): New vector immediate constraint.
10742 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
10743 (aarch64_reg_or_bic_imm): Likewise.
10745 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10747 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
10748 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
10750 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10754 PR rtl-optimization/82396
10755 * haifa-sched.c (autopref_multipass_init): Simplify
10757 (autopref_rank_data): Simplify sort order.
10758 * sched-int.h (autopref_multipass_data_): Remove
10759 multi_mem_insn_p, min_offset and max_offset.
10761 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10763 * doc/sourcebuild.texi: Document vect_peeling_profitable.
10765 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10767 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
10768 vect_doubleint_cvt.
10770 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10772 * doc/sourcebuild.texi: Document vect_long_mult.
10774 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
10776 PR tree-optimization/82413
10777 * fold-const.c (build_range_check): Use widest_int when comparing
10778 the maximum ETYPE value with HIGH.
10780 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10782 PR rtl-optimization/82396
10783 * haifa-sched.c (autopref_multipass_init): Simplify
10785 (autopref_rank_data): Simplify sort order.
10786 * sched-int.h (autopref_multipass_data_): Remove
10787 multi_mem_insn_p, min_offset and max_offset.
10789 2017-10-04 Jakub Jelinek <jakub@redhat.com>
10791 PR tree-optimization/82381
10792 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
10793 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
10796 PR tree-optimization/82374
10797 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
10798 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
10799 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
10800 current_function_decl to the new decl.
10802 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
10804 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
10805 helper macro for IEEE float128 hardware built-in functions.
10806 (SQRTF128_ODD): Add built-in functions with the round-to-odd
10808 (TRUNCF128_ODD): Likewise.
10809 (ADDF128_ODD): Likewise.
10810 (SUBF128_ODD): Likewise.
10811 (MULF128_ODD): Likewise.
10812 (DIVF128_ODD): Likewise.
10813 (FMAF128_ODD): Likewise.
10814 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
10815 UNSPEC_TRUNC_ROUND_TO_ODD.
10816 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
10817 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
10818 floating point round to odd instructions.
10819 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
10820 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
10821 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
10822 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
10823 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
10824 (trunc<mode>sf2_hw): Change the truncate with round to odd
10825 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
10826 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
10827 to odd hardware instructions.
10828 (sub<mode>3_odd): Likewise.
10829 (mul<mode>3_odd): Likewise.
10830 (div<mode>3_odd): Likewise.
10831 (sqrt<mode>2_odd): Likewise.
10832 (fma<mode>4_odd): Likewise.
10833 (fms<mode>4_odd): Likewise.
10834 (nfma<mode>4_odd): Likewise.
10835 (nfms<mode>4_odd): Likewise.
10836 (trunc<mode>df2_odd): Change the truncate with round to odd
10837 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
10839 * doc/extend.texi (PowerPC built-in functions): Update documentation
10840 for existing IEEE float128-bit built-in functions. Add built-in
10841 functions that generate the IEEE 128-bit floating point round to
10844 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
10846 PR rtl-optimization/77729
10847 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
10848 to (X&(C1&~C2))|C2 transformations.
10850 2017-10-03 Martin Jambor <mjambor@suse.cz>
10852 PR tree-optimization/82363
10853 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
10854 mismatch, mark lacc written regardless of racc.
10856 2017-10-03 Jakub Jelinek <jakub@redhat.com>
10858 PR tree-optimization/82381
10859 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
10860 stmt_to_insert nor wheather SSA_NAMEs are default defs.
10861 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
10862 fallthrough into reassoc_stmt_dominates_stmt_p.
10865 * combine.c (combine_instructions): Don't combine in unreachable
10868 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
10871 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
10872 function to not use the have_cpu variable. Do not set cpu_index,
10873 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
10874 or the default cpu.
10875 (rs6000_valid_attribute_p): Remove duplicate initializations of
10876 old_optimize and func_optimize.
10877 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
10878 (rs6000_activate_target_options): Make global.
10879 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
10882 2017-10-02 Jakub Jelinek <jakub@redhat.com>
10884 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
10885 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
10886 Set *poffset to 0 and *psize and *pmax_size to -1 if
10887 *poffset + *psize overflows in HOST_WIDE_INT.
10889 PR tree-optimization/82387
10890 PR tree-optimization/82388
10891 PR tree-optimization/82389
10892 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
10893 instead of live_bytes non-NULL.
10895 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
10898 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
10901 2017-10-02 Richard Biener <rguenther@suse.de>
10903 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
10904 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
10905 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
10907 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10909 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
10910 requested precision matches the type's.
10911 * calls.c (alloc_max_size): Calculate the new candidate size as
10912 a widest_int and use wi::to_widest when comparing it with the
10913 current candidate size.
10914 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
10915 zero rather than integer_zero_node.
10916 * match.pd: Check for a no-op conversion before using wi::add
10917 rather than after. Use tree_to_uhwi when summing small shift
10918 counts into an unsigned int.
10920 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10921 Alan Hayward <alan.hayward@arm.com>
10922 David Sherwood <david.sherwood@arm.com>
10925 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
10926 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
10927 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
10928 POINTER_AND_FP_REGS.
10930 2017-10-02 Richard Biener <rguenther@suse.de>
10932 PR tree-optimization/82355
10933 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
10934 a mapping for the enclosing loop but avoid generating one for
10935 the loop tree root.
10936 (copy_bb_and_scalar_dependences): Remove premature codegen
10937 error on PHIs in blocks duplicated into multiple places.
10938 * graphite-scop-detection.c
10939 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
10940 in the region use it as loop and nest to analyze the DR in.
10941 (try_generate_gimple_bb): Likewise.
10942 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
10943 (add_loop_constraints): For blocks in a loop not in the region
10944 create a dimension with a single iteration.
10945 * sese.h (gbb_loop_at_index): Remove assert.
10947 2017-10-01 Kevin Buettner <kevinb@redhat.com>
10949 * omp-expand.c (adjust_context_scope): New function.
10950 (expand_parallel_call): Call adjust_context_scope.
10952 2017-10-01 Jeff Law <law@redhat.com>
10954 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
10955 dom_opt_dom_walker class with direct access to private members.
10956 Add comments. Call test_for_singularity.
10957 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
10958 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
10959 m_dummy_cond anymore.
10960 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
10962 (pass_dominator:execute): Build the dummy_cond here and pass it
10963 to the dom_opt_dom_walker ctor.
10964 (test_for_singularity): New function.
10966 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
10967 Maya Rashish <coypu@sdf.org>
10969 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
10970 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
10971 (alpha*-*-netbsd*) Use nbsd_tm_file.
10972 (arm*-*-netbsdelf*) Likewise.
10973 (i[34567]86-*-netbsdelf*) Likewise.
10974 (x86_64-*-netbsd*) Likewise.
10975 (mips*-*-netbsd*) Likewise.
10976 (powerpc-*-netbsd*) Likewise.
10977 (sh*-*-netbsd*) Likewise.
10978 (sparc-*-netbsdelf*) Likewise.
10979 (sparc64-*-netbsd*) Likewise.
10980 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
10982 (vax-*-netbsdelf*) Likewise.
10983 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
10984 (UINT_FAST8_TYPE) Likewise.
10985 (INT_FAST16_TYPE) Check CHAR_FAST16.
10986 (UINT_FAST16_TYPE) Likewise.
10988 2017-09-30 Jakub Jelinek <jakub@redhat.com>
10991 * config/i386/i386.md
10992 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
10993 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
10994 *divmodsi4_zext_2): New define_insn_and_split.
10995 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
10996 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
10997 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
10998 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
10999 New define_insn_and_split.
11000 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
11001 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
11002 operands[1] having DImode when mode is SImode.
11004 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
11005 always SImode for DIV and MOD in REG_EQUAL notes.
11007 2017-09-29 Yury Gribov <tetra2005@gmail.com>
11009 PR middle-end/82319
11010 * match.pd: Fix handling of NaNs in pattern.
11012 2017-09-29 Jeff Law <law@redhat.com>
11014 * sbitmap.c (bitmap_bit_in_range_p): New function.
11015 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
11016 * tree-ssa-dse.c (live_bytes_read): New function.
11017 (dse_classify_store): Ignore reads of dead bytes.
11019 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
11020 typos and whitespace errors.
11021 * config/i386/predicates.md (address_no_seg_operand): Likewise.
11022 * config/s390/s390.c (s390_emit_prologue): Likewise.
11024 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
11027 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
11028 with a symbol for LRA.
11030 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
11032 PR rtl-optimization/82338
11033 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
11035 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
11037 * genmodes.c (calc_wider_mode): Suppress qsort macro.
11038 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
11039 (qsort_chk): Declare.
11040 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
11041 (qsort_chk): New function.
11043 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11045 PR tree-optimization/82337
11046 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
11047 phi definition if the PHI result appears in an abnormal PHI.
11048 (find_basis_for_base_expr): Don't record a basis if the LHS of the
11049 basis appears in an abnormal PHI.
11051 2017-09-29 Richard Biener <rguenther@suse.de>
11053 * graphite-isl-ast-to-gimple.c
11054 (translate_isl_ast_to_gimple::set_codegen_error): New function.
11055 (binary_op_to_tree): Use it.
11056 (get_rename_from_scev): Likewise.
11057 (copy_loop_phi_nodes): Likewise.
11058 (copy_bb_and_scalar_dependences): Likewise.
11059 (translate_pending_phi_nodes): Likewise.
11061 2017-09-29 Jakub Jelinek <jakub@redhat.com>
11064 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
11065 for movabsq $(i32 << shift), r64.
11067 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
11069 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
11070 index when encoding %esp as %rsp to avoid 0x67 prefix.
11072 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
11074 * config/i386/i386.md (*movsf_internal, *movdf_internal):
11075 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
11077 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
11079 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
11080 Extensions with more than 16 double VFP registers.
11081 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
11082 to_clear_mask and all code related to it. Replace the remaining
11083 entry by a sbitmap and adapt code accordingly.
11085 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
11087 * brig-builtins.def: Change pure attributes to const.
11089 2017-09-28 Joseph Myers <joseph@codesourcery.com>
11091 * config.gcc (default_gnu_indirect_function): Default to yes for
11092 sparc*-*-linux* with glibc.
11094 2017-09-28 Joseph Myers <joseph@codesourcery.com>
11096 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
11097 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
11098 when creating .init_array and .fini_array sections with priority
11101 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
11104 * config/aarch64/aarch64.c
11105 (aarch64_builtin_support_vector_misalignment): Always return false
11106 when misalignment is unknown.
11108 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
11110 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
11111 this function to return false if the definition used by the swap
11112 instruction is artificial, or if the memory address from which the
11113 constant value is loaded is not represented by a base address held
11114 in a register or if the base address register is a frame or stack
11115 pointer. Additionally, return false if the base address of the
11116 loaded constant is a SYMBOL_REF but is not considered to be a
11118 (replace_swapped_load_constant): New function.
11119 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
11120 loaded constant vector with a load of a swapped constant vector.
11122 2017-09-27 Carl Love <cel@us.ibm.com>
11124 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
11125 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
11126 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
11129 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
11131 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
11132 first, always call autopref_rank_data otherwise.
11134 2017-09-27 Richard Biener <rguenther@suse.de>
11136 * graphite-scop-detection.c (find_scop_parameters): Move
11137 loop bound handling ...
11138 (gather_bbs::before_dom_children): ... here, avoiding the need
11139 to build scop_info->loop_nest.
11140 (record_loop_in_sese): Remove.
11141 * sese.h (sese_info_t::loop_nest): Remove.
11142 * sese.c (new_sese_info): Do not allocate loop_nest.
11143 (free_sese_info): Do not free loop_nest.
11145 2017-09-27 Jakub Jelinek <jakub@redhat.com>
11148 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
11149 lhs from calls if the lhs has addressable type.
11151 2017-09-27 Richard Biener <rguenther@suse.de>
11153 * graphite.h (scop::max_alias_set): New member.
11154 * graphite-scop-detection.c: Remove references to non-existing
11155 --param in comments.
11156 (build_alias_sets): Record the maximum alias set used for drs.
11157 (build_scops): Support zero as unlimited for
11158 --param graphite-max-arrays-per-scop.
11159 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
11160 and inline into ...
11161 (build_poly_sr_1): ... here. Compute alias set based on the
11162 maximum alias set used for drs rather than
11163 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
11165 2017-09-27 Richard Biener <rguenther@suse.de>
11167 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
11168 --param loop-block-tile-size=0 to disable tiling.
11170 2017-09-27 Richard Biener <rguenther@suse.de>
11172 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
11173 (graphite-max-nb-scop-params): Document special value zero.
11174 * domwalk.h (dom_walker::STOP): New symbolical constant.
11175 (dom_walker::dom_walker): Add optional parameter for bb to
11177 (dom_walker::~dom_walker): Declare.
11178 (dom_walker::before_dom_children): Document STOP return value.
11179 (dom_walker::m_user_bb_to_rpo): New member.
11180 (dom_walker::m_bb_to_rpo): Likewise.
11181 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
11182 mapping here if not provided by the user.
11183 (dom_walker::~dom_walker): Free bb to RPO mapping if not
11184 provided by the user.
11185 (dom_walker::STOP): Define.
11186 (dom_walker::walk): Do not compute bb to RPO mapping here.
11187 Support STOP return value from before_dom_children to stop
11189 * graphite-optimize-isl.c (optimize_isl): If the schedule
11190 is the same still generate code if -fgraphite-identity
11191 or -floop-parallelize-all are given.
11192 * graphite-scop-detection.c: Include cfganal.h.
11193 (gather_bbs::gather_bbs): Get and pass through bb to RPO
11195 (gather_bbs::before_dom_children): Return STOP for BBs
11197 (build_scops): Compute bb to RPO mapping and pass it to
11198 the domwalk. Treat --param graphite-max-nb-scop-params=0
11199 as not limiting the number of params.
11200 * graphite.c (graphite_initialize): Remove limit on the
11201 number of basic-blocks in a function.
11202 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
11203 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
11204 default value of 10.
11206 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11208 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
11209 Adjust code to eliminate needing to do the shift right 32-bits
11210 operation after XSCVDPSPN.
11212 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11214 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
11215 ((X / Y) != 0 -> X >= Y): Likewise.
11217 2017-09-26 Carl Love <cel@us.ibm.com>
11219 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
11220 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
11221 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
11222 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
11223 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
11224 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
11225 definitions and overloading.
11226 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
11227 statement for P9V_BUILTIN_XST_LEN_R.
11228 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
11229 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
11230 define_expand and define_insn for the instructions and builtins.
11231 * doc/extend.texi: Update the built-in documentation file for the new
11232 built-in functions.
11233 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
11234 define_insn for the instructions
11236 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
11239 * gcc/config/netbsd-protos.h: New file.
11240 * gcc/config/netbsd.c: New file.
11241 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
11242 * gcc/config/t-netbsd: New file.
11243 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
11244 (tmake_file) Add t-netbsd.
11245 (extra_objs) Add netbsd.o.
11247 2017-09-26 Janus Weil <janus@gcc.gnu.org>
11251 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
11253 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11255 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
11256 sign extension from a vector register to a GPR by doing a 32-bit
11257 direct move and then an EXTSW.
11258 (extendsi<mode>2 splitter): Likewise.
11259 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
11260 right or vector extract after doing XSCVDPSPN. Use
11261 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
11263 (movdi_from_sf_zero_ext): Likewise.
11264 (reload_gpr_from_vsxsf): Likewise.
11265 (p8_mfvsrd_4_disf): Delete, no longer used.
11266 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
11267 then needing to move the SFmode to a GPR to use the XSCVDPSP
11268 instruction instead of FRSP and XSCVDPSPN.
11269 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
11270 it is adjacent to the other XSCVSPDP insns.
11271 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
11272 SFmode to be in traditional Altivec registers.
11273 (vsx_xscvdpspn): Eliminate useless alternative constraint.
11274 (vsx_xscvspdpn): Likewise.
11275 (vsx_xscvspdpn_scalar): Likewise.
11277 2017-09-26 Martin Jambor <mjambor@suse.cz>
11279 * tree-sra.c (compare_access_positions): Put integral types first,
11280 stabilize sorting of integral types, remove conditions putting
11281 non-full-precision integers last.
11282 (sort_and_splice_var_accesses): Disable scalarization if a
11283 non-integert would be represented by a non-full-precision integer.
11285 2017-09-26 Joseph Myers <joseph@codesourcery.com>
11287 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
11288 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
11289 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
11290 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
11291 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
11292 conditionals inside the function instead of around it. Call
11293 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
11294 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
11296 2017-09-26 Richard Biener <rguenther@suse.de>
11298 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
11300 (scop_detection::build_scop_breadth): ... this. Removed.
11301 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
11302 (scop_detection::harmful_stmt_in_bb): Likewise.
11303 (scop_detection::graphite_can_represent_stmt): Likewise.
11304 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
11305 (scop_detection::can_represent_loop): Remove recursion, fold in ...
11306 (scop_detection::can_represent_loop_1): ... this. Removed.
11307 (scop_detection::harmful_loop_in_region): Simplify after inlining
11308 the above and remove more quadraticness.
11309 (build_scops): Adjust.
11310 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
11313 2017-09-26 Jakub Jelinek <jakub@redhat.com>
11316 * config/i386/i386.c (ix86_print_operand_address_as): Only test
11317 REGNO (base) == SP_REG if base is a REG.
11319 PR middle-end/35691
11320 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
11321 if it is different SSA_NAME.
11322 (optimize_range_tests_cmp_bitwise): New function.
11323 (optimize_range_tests): Call it.
11325 2017-09-26 Richard Biener <rguenther@suse.de>
11327 PR tree-optimization/82321
11328 * graphite.c (canonicalize_loop_closed_ssa): Properly check
11329 for the def being inside the loop.
11331 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11333 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
11335 * config/s390/s390-builtins.def: Fix constraint on op4.
11337 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11339 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
11340 independent expanders.
11341 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
11342 ("vec_ordered", "vec_unordered"): New expanders.
11344 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11346 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
11349 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11351 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
11352 vec_unpacks_lo_v16qi.
11353 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
11355 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11357 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
11358 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
11359 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
11361 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11363 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
11365 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
11367 ("*vec_slb<mode>"): New insn pattern.
11368 ("vec_shr_<mode>"): New expander.
11369 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
11370 and force the shift count operand to V16QImode.
11371 ("vec_srb<mode>"): Set shift count mode to V16QI.
11373 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11375 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
11376 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
11377 ("vec_widen_smult_hi_<mode>"): New expander definitions.
11379 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
11382 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
11384 2017-09-26 Richard Biener <rguenther@suse.de>
11386 PR tree-optimization/82320
11387 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
11390 2017-09-25 Jeff Law <law@redhat.com>
11392 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
11393 prototype for new argument.
11394 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
11395 mostly extracted from rs6000_emit_allocate_stack.
11396 (rs6000_emit_probe_stack_range_stack_clash): New function.
11397 (rs6000_emit_allocate_stack): Call
11398 rs6000_emit_probe_stack_range_stack_clash as needed.
11399 (rs6000_emit_probe_stack_range): Add additional argument
11400 to call to gen_probe_stack_range{si,di}.
11401 (output_probe_stack_range): New.
11402 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
11403 (output_probe_stack_range_stack_clash): New.
11404 (rs6000_emit_prologue): Emit notes into dump file as requested.
11405 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
11406 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
11407 Add additional operand and pass it to output_probe_stack_range.
11409 2017-09-25 Bin Cheng <bin.cheng@arm.com>
11411 PR tree-optimization/82163
11412 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
11413 (checking_verify_loop_closed_ssa): New parameter.
11414 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
11415 (check_loop_closed_ssa_stmt): Delete.
11416 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
11417 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
11418 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
11421 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
11423 * brig-builtins.def: Treat HSAIL barrier builtins as
11424 setjmp/longjump style functions.
11426 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
11428 * target.def (constant_alignment): New hook.
11429 * defaults.h (CONSTANT_ALIGNMENT): Delete.
11430 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
11431 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
11432 * doc/tm.texi: Regenerate.
11433 * targhooks.h (default_constant_alignment): Declare.
11434 (constant_alignment_word_strings): Likewise.
11435 * targhooks.c (default_constant_alignment): New function.
11436 (constant_alignment_word_strings): Likewise.
11437 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
11438 instead of CONSTANT_ALIGNMENT.
11439 * varasm.c (align_variable, get_variable_align, build_constant_desc)
11440 (force_const_mem): Likewise.
11441 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
11442 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
11443 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
11444 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11445 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
11447 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
11448 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11449 constant_alignment_word_strings.
11450 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
11451 (CONSTANT_ALIGNMENT): Likewise.
11452 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11453 (arm_constant_alignment): New function.
11454 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
11455 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11456 constant_alignment_word_strings.
11457 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
11458 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11459 constant_alignment_word_strings.
11460 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
11461 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11462 (cris_constant_alignment): New function.
11463 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
11464 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11465 (epiphany_constant_alignment): New function.
11466 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
11467 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11468 constant_alignment_word_strings.
11469 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
11470 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11471 constant_alignment_word_strings.
11472 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
11473 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11474 constant_alignment_word_strings.
11475 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
11476 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
11477 * config/i386/i386.c (ix86_constant_alignment): Make static.
11478 Use the same interface as the target hook.
11479 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11480 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
11481 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11482 constant_alignment_word_strings.
11483 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
11484 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
11485 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11486 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
11487 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11488 constant_alignment_word_strings.
11489 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
11490 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11491 constant_alignment_word_strings.
11492 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
11493 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11494 constant_alignment_word_strings.
11495 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
11496 * config/microblaze/microblaze.c (microblaze_constant_alignment):
11498 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11499 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
11500 * config/mips/mips.c (mips_constant_alignment): New function.
11501 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11502 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
11503 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
11504 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11505 (mmix_constant_alignment): Make static. Use the same interface
11506 as the target hook.
11507 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
11508 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11509 constant_alignment_word_strings.
11510 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
11511 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11512 constant_alignment_word_strings.
11513 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
11514 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11515 constant_alignment_word_strings.
11516 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
11517 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11518 (rs6000_constant_alignment): New function.
11519 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
11520 * config/riscv/riscv.c (riscv_constant_alignment): New function.
11521 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11522 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
11523 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11524 (rs6000_constant_alignment): New function.
11525 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
11526 * config/s390/s390.c (s390_constant_alignment): New function.
11527 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11528 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
11529 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11530 constant_alignment_word_strings.
11531 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
11532 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11533 (sparc_constant_alignment): New function.
11534 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
11535 * config/spu/spu.c (spu_constant_alignment): New function.
11536 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11537 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
11538 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11539 constant_alignment_word_strings.
11540 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
11541 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11542 constant_alignment_word_strings.
11543 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
11544 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11545 constant_alignment_word_strings.
11546 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
11547 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11548 (visium_constant_alignment): New function.
11549 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
11550 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11551 (xtensa_constant_alignment): New function.
11552 * system.h (CONSTANT_ALIGNMENT): Poison.
11554 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
11556 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11557 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
11558 (rs6000_builtin_valid_without_lhs): New helper function.
11559 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11560 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
11562 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
11564 * target.h (vec_perm_indices): Use unsigned short rather than
11566 (auto_vec_perm_indices): Likewise.
11567 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
11568 Use unsigned int rather than unsigned char.
11569 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
11571 2017-09-25 Richard Biener <rguenther@suse.de>
11573 * cfgloop.h (sort_sibling_loops): Declare.
11574 * cfgloop.c (sort_sibling_loops_cmp): New helper.
11575 (sort_sibling_loops): New function sorting the sibling loop list
11577 * graphite.c (graphite_transform_loops): Sort sibling loops.
11579 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
11581 * target.def (vec_perm_const_ok): Change sel parameter to
11583 * optabs-query.c (can_vec_perm_p): Update accordingly.
11584 * doc/tm.texi: Regenerate.
11585 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
11586 auto_vec_perm_indices and remove separate nelt field.
11587 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
11588 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
11589 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
11590 (aarch64_expand_vec_perm_const): Update accordingly.
11591 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
11592 to vec_perm_indices.
11593 * config/arm/arm.c (expand_vec_perm_d): Change perm to
11594 auto_vec_perm_indices and remove separate nelt field.
11595 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
11596 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
11597 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
11599 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
11600 to vec_perm_indices.
11601 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
11602 sel to vec_perm_indices.
11603 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
11604 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
11605 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
11607 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
11610 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
11613 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
11614 on the FUNCTION_DECL function context if it has a DIE that is a
11617 2017-09-25 Richard Biener <rguenther@suse.de>
11619 PR tree-optimization/82285
11620 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
11623 2017-09-25 Tom de Vries <tom@codesourcery.com>
11627 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
11630 2017-09-25 Richard Biener <rguenther@suse.de>
11632 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
11633 ISL errors other than isl_error_quota happen. Dump if the
11634 schedule is the same.
11635 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
11636 errors instead of aborting inside ISL.
11638 2017-09-25 Iain Sandoe <iain@codesourcery.com>
11641 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
11642 of libgcc_eh for m64.
11643 * config/i386/darwin64.h: Likewise.
11645 2017-09-25 Richard Biener <rguenther@suse.de>
11647 PR middle-end/82144
11648 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
11649 attribute for incomplete types nor twice for complete ones.
11651 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
11654 * config/i386/i386.c (ix86_print_operand_address_as): Encode
11655 %esp as %rsp to avoid 0x67 prefix if there is no index or base
11658 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
11661 * config/i386/i386.opt (mprefer-avx256): Use
11662 ix86_target_flags variable.
11663 * config/i386/i386.c (ix86_target_string): Move
11664 -mprefer-avx256 to flag2_opts.
11666 2017-09-22 Jakub Jelinek <jakub@redhat.com>
11668 PR middle-end/35691
11669 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
11670 and x != -1 | y != -1 into (x & y) != -1.
11672 2017-09-22 Steve Ellcey <sellcey@cavium.com>
11674 * config.gcc: Add new case statement to set
11675 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
11676 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
11677 s390x-*-linux* case statements. Added aarch64 to the list of
11678 supported architectures.
11680 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11682 PR tree-optimization/82289
11683 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
11684 STMT_VINFO_RELEVANT_P.
11686 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11687 Alan Hayward <alan.hayward@arm.com>
11688 David Sherwood <david.sherwood@arm.com>
11690 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
11691 for VR_RANGE only; don't allow VR_ANTI_RANGE.
11692 (extract_range_from_binary_expr_1): Don't call
11693 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
11695 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11696 Alan Hayward <alan.hayward@arm.com>
11697 David Sherwood <david.sherwood@arm.com>
11699 * target.def (preferred_vector_alignment): New hook.
11700 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
11702 * doc/tm.texi: Regenerate.
11703 * targhooks.h (default_preferred_vector_alignment): Declare.
11704 * targhooks.c (default_preferred_vector_alignment): New function.
11705 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
11707 (DR_TARGET_ALIGNMENT): New macro.
11708 (aligned_access_p): Update commentary.
11709 (vect_known_alignment_in_bytes): New function.
11710 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
11712 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
11713 Calculate the misalignment based on the target alignment rather than
11715 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
11716 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
11717 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
11718 the target alignment, rather than masking the element misalignment
11719 with the number of elements in a vector. Also use the target
11720 alignment when calculating the maximum number of peels.
11721 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
11722 instead of TYPE_ALIGN_UNIT.
11723 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
11724 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
11725 (vect_create_addr_base_for_vector_ref): Update call accordingly.
11726 (vect_create_data_ref_ptr): Likewise.
11727 (vect_setup_realignment): Realign by ANDing with
11728 -DR_TARGET_MISALIGNMENT.
11729 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
11730 the number of peels based on DR_TARGET_ALIGNMENT.
11731 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
11732 with the guaranteed alignment boundary when deciding whether
11734 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
11735 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
11736 (ensure_base_align): Remove stmt_info parameter. Get the
11737 target base alignment from DR_TARGET_ALIGNMENT.
11738 (vectorizable_store): Update call accordingly. Interpret
11739 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
11741 (vectorizable_load): Likewise.
11743 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11744 Alan Hayward <alan.hayward@arm.com>
11745 David Sherwood <david.sherwood@arm.com>
11747 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
11748 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
11749 (vect_enhance_data_refs_alignment): Likewise.
11751 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11753 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
11754 error. Only quit immediately if parsing is complete.
11755 (BEGIN): Initialize fatal_err and parse_done.
11756 (begin fpu, end fpu): Check number of arguments.
11757 (begin arch, end arch): Likewise.
11758 (begin cpu, end cpu): Likewise.
11759 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
11760 (optalias): Likewise.
11762 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11764 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
11765 * config/arm/arm-isa.h: Delete. Move definitions to ...
11766 * arm-cpus.in: ... here. Use new feature and fgroup values.
11767 * config/arm/arm.c (arm_option_override): Use lower case for feature
11769 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
11770 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
11771 * config/arm/parsecpu.awk (END): Add new command 'isa'.
11773 (print_isa_bits_for): New function.
11774 (gen_isa): New function.
11775 (gen_comm_data): Use print_isa_bits_for.
11776 (define feature): New keyword.
11777 (define fgroup): New keyword.
11778 * config/arm/t-arm (TM_H): Remove.
11779 (GTM_H): Add arm-isa.h.
11780 (arm-isa.h): Add rule to generate file.
11781 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
11782 case for feature bit names.
11784 2017-09-22 Richard Biener <rguenther@suse.de>
11786 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
11788 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
11789 print of no dependency loops ...
11790 * graphite.c (graphite_transform_loops): ... here.
11791 (canonicalize_loop_closed_ssa_form): Work from inner to outer
11793 (same_close_phi_node, remove_duplicate_close_phi,
11794 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
11795 (canonicalize_loop_closed_ssa): ... here and simplify.
11796 * graphite-optimize-isl.c: Include tree-vectorizer.h.
11797 (optimize_isl): Use dump_printf_loc to tell when we stopped
11798 optimizing because of an ISL timeout.
11800 2017-09-22 Richard Biener <rguenther@suse.de>
11802 PR tree-optimization/82291
11803 * tree-if-conv.c (predicate_mem_writes): Make sure to
11804 remove writes in blocks predicated with false.
11806 2017-09-22 Richard Biener <rguenther@suse.de>
11808 * sese.c: Include cfganal.h.
11809 (if_region_set_false_region): Remove.
11810 (create_if_region_on_edge): Likewise.
11811 (move_sese_in_condition): Re-implement without destroying
11814 2017-09-22 Richard Biener <rguenther@suse.de>
11816 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
11817 Verify both BBs contain loop PHI nodes before dispatching to
11818 copy_loop_phi_args.
11819 (graphite_regenerate_ast_isl): Do not recompute dominators,
11820 do not verify three times. Restructure for clarity.
11821 * graphite-scop-detection.c (same_close_phi_node,
11822 remove_duplicate_close_phi, make_close_phi_nodes_unique,
11823 defined_in_loop_p, canonicalize_loop_closed_ssa,
11824 canonicalize_loop_closed_ssa_form): Simplify, remove excess
11825 checking and SSA rewrite, move to ...
11826 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
11827 (graphite_initialize): Do not pass in ctx, do not reset the
11828 SCEV cache, compute only dominators.
11829 (graphite_transform_loops): Allocate ISL ctx after
11830 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
11831 Maintain post-dominators only around build_scops.
11832 * sese.c (if_region_set_false_region): Make static. Free
11833 and recompute dominators.
11834 (move_sese_in_condition): Assert we don't get called with
11835 post-dominators computed.
11836 * sese.h (if_region_set_false_region): Remove.
11838 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
11840 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
11841 mode attribute for TARGET_AVX512VL.
11843 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
11845 * config/i386/i386.opt (mprefer-avx256): New option.
11846 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
11848 (ix86_preferred_simd_mode): Return 256-bit AVX modes
11849 for TARGET_PREFER_AVX256.
11850 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
11852 2017-09-21 Jeff Law <law@redhat.com>
11854 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
11855 Fix dump output if the only stack space is for pushed registers.
11857 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11859 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
11862 2017-09-21 Martin Sebor <msebor@redhat.com>
11865 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
11866 code (in C++) or code that triggers warnings.
11868 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
11870 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
11872 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
11874 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
11875 * sched-rgn.c: Ditto.
11876 * sel-sched-ir.c: Ditto.
11878 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
11880 * toplev.h (set_random_seed): Adjust return type.
11881 * toplev.c (init_local_tick): Move eager initialization of random_seed
11882 to get_random_seed. Adjust comment.
11883 (init_random_seed): Inline to get_random_seed, delete.
11884 (get_random_seed): Initialize random_seed lazily.
11885 (set_random_seed): Do not return previous value.
11886 (print_switch_value): Do not call get_random_seed.
11888 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
11890 * cgraph.c (delete_function_version): New, broken out from...
11891 (cgraph_node::delete_function_version): ...here. Rename to
11892 cgraph_node::delete_function_version_by_decl. Update all uses.
11893 (cgraph_node::remove): Call delete_function_version.
11895 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11898 * tree-inline.c (expand_call_inline): Emit clobber stmts for
11899 VAR_DECLs to which addressable non-volatile parameters are mapped
11900 and for id->retvar after the return value assignment. Clear
11901 id->retval and id->retbnd after inlining.
11903 2017-09-21 Richard Biener <rguenther@suse.de>
11905 PR tree-optimization/82276
11906 PR tree-optimization/82244
11907 * tree-vrp.c (build_assert_expr_for): Set
11908 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
11910 (remove_range_assertions): Revert earlier change.
11912 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
11915 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
11917 2017-09-21 Richard Biener <rguenther@suse.de>
11919 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11920 Restore valid IL after code generation errors.
11921 * graphite.c (graphite_transform_loops): Diagnose code
11922 generation issues as MSG_MISSED_OPTIMIZATION and continue
11923 with processing SCOPs.
11925 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11926 Alan Hayward <alan.hayward@arm.com>
11927 David Sherwood <david.sherwood@arm.com>
11929 * calls.c (compute_argument_addresses): Use simplify_gen_binary
11930 rather than choosing between plus_constant and gen_rtx_<CODE>.
11931 * expr.c (emit_push_insn): Likewise.
11932 (expand_expr_real_2): Likewise.
11934 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11935 Alan Hayward <alan.hayward@arm.com>
11936 David Sherwood <david.sherwood@arm.com>
11938 * loop-unroll.c (split_iv): Call copy_rtx on the step.
11940 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11941 Alan Hayward <alan.hayward@arm.com>
11942 David Sherwood <david.sherwood@arm.com>
11944 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
11945 calling tree_to_uhwi.
11947 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11948 Alan Hayward <alan.hayward@arm.com>
11949 David Sherwood <david.sherwood@arm.com>
11951 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
11952 INTEGER_CST rather than a negative test for ADDR_EXPR.
11954 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11955 Alan Hayward <alan.hayward@arm.com>
11956 David Sherwood <david.sherwood@arm.com>
11958 * tree-vrp.c (extract_range_from_binary_expr_1): Check
11959 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
11961 2017-09-21 Richard Biener <rguenther@suse.de>
11963 PR tree-optimization/71351
11964 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
11965 graphite_create_new_loop_guard): Remove, fold remaining parts
11967 (translate_isl_ast_node_for): ... here and simplify.
11969 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11972 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
11973 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
11974 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
11975 alternative always use QI mode, for -Os imov (=R,R) alternative
11976 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
11979 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11980 Jeff Law <law@redhat.com>
11982 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
11983 (allocate_stack_space): New function, partially extracted from
11984 s390_emit_prologue.
11985 (s390_emit_prologue): Track offset to most recent stack probe.
11986 Code to allocate space moved into allocate_stack_space.
11987 Dump actions when no stack is allocated.
11988 (s390_prologue_plus_offset): New function.
11989 (s390_emit_stack_probe): Likewise.
11991 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
11993 * common.opt (Wa, Wl, Wp, g, gz=): Add
11995 (gno-column-info): Remove.
11996 (gcolumn-info): Drop RejectNegative.
11997 (gno-): New prefix.
11998 (gno-record-gcc-switches): Remove.
11999 (grecord-gcc-switches): Drop RejectNegative.
12000 (gno-split-dwarf): Remove.
12001 (gsplit-dwarf): Drop RejectNegative.
12002 (gno-strict-dwarf): Remove.
12003 (gstrict-dwarf): Drop RejectNegative.
12004 * config/darwin.opt (gfull, gused): Add RejectNegative.
12005 * dwarf2out.c (gen_producer_string): Drop
12006 gno-record-gcc-switches handler.
12007 * optc-gen.awk: Add g to prefixes with negative forms.
12008 * opts-common.c (remapping_prefix_p): New.
12009 (find_opt): Check it.
12010 (generate_canonical_option): Test g prefix.
12011 (option_map): Add -gno- mapping.
12012 (add_misspelling_candidates): Check remapping_prefix_p.
12014 2017-09-20 Jeff Law <law@redhat.com>
12016 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
12017 thinko in stack clash protection support.
12019 * explow.c (compute_stack_clash_protection_loop_data): Use
12020 CONST_INT_P instead of explicit test. Verify object is a
12021 CONST_INT_P before looking at INTVAL.
12022 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
12023 instead of explicit test.
12025 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
12028 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
12029 address instead of to r1 and r11.
12031 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
12033 * config.gcc: Support "knm".
12034 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
12035 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
12037 * config/i386/i386.c (m_KNM): Define.
12038 (processor_target_table): Add "knm".
12040 (ix86_option_override_internal): Add "knm".
12041 (ix86_issue_rate): Add PROCESSOR_KNM.
12042 (ix86_adjust_cost): Ditto.
12043 (ia32_multipass_dfa_lookahead): Ditto.
12044 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
12045 (fold_builtin_cpu): Add M_INTEL_KNM.
12046 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
12047 (processor_type): Add PROCESSOR_KNM.
12048 * config/i386/x86-tune.def: Add m_KNM.
12049 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
12051 2017-09-20 Richard Biener <rguenther@suse.de>
12053 PR tree-optimization/80213
12054 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
12055 are allowed in empty BBs as well.
12056 (canonicalize_loop_closed_ssa): Also look for other complex
12058 (scop_detection::get_sese): Include the loop-closed PHI block
12060 (scop_detection::merge_sese): Remove code adding extra blocks.
12061 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
12062 (build_scops): Assert the final returned scop is invalid.
12064 2017-09-20 Richard Biener <rguenther@suse.de>
12066 PR tree-optimization/82264
12067 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
12069 (vn_phi_lookup): Likewise.
12070 (vn_phi_insert): Likewise.
12072 2017-09-20 Jakub Jelinek <jakub@redhat.com>
12074 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
12075 that fits into uhwi or shwi, add DW_AT_const_value regardless
12076 of early_dwarf without going through RTL, using add_AT_unsigned
12079 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
12080 (DEBUG_LTO_ABBREV_SECTION): Likewise.
12081 (DEBUG_LTO_MACINFO_SECTION): Likewise.
12082 (DEBUG_MACRO_SECTION): Likewise.
12083 (DEBUG_LTO_MACRO_SECTION): Likewise.
12084 (DEBUG_STR_DWO_SECTION): Likewise.
12085 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
12086 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
12087 (DEBUG_LTO_DWO_LINE_SECTION): Define.
12088 (DEBUG_LTO_LINE_STR_SECTION): Define.
12089 (init_sections_and_labels): Initialize debug_line_str_section
12090 variable. Initialize debug_loc_section for -gdwarf-5 to
12091 DEBUG_LOCLISTS_SECTION. Formatting fixes.
12093 2017-09-20 Richard Biener <rguenther@suse.de>
12095 * graphite-sese-to-poly.c (extract_affine): Properly handle
12096 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
12098 2017-09-20 Richard Biener <rguenther@suse.de>
12100 PR tree-optimization/81373
12101 * graphite-scop-detection.c (build_cross_bb_scalars_def):
12102 Force SESE live-out defs to be handled even if they are
12105 2017-09-19 Jeff Law <law@redhat.com>
12107 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
12108 nothing for stack adjustments with REG_STACK_CHECK.
12109 * sched-deps.c (parse_add_or_inc): Reject insns with
12110 REG_STACK_CHECK from dependency breaking.
12111 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
12112 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
12113 * reg-notes.def (STACK_CHECK): New note.
12115 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
12116 (ix86_expand_prologue): Dump stack clash info as needed.
12117 Call ix86_adjust_stack_and_probe_stack_clash as needed.
12119 * function.c (dump_stack_clash_frame_info): New function.
12120 * function.h (dump_stack_clash_frame_info): Prototype.
12121 (enum stack_clash_probes): New enum.
12123 * config/alpha/alpha.c (alpha_expand_prologue): Also check
12124 flag_stack_clash_protection.
12125 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
12126 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
12127 (arm_frame_pointer_required): Likewise.
12128 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
12129 (ia64_expand_prologue): Likewise.
12130 * config/mips/mips.c (mips_expand_prologue): Likewise.
12131 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
12132 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
12133 (sparc_flat_expand_prologue): Likewise.
12134 * config/spu/spu.c (spu_expand_prologue): Likewise.
12136 * explow.c: Include "params.h".
12137 (anti_adjust_stack_and_probe_stack_clash): New function.
12138 (get_stack_check_protect): Likewise.
12139 (compute_stack_clash_protection_loop_data): Likewise.
12140 (emit_stack_clash_protection_loop_start): Likewise.
12141 (emit_stack_clash_protection_loop_end): Likewise.
12142 (allocate_dynamic_stack_space): Use get_stack_check_protect.
12143 Use anti_adjust_stack_and_probe_stack_clash.
12144 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
12145 (emit_stack_clash_protection_loop_start): Likewise.
12146 (emit_stack_clash_protection_loop_end): Likewise.
12147 * rtl.h (get_stack_check_protect): Prototype.
12148 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
12149 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
12150 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
12152 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
12154 * doc/tm.texi: Rebuilt.
12155 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
12156 get_stack_check_protect.
12157 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
12158 * config/arm/arm.c (arm_expand_prologue): Likewise.
12159 (arm_frame_pointer_required): Likewise.
12160 * config/i386/i386.c (ix86_expand_prologue): Likewise.
12161 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
12162 * config/mips/mips.c (mips_expand_prologue): Likewise.
12163 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
12164 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
12165 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
12166 (sparc_flat_expand_prologue): Likewise.
12168 * common.opt (-fstack-clash-protection): New option.
12169 * flag-types.h (enum stack_check_type): Note difference between
12170 -fstack-check= and -fstack-clash-protection.
12171 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
12172 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
12173 * toplev.c (process_options): Issue warnings/errors for cases
12174 not handled with -fstack-clash-protection.
12175 * doc/invoke.texi (-fstack-clash-protection): Document new option.
12176 (-fstack-check): Note additional problem with -fstack-check=generic.
12177 Note that -fstack-check is primarily for Ada and refer users
12178 to -fstack-clash-protection for stack-clash-protection.
12179 Document new params for stack clash protection.
12181 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
12183 * config/i386/i386.c (ix86_split_long_move): Do not handle
12184 address used for LEA in a special way.
12186 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
12188 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
12190 2017-09-19 Martin Sebor <msebor@redhat.com>
12193 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
12194 of incompatible types.
12196 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
12198 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12199 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
12200 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12201 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
12203 2017-09-19 Richard Biener <rguenther@suse.de>
12205 PR tree-optimization/82244
12206 * tree-vrp.c (remove_range_assertions): Do not propagate
12207 a constant to abnormals but replace the assert with a copy.
12209 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
12211 PR rtl-optimization/57878
12212 PR rtl-optimization/68988
12213 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
12214 avoidance test involving non_reload_pseudos. Move frequency test
12215 below the general fragmentation avoidance test.
12217 2017-09-19 Richard Biener <rguenther@suse.de>
12219 PR tree-optimization/69728
12220 * graphite-sese-to-poly.c (schedule_error): New global.
12221 (add_loop_schedule): Handle empty domain by failing the
12223 (build_original_schedule): Handle schedule_error.
12225 2017-09-19 Richard Biener <rguenther@suse.de>
12227 * graphite-scop-detection.c (scop_detection::can_represent_loop):
12228 Do not iterate to sibling loops but only to siblings of inner
12231 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
12234 * config/m68k/m68k.md (moveq feeding equality comparison): Check
12235 that the registers are different.
12237 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
12239 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
12240 to processor_model and "amdfam17h" to arch_names_table.
12241 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
12243 2017-09-18 Jakub Jelinek <jakub@redhat.com>
12246 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
12248 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12249 Alan Hayward <alan.hayward@arm.com>
12250 David Sherwood <david.sherwood@arm.com>
12252 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
12254 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
12256 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
12257 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
12258 vect_schedule_slp_instance.
12259 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
12260 Update call to vect_slp_analyze_node_operations. Simplify return
12262 (vect_slp_analyze_bb_1): Update call accordingly.
12263 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
12264 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
12265 (vect_schedule_slp): Update call accordingly.
12267 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12268 Alan Hayward <alan.hayward@arm.com>
12269 David Sherwood <david.sherwood@arm.com>
12271 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
12272 with types that aren't in fact scalar.
12274 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12276 * tree-vect-slp.c (vect_record_max_nunits): New function,
12278 (vect_build_slp_tree_1): ...here.
12279 (vect_build_slp_tree_2): Call it for phis too.
12281 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12283 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
12284 to vect_get_vec_def_for_operand when getting the mask operand.
12286 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12287 Alan Hayward <alan.hayward@arm.com>
12288 David Sherwood <david.sherwood@arm.com>
12290 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
12293 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12294 Alan Hayward <alan.hayward@arm.com>
12295 David Sherwood <david.sherwood@arm.com>
12297 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
12298 calculation for vector booleans.
12300 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12301 Alan Hayward <alan.hayward@arm.com>
12302 David Sherwood <david.sherwood@arm.com>
12304 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
12306 (vect_transform_stmt): ...here.
12307 (vect_analyze_stmt): Use it instead of calling
12308 vectorizable_live_operation directly.
12310 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
12312 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
12313 non-SIMT targets in acc vector loops.
12315 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
12317 * configure.ac: Add arc and check if assembler supports gdwarf2.
12318 * configure: Regenerate.
12320 2017-09-18 Richard Biener <rguenther@suse.de>
12322 PR tree-optimization/82220
12323 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
12324 epilogue niters from the min_profitable_iters compute.
12326 2017-09-18 Jakub Jelinek <jakub@redhat.com>
12329 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
12330 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
12331 (ix86_init_pic_reg): Revert 2017-09-01 changes.
12333 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
12336 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
12337 switching to a new text section.
12339 2017-09-18 Richard Biener <rguenther@suse.de>
12341 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
12343 (build_alias_set): Reject aliases with no access function.
12345 2017-09-18 Richard Biener <rguenther@suse.de>
12347 PR tree-optimization/79622
12348 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
12350 (build_cross_bb_scalars_use): Likewise.
12352 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
12354 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
12356 2017-09-18 Alan Modra <amodra@gmail.com>
12359 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
12360 stack_pointer_rtx for count 0. Update comments. Break up
12361 large rtl expression.
12363 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
12365 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
12366 Increase to 20 bytes.
12367 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
12368 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
12369 or avx version of the stub.
12371 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
12374 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
12375 compute the minimum stack alignment. Also update preferred stack
12376 boundary for leaf functions.
12378 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
12380 PR tree-optimization/82228
12381 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
12384 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
12386 * common/config/nds32/nds32-common.c
12387 (nds32_option_optimization_table): Refine formatting.
12388 (nds32_option_optimization_table): Use -fsched-pressure and
12389 -fomit-frame-pointer for specific optimization level.
12391 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
12393 * config/nds32/nds32.c: Refine formatting and comments.
12394 * config/nds32/nds32.h: Likewise.
12395 * config/nds32/nds32.md: Likewise.
12396 * config/nds32/nds32-cost.c: Likewise.
12397 * config/nds32/nds32-isr.c: Likewise.
12398 * config/nds32/nds32-md-auxiliary.c: Likewise.
12399 * config/nds32/nds32-multiple.md: Likewise.
12400 * config/nds32/nds32-predicates.c: Likewise.
12402 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
12403 Jakub Jelinek <jakub@redhat.com>
12405 Add support for -std=c++2a.
12406 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
12408 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
12410 2017-09-15 Steve Ellcey <sellcey@cavium.com>
12413 * doc/extend.texi (Common Function Attributes): Add
12414 references to ARM, AArch64, and S/390 specific attributes.
12415 (Function Specific Option Pragmas): Add AArch64 and S/390
12416 to list of back ends that support the target pragma.
12418 2017-09-15 Nathan Sidwell <nathan@acm.org>
12420 * doc/standards.texi: Fix C++17 description. Update URLs for
12423 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
12425 * common.opt (Wcast-align=strict): New warning option.
12426 * doc/invoke.texi: Document -Wcast-align=strict.
12428 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
12430 * cgraph.h (cgraph_thunk_info): Add comments.
12431 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
12432 assert for VIRTUAL_* arguments stricter.
12434 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
12436 PR tree-optimization/71026
12437 * match.pd: Move RDIV patterns from fold-const.c
12438 * fold-const.c (distribute_real_division): Removed.
12439 (fold_binary_loc): Remove calls to distribute_real_divison.
12441 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12443 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
12444 c++1z and gnu++1z as deprecated. Change other references to
12445 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
12446 Change -Wc++1z-compat to -Wc++17-compat.
12447 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
12448 * dwarf2out.c (highest_c_language): Handle C++17.
12449 (gen_compile_unit_die): Likewise.
12451 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12453 PR rtl-optimization/82192
12454 * combine.c (make_extraction): Don't look through non-paradoxical
12455 SUBREGs or TRUNCATE if pos + len is or might be bigger than
12458 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
12459 Alan Hayward <alan.hayward@arm.com>
12460 David Sherwood <david.sherwood@arm.com>
12462 * target.def (function_arg_offset): New hook.
12463 * targhooks.h (default_function_arg_offset): Declare.
12464 * targhooks.c (default_function_arg_offset): New function.
12465 * function.c (locate_and_pad_parm): Use
12466 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
12467 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
12468 (TARGET_FUNCTION_ARG_OFFSET): ...this.
12469 * doc/tm.texi: Regenerate.
12470 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
12471 * config/spu/spu.c (spu_function_arg_offset): New function.
12472 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
12473 * system.h (FUNCTION_ARG_OFFSET): Poison.
12475 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
12476 Alan Hayard <alan.hayward@arm.com>
12477 David Sherwood <david.sherwood@arm.com>
12479 * target.def (truly_noop_truncation): New hook.
12480 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
12481 than TRULY_NOOP_TRUNCATION.
12482 * hooks.h (hook_bool_uint_uint_true): Declare.
12483 * hooks.c (hook_bool_uint_uint_true): New function.
12484 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
12485 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
12486 * doc/tm.texi: Regenerate.
12487 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
12488 rather than TRULY_NOOP_TRUNCATION in comments.
12489 (simplify_comparison): Likewise.
12490 (record_truncated_value): Likewise.
12491 * expmed.c (extract_bit_field_1): Likewise.
12492 (extract_split_bit_field): Likewise.
12493 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
12494 instead of TRULY_NOOP_TRUNCATION.
12495 * function.c (assign_parm_setup_block): Likewise.
12496 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
12497 * rtlhooks.c: Include target.h.
12498 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
12499 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
12500 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
12501 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
12502 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
12503 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
12504 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
12505 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
12506 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
12507 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
12508 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
12509 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
12510 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
12511 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
12512 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
12513 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
12514 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
12515 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
12516 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
12517 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
12518 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
12519 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
12520 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
12521 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
12522 * config/mips/mips.c (mips_truly_noop_truncation): New function.
12523 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
12524 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
12525 rather than TRULY_NOOP_TRUNCATION in comments.
12526 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
12527 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
12528 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
12529 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
12530 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
12531 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
12532 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
12533 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
12534 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
12535 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
12536 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
12537 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
12538 rather than TRULY_NOOP_TRUNCATION in comments.
12539 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
12540 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
12541 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
12542 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
12543 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
12544 TRULY_NOOP_TRUNCATION condition.
12545 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
12546 (TRULY_NOOP_TRUNCATION): Delete.
12547 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
12548 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
12549 * config/spu/spu.c (spu_truly_noop_truncation): New function.
12550 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
12551 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
12552 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
12553 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
12554 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
12555 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
12556 rather than TRULY_NOOP_TRUNCATION in comments.
12557 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
12558 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
12559 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
12560 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
12561 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
12562 * system.h (TRULY_NOOP_TRUNCATION): Poison.
12564 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
12567 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
12568 (*cmp_ior): Likewise.
12569 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
12570 (*ior_scc_scc_cmp): Likewise.
12571 (*and_scc_scc): Likewise.
12572 (*and_scc_scc_cmp): Likewise.
12574 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
12575 Alan Hayard <alan.hayward@arm.com>
12576 David Sherwood <david.sherwood@arm.com>
12578 * target.def (can_change_mode_class): New hook.
12579 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
12580 (hard_regno_nregs): Likewise.
12581 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
12582 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
12583 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
12584 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
12585 (LOAD_EXTEND_OP): Update accordingly.
12586 * doc/tm.texi: Regenerate.
12587 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
12588 CANNOT_CHANGE_MODE_CLASS.
12589 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
12590 (REG_CAN_CHANGE_MODE_P): ...this new macro.
12591 * combine.c (simplify_set): Update accordingly.
12592 * emit-rtl.c (validate_subreg): Likewise.
12593 * recog.c (general_operand): Likewise.
12594 * regcprop.c (mode_change_ok): Likewise.
12595 * reload1.c (choose_reload_regs): Likewise.
12596 (inherit_piecemeal_p): Likewise.
12597 * rtlanal.c (simplify_subreg_regno): Likewise.
12598 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
12599 instead of CANNOT_CHANGE_MODE_CLASS.
12600 (reload_cse_simplify_operands): Likewise.
12601 * reload.c (push_reload): Use targetm.can_change_mode_class
12602 instead of CANNOT_CHANGE_MODE_CLASS.
12603 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
12604 REG_CANNOT_CHANGE_MODE_P.
12605 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12606 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
12607 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12608 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12609 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12610 (arm_can_change_mode_class): New function.
12611 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
12612 than CANNOT_CHANGE_MODE_CLASS in comments.
12613 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12614 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
12615 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
12616 (ix86_can_change_mode_class): ...this new function, inverting the
12617 sense of the return value.
12618 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12619 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12620 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12621 (ia64_can_change_mode_class): New function.
12622 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12623 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
12624 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
12625 (m32c_can_change_mode_class): ...this new function, inverting the
12626 sense of the return value.
12627 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12628 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12629 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
12630 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
12631 (mips_can_change_mode_class): ...this new function, inverting the
12632 sense of the return value.
12633 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12634 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12635 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12636 (msp430_can_change_mode_class): New function.
12637 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12638 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
12639 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12640 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12641 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12642 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
12643 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12644 (pa_cannot_change_mode_class): Replace with...
12645 (pa_can_change_mode_class): ...this new function, inverting the
12646 sense of the return value.
12647 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
12648 than CANNOT_CHANGE_MODE_CLASS in comments.
12649 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12650 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
12651 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12652 (pdp11_cannot_change_mode_class): Replace with...
12653 (pdp11_can_change_mode_class): ...this new function, inverting the
12654 sense of the return value.
12655 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12656 * config/powerpcspe/powerpcspe-protos.h
12657 (rs6000_cannot_change_mode_class_ptr): Delete.
12658 * config/powerpcspe/powerpcspe.c
12659 (rs6000_cannot_change_mode_class_ptr): Delete.
12660 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12661 (rs6000_option_override_internal): Assign to
12662 targetm.can_change_mode_class instead of
12663 rs6000_cannot_change_mode_class_ptr.
12664 (rs6000_cannot_change_mode_class): Replace with...
12665 (rs6000_can_change_mode_class): ...this new function, inverting the
12666 sense of the return value.
12667 (rs6000_debug_cannot_change_mode_class): Replace with...
12668 (rs6000_debug_can_change_mode_class): ...this new function.
12669 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12670 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
12671 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12672 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12673 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
12675 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
12676 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12677 (rs6000_option_override_internal): Assign to
12678 targetm.can_change_mode_class instead of
12679 rs6000_cannot_change_mode_class_ptr.
12680 (rs6000_cannot_change_mode_class): Replace with...
12681 (rs6000_can_change_mode_class): ...this new function, inverting the
12682 sense of the return value.
12683 (rs6000_debug_cannot_change_mode_class): Replace with...
12684 (rs6000_debug_can_change_mode_class): ...this new function.
12685 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12686 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
12687 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
12688 (s390_can_change_mode_class): ...this new function, inverting the
12689 sense of the return value.
12690 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12691 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12692 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
12693 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12694 (sh_cannot_change_mode_class): Replace with...
12695 (sh_can_change_mode_class): ...this new function, inverting the
12696 sense of the return value.
12697 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12698 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12699 (sparc_can_change_mode_class): New function.
12700 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12701 * config/spu/spu.c (spu_can_change_mode_class): New function.
12702 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12703 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12704 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12705 (visium_can_change_mode_class): New function.
12706 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
12708 2017-09-15 Richard Biener <rguenther@suse.de>
12710 PR tree-optimization/82217
12711 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
12712 but not undefined case.
12714 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12717 * postreload.c (reload_cse_simplify_operands): Skip
12718 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
12720 2017-09-15 Richard Biener <rguenther@suse.de>
12722 PR tree-optimization/68823
12723 * graphite-scop-detection.c (build_alias_set): If we have a
12724 possible dependence check whether we can handle them by just
12725 looking at the DRs DR_ACCESS_FNs.
12726 (build_scops): If build_alias_set fails, fail the SCOP.
12728 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
12730 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
12731 to support float128 built-in functions that require the ISA 3.0
12733 (BU_FLOAT128_3_HW): Likewise.
12734 (SQRTF128): Add support for the IEEE 128-bit square root and fma
12735 built-in functions.
12736 (FMAF128): Likewise.
12738 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
12739 support for built-in functions that need the ISA 3.0 IEEE 128-bit
12740 floating point instructions.
12741 (rs6000_invalid_builtin): Likewise.
12742 (rs6000_builtin_mask_names): Likewise.
12743 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
12744 (RS6000_BTM_FLOAT128_HW): Likewise.
12745 (RS6000_BTM_COMMON): Likewise.
12746 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
12748 * doc/extend.texi (RS/6000 built-in functions): Document the
12749 IEEE 128-bit floating point square root and fused multiply-add
12750 built-in functions.
12752 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
12754 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
12755 reg (r2) isn't in the set of registers defined in the prologue.
12757 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12758 Alan Hayward <alan.hayward@arm.com>
12759 David Sherwood <david.sherwood@arm.com>
12761 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
12762 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
12763 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
12764 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
12765 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
12767 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
12768 max_vectorization_factor.
12769 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
12771 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12772 Alan Hayward <alan.hayward@arm.com>
12773 David Sherwood <david.sherwood@arm.com>
12775 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
12776 (vect_worthwhile_without_simd_p): Declare.
12777 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
12778 (vectorizable_reduction): Use it.
12779 * tree-vect-stmts.c (vectorizable_shift): Likewise.
12780 (vectorizable_operation): Likewise.
12782 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12783 Alan Hayward <alan.hayward@arm.com>
12784 David Sherwood <david.sherwood@arm.com>
12786 * tree-vectorizer.h (vect_get_num_copies): New function.
12787 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
12788 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12789 (vectorizable_induction): Likewise.
12790 (vectorizable_live_operation): Likewise.
12791 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12792 (vectorizable_bswap): Likewise.
12793 (vectorizable_call): Likewise.
12794 (vectorizable_conversion): Likewise.
12795 (vectorizable_assignment): Likewise.
12796 (vectorizable_shift): Likewise.
12797 (vectorizable_operation): Likewise.
12798 (vectorizable_store): Likewise.
12799 (vectorizable_load): Likewise.
12800 (vectorizable_condition): Likewise.
12801 (vectorizable_comparison): Likewise.
12802 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
12804 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12805 Alan Hayward <alan.hayward@arm.com>
12806 David Sherwood <david.sherwood@arm.com>
12808 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
12809 of vect_init_vector.
12811 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12812 Alan Hayward <alan.hayward@arm.com>
12813 David Sherwood <david.sherwood@arm.com>
12815 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
12816 an inline wrapper that provides a location.
12817 (gimple_build_vector): Likewise.
12818 * gimple-fold.c (gimple_build_vector_from_val): New function.
12819 (gimple_build_vector): Likewise.
12820 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
12821 functions to build the initial value. Always return a gimple value.
12822 (get_initial_defs_for_reduction): Likewise. Only compute
12824 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
12825 vect_init_vector on the results from get_initial_def(s)_for_reduction.
12826 (vectorizable_induction): Use gimple_build_vector rather than
12829 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12830 Alan Hayward <alan.hayward@arm.com>
12831 David Sherwood <david.sherwood@arm.com>
12833 * target.h (vec_perm_indices): New typedef.
12834 (auto_vec_perm_indices): Likewise.
12835 * optabs-query.h: Include target.h
12836 (can_vec_perm_p): Take a vec_perm_indices *.
12837 * optabs-query.c (can_vec_perm_p): Likewise.
12838 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
12839 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12840 * tree-vect-generic.c (lower_vec_perm): Likewise.
12841 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
12842 (vect_grouped_load_supported): Likewise.
12843 (vect_shift_permute_load_chain): Likewise.
12844 (vect_permute_store_chain): Use auto_vec_perm_indices.
12845 (vect_permute_load_chain): Likewise.
12846 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
12847 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
12848 Update uses of can_vec_perm_p.
12849 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
12850 mode with a number of elements. Take a vec_perm_indices *.
12851 (vect_create_epilog_for_reduction): Update accordingly.
12852 Use auto_vec_perm_indices.
12853 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
12854 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
12855 (vect_transform_slp_perm_load): Likewise.
12856 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
12857 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
12858 (vect_gen_perm_mask_checked): Likewise.
12859 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
12860 (vect_gen_perm_mask_checked): Likewise.
12861 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
12862 (vectorizable_store): Likewise.
12863 (vectorizable_load): Likewise.
12864 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
12865 (vectorizable_bswap): Likewise.
12867 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12868 Alan Hayward <alan.hayward@arm.com>
12869 David Sherwood <david.sherwood@arm.com>
12871 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
12872 * tree.c (build_vector): Likewise.
12873 (build_vector_from_ctor): Update accordingly.
12874 (build_vector_from_val): Likewise.
12875 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
12876 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12877 * tree-vect-generic.c (add_rshift): Likewise.
12878 (expand_vector_divmod): Likewise.
12879 (optimize_vector_constructor): Likewise.
12880 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
12881 (vect_transform_slp_perm_load): Likewise.
12882 (vect_schedule_slp_instance): Likewise.
12883 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
12884 (vectorizable_call): Likewise.
12885 (vect_gen_perm_mask_any): Likewise. Add elements in order.
12886 * expmed.c (make_tree): Likewise.
12887 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
12888 a vector passed to build_vector.
12889 (fold_convert_const): Likewise.
12890 (exact_inverse): Likewise.
12891 (fold_ternary_loc): Likewise.
12892 (fold_relational_const): Likewise.
12893 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
12894 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
12895 (const_unop): Likewise. Store the reduction accumulator in a
12896 variable rather than an array.
12897 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
12898 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
12899 the new vector, rather than constructing it after the input arrays.
12900 (native_interpret_vector): Use auto_vec<tree> when building
12901 a vector passed to build_vector. Add elements in order.
12902 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
12903 auto_vec<tree> when building a vector passed to build_vector.
12904 (vect_create_epilog_for_reduction): Likewise.
12905 (vectorizable_induction): Likewise.
12906 (get_initial_def_for_reduction): Likewise. Fix indentation of
12908 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
12910 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
12911 passed to build_vector.
12913 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12914 Alan Hayward <alan.hayward@arm.com>
12915 David Sherwood <david.sherwood@arm.com>
12917 * tree-core.h (tree_base::u): Add an "nelts" field.
12918 (tree_vector): Use VECTOR_CST_NELTS as the length.
12919 * tree.c (tree_size): Likewise.
12920 (make_vector): Initialize VECTOR_CST_NELTS.
12921 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
12922 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
12923 TYPE_VECTOR_SUBPARTS.
12924 * expr.c (const_vector_mask_from_tree): Consistently use "units"
12925 as the number of units, setting it from VECTOR_CST_NELTS.
12926 (const_vector_from_tree): Likewise.
12927 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
12928 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
12929 (fold_negate_expr_1): Likewise.
12930 (fold_convert_const): Likewise.
12931 (const_binop): Likewise. Differentiate the number of output and
12933 (const_unop): Likewise.
12934 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
12935 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
12936 in cases that did the opposite.
12938 2017-09-14 Richard Biener <rguenther@suse.de>
12940 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
12943 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
12945 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
12947 2017-09-14 Jakub Jelinek <jakub@redhat.com>
12950 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
12951 if and where to split a bb, except for splitting before debug insn
12952 sequences followed by non-label real insn. Delete debug insns
12953 in between basic blocks.
12955 * combine.c (make_compound_operation_int): Formatting fixes.
12957 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
12958 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12959 * config/netbsd.h (LINK_EH_SPEC): Likewise.
12960 * config/sol2.h (LINK_EH_SPEC): Likewise.
12961 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12962 * config/s390/linux.h (LINK_SPEC): Likewise.
12963 * config/freebsd.h (LINK_EH_SPEC): Likewise.
12964 * config/openbsd.h (LINK_EH_SPEC): Likewise.
12965 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12966 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
12967 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
12968 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12969 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
12970 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
12972 2017-09-13 Jakub Jelinek <jakub@redhat.com>
12974 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
12976 (ENDFILE_LINUX_SPEC): Likewise.
12977 (LINK_EH_SPEC): Likewise.
12978 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
12979 (LINK_OS_LINUX_SPEC32): Likewise.
12980 (LINK_OS_LINUX_SPEC64): Likewise.
12981 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
12982 (LINK_OS_LINUX_SPEC): Likewise.
12984 2017-09-13 Martin Liska <mliska@suse.cz>
12986 PR middle-end/82154
12987 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
12988 CASE_HIGH is NULL_TREE.
12990 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12991 Alan Hayward <alan.hayward@arm.com>
12992 David Sherwood <david.sherwood@arm.com>
12994 * target.def (secondary_memory_needed): New hook.
12995 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
12996 instead of SECONDARY_MEMORY_NEEDED.
12997 (secondary_memory_needed_mode): Likewise.
12998 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
12999 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
13000 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
13001 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
13002 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
13003 * doc/tm.texi: Regenerate.
13004 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
13005 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
13006 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13007 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
13008 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
13009 * config/i386/i386.c (inline_secondary_memory_needed): Put the
13010 mode argument first and change the reg_class arguments to reg_class_t.
13011 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
13012 Make static. Update the call to inline_secondary_memory_needed.
13013 (ix86_register_move_cost): Update the call to
13014 inline_secondary_memory_needed.
13015 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13016 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
13018 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
13019 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
13021 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
13022 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
13023 * config/mips/mips.c (mips_secondary_memory_needed): Make static
13024 and match hook interface. Add comment from mips.h.
13025 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13026 * config/mmix/mmix.md (truncdfsf2): Refer to
13027 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
13029 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
13030 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
13031 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13032 (pa_secondary_memory_needed): New function.
13033 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
13034 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
13035 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13036 (pdp11_secondary_memory_needed): Make static and match hook interface.
13037 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
13038 * config/powerpcspe/powerpcspe-protos.h
13039 (rs6000_secondary_memory_needed_ptr): Delete.
13040 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
13042 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13043 (rs6000_option_override_internal): Assign to
13044 targetm.secondary_memory_needed rather than
13045 rs6000_secondary_memory_needed_ptr.
13046 (rs6000_secondary_memory_needed): Match hook interface.
13047 (rs6000_debug_secondary_memory_needed): Likewise.
13048 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
13049 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
13050 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
13051 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13052 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
13053 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
13055 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
13056 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13057 (rs6000_option_override_internal): Assign to
13058 targetm.secondary_memory_needed rather than
13059 rs6000_secondary_memory_needed_ptr.
13060 (rs6000_secondary_memory_needed): Match hook interface.
13061 (rs6000_debug_secondary_memory_needed): Likewise.
13062 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
13063 * config/s390/s390.c (s390_secondary_memory_needed): New function.
13064 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13065 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
13066 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13067 (sparc_secondary_memory_needed): New function.
13068 * lra-constraints.c (check_and_process_move): Refer to
13069 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
13071 (curr_insn_transform): Likewise.
13072 (process_alt_operands): Use targetm.secondary_memory_needed
13073 instead of TARGET_SECONDARY_MEMORY_NEEDED.
13074 (check_secondary_memory_needed_p): Likewise.
13075 (choose_split_class): Likewise.
13076 * reload.c: Unconditionally include code that was previously
13077 conditional on SECONDARY_MEMORY_NEEDED.
13078 (push_secondary_reload): Use targetm.secondary_memory_needed
13079 instead of TARGET_SECONDARY_MEMORY_NEEDED.
13080 (push_reload): Likewise.
13081 * reload1.c: Unconditionally include code that was previously
13082 conditional on SECONDARY_MEMORY_NEEDED.
13083 (choose_reload_regs): Use targetm.secondary_memory_needed
13084 instead of TARGET_SECONDARY_MEMORY_NEEDED.
13085 (gen_reload): Likewise.
13086 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
13088 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
13089 Alan Hayward <alan.hayward@arm.com>
13090 David Sherwood <david.sherwood@arm.com>
13092 * target.def (secondary_memory_needed_mode): New hook:
13093 * targhooks.c (default_secondary_memory_needed_mode): Declare.
13094 * targhooks.h (default_secondary_memory_needed_mode): New function.
13095 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
13096 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
13097 * doc/tm.texi: Regenerate.
13098 * lra-constraints.c (check_and_process_move): Use
13099 targetm.secondary_memory_needed_mode instead of
13100 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
13101 (curr_insn_transform): Likewise.
13102 * reload.c (get_secondary_mem): Likewise.
13103 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13104 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
13106 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13107 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13108 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
13109 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13110 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
13112 * config/powerpcspe/powerpcspe-protos.h
13113 (rs6000_secondary_memory_needed_mode): Delete.
13114 * config/powerpcspe/powerpcspe.c
13115 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13116 (rs6000_secondary_memory_needed_mode): Make static.
13117 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13118 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
13120 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
13122 (rs6000_secondary_memory_needed_mode): Make static.
13123 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13124 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
13125 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13126 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13127 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
13129 (sparc_secondary_memory_needed_mode): New function.
13130 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
13132 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
13134 * config/aarch64/constraints.md (Umq): New constraint.
13135 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
13137 (mov<mode>): Update condition.
13139 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13141 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
13142 when bitposition is the same.
13144 2017-09-13 Richard Biener <rguenther@suse.de>
13146 * dwarf2out.c (output_die_symbol): Remove.
13147 (output_die): Do not output a DIEs symbol.
13149 2017-09-13 Richard Biener <rguenther@suse.de>
13151 PR middle-end/82128
13152 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
13153 default-def to avoid breaking iterator update with the weird
13154 interaction with cgraph_update_edges_for_call_stmt_node.
13156 2017-09-13 Richard Biener <rguenther@suse.de>
13158 * tree-cfg.c (verify_gimple_assign_binary): Add verification
13159 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
13160 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
13161 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
13163 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
13165 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
13166 Disable pc relative literal load irrespective of
13167 TARGET_FIX_ERR_A53_84341 for default.
13169 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
13171 * config/sparc/sparc.c (output_return): Output the source location of
13172 the insn in the delay slot, if any.
13173 (output_sibcall): Likewise.
13175 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
13178 * common.opt (-static-pie): New alias.
13179 (shared): Negate static-pie.
13180 (-no-pie): Update help text.
13182 (static-pie): New option.
13183 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
13184 -static-pie support.
13185 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
13186 (LINK_EH_SPEC): Likewise.
13187 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13188 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
13189 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
13190 * gcc.c (LINK_COMMAND_SPEC): Likewise.
13191 (init_gcc_specs): Likewise.
13192 (init_spec): Likewise.
13193 (display_help): Update help message for -pie.
13194 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
13197 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
13199 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
13200 (movdi_aarch64): Likewise.
13201 (movti_aarch64): Likewise.
13203 2017-09-12 Simon Wright <simon@pushface.org>
13206 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
13207 calculation of the minor version, always output as 0.
13209 2017-09-12 Jakub Jelinek <jakub@redhat.com>
13212 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
13213 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
13214 on it early, rather than manual conversion late. For
13215 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
13216 instead of performing manual conversion.
13218 2017-09-12 Carl Love <cel@us.ibm.com>
13220 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
13221 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
13223 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13224 VMULOSW): Add definitions.
13225 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13226 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13227 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13228 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13229 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13231 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
13233 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
13235 (movti_aarch64): Likewise.
13236 (movdf_aarch64): Likewise.
13237 (movtf_aarch64): Likewise.
13238 (load_pairdi): Likewise.
13239 (store_pairdi): Likewise.
13240 (load_pairdf): Likewise.
13241 (store_pairdf): Likewise.
13242 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
13243 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
13244 (ldr_got_small_<mode>): Likewise.
13245 (ldr_got_small_28k_<mode>): Likewise.
13246 (ldr_got_tiny): Likewise.
13247 * config/aarch64/iterators.md (ldst_sz): New.
13248 (ldpstp_sz): Likewise.
13249 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
13251 (thunderx_load): Split load_8 to load_16.
13252 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
13254 (thunderx2t99_storepair_basic): Split store_8 to store_16.
13255 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
13256 (xgene1_store_pair): Split store_8 to store_16.
13257 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
13258 (falkor_st_0_st_sd): Split store_8 to store_16.
13260 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
13262 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
13263 and store1/2/3/4 to store_4/8/12/16.
13264 * config/aarch64/aarch64.md: Update for rename.
13265 * config/arm/arm.md: Likewise.: Likewise.
13266 * config/arm/arm.c: Likewise.
13267 * config/arm/thumb1.md: Likewise.
13268 * config/arm/thumb2.md: Likewise.
13269 * config/arm/vfp.md: Likewise.
13270 * config/arm/arm-generic.md: Likewise.
13271 * config/arm/arm1020e.md: Likewise.
13272 * config/arm/arm1026ejs.md: Likewise.
13273 * config/arm/arm1136jfs.md: Likewise.
13274 * config/arm/arm926ejs.md: Likewise.
13275 * config/arm/cortex-a15.md: Likewise.
13276 * config/arm/cortex-a17.md: Likewise.
13277 * config/arm/cortex-a5.md: Likewise.
13278 * config/arm/cortex-a53.md: Likewise.
13279 * config/arm/cortex-a57.md: Likewise.
13280 * config/arm/cortex-a7.md: Likewise.
13281 * config/arm/cortex-a8.md: Likewise.
13282 * config/arm/cortex-a9.md: Likewise.
13283 * config/arm/cortex-m4.md: Likewise.
13284 * config/arm/cortex-m7.md: Likewise.
13285 * config/arm/cortex-r4.md: Likewise.
13286 * config/arm/exynos-m1.md: Likewise.
13287 * config/arm/fa526.md: Likewise.
13288 * config/arm/fa606te.md: Likewise.
13289 * config/arm/fa626te.md: Likewise.
13290 * config/arm/fa726te.md: Likewise.
13291 * config/arm/fmp626.md: Likewise.
13292 * config/arm/iwmmxt.md: Likewise.
13293 * config/arm/ldmstm.md: Likewise.
13294 * config/arm/marvell-pj4.md: Likewise.
13295 * config/arm/xgene1.md: Likewise.
13296 * config/aarch64/thunderx.md: Likewise.
13297 * config/aarch64/thunderx2t99.md: Likewise.
13298 * config/aarch64/falkor.md: Likewise.
13300 2017-09-12 Martin Liska <mliska@suse.cz>
13302 * attribs.c (private_lookup_attribute): New function.
13303 * attribs.h (private_lookup_attribute): Declared here.
13304 (lookup_attribute): Called from this place.
13306 2017-09-12 Richard Biener <rguenther@suse.de>
13308 PR tree-optimization/82157
13309 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
13310 stmts with side-effects.
13312 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13313 Alan Hayward <alan.hayward@arm.com>
13314 David Sherwood <david.sherwood@arm.com>
13316 * target.def (hard_regno_nregs): New hook.
13317 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
13318 * targhooks.h (default_hard_regno_nregs): Declare.
13319 * targhooks.c (default_hard_regno_nregs): New function.
13320 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
13321 (TARGET_HARD_REGNO_NREGS): ...this hook.
13322 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
13323 (CLASS_MAX_NREGS): Likewise.
13324 * doc/tm.texi: Regenerate.
13325 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
13326 instead of HARD_REGNO_NREGS.
13327 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
13328 HARD_REGNO_NREGS in the comment.
13329 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
13330 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
13331 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
13332 Return an unsigned int.
13333 (TARGET_HARD_REGNO_NREGS): Redefine.
13334 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
13335 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
13336 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
13337 (arc_hard_regno_nregs): New function.
13338 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
13339 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
13340 (arm_hard_regno_nregs): New function.
13341 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
13342 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
13343 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
13344 (TARGET_HARD_REGNO_NREGS): Redefine.
13345 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
13346 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
13347 (HARD_REGNO_NREGS): Delete.
13348 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
13349 (cr16_hard_regno_nregs): New function.
13350 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
13351 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
13352 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
13353 (cris_hard_regno_nregs): New function.
13354 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
13355 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
13356 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
13357 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
13358 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
13359 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
13360 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
13361 (frv_hard_regno_nregs): Make static. Take and return an
13363 (frv_class_max_nregs): Remove outdated copy of documentation.
13364 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
13365 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
13366 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
13367 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
13368 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
13369 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
13370 (TARGET_HARD_REGNO_NREGS): Redefine.
13371 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
13372 (CLASS_MAX_NREGS): Update comment.
13373 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
13374 (ia64_hard_regno_nregs): New function.
13375 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
13376 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
13377 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
13378 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
13379 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
13381 (m32c_hard_regno_nregs): Likewise. Make static.
13382 (TARGET_HARD_REGNO_NREGS): Redefine.
13383 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
13384 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
13385 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
13386 (m68k_hard_regno_nregs): New function.
13387 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
13388 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
13389 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
13390 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
13391 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
13392 Take and return an unsigned int.
13393 (TARGET_HARD_REGNO_NREGS): Redefine.
13394 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
13395 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
13396 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
13397 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
13398 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
13399 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
13400 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
13401 (msp430_hard_regno_nregs): Make static. Take and return an
13403 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
13404 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
13405 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
13406 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
13407 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
13408 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
13409 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
13410 (TARGET_HARD_REGNO_NREGS): Redefine.
13411 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
13412 (PA_HARD_REGNO_NREGS): ...this.
13413 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
13414 (PA_HARD_REGNO_NREGS): ...this.
13415 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
13416 (pa_hard_regno_nregs): New function.
13417 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
13418 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
13419 (pdp11_hard_regno_nregs): New function.
13420 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
13421 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
13422 (rs6000_hard_regno_nregs_hook): New function.
13423 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
13424 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
13425 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
13426 Take and return an unsigned int. Move earlier in file.
13427 (TARGET_HARD_REGNO_NREGS): Redefine.
13428 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
13429 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
13430 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
13431 (rl78_hard_regno_nregs): Make static. Take and return an
13433 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
13434 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
13435 (rs6000_hard_regno_nregs_hook): New function.
13436 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
13437 * config/rx/rx.c (rx_hard_regno_nregs): New function.
13438 (TARGET_HARD_REGNO_NREGS): Redefine.
13439 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
13440 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
13441 instead of HARD_REGNO_NREGS.
13442 (s390_hard_regno_nregs): New function.
13443 (s390_hard_regno_mode_ok): Add comment from s390.h.
13444 (TARGET_HARD_REGNO_NREGS): Redefine.
13445 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
13446 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
13447 (sh_hard_regno_nregs): New function.
13448 (sh_pass_in_reg_p): Use it.
13449 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
13450 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
13451 (sparc_hard_regno_nregs): New function.
13452 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
13453 * config/spu/spu.c (spu_hard_regno_nregs): New function.
13454 (spu_function_arg_advance): Use it, supplying a valid register number.
13455 (TARGET_HARD_REGNO_NREGS): Redefine.
13456 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
13457 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
13458 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
13459 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
13460 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
13461 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
13462 (CLASS_MAX_NREGS): Remove copy of old documentation.
13463 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
13464 (visium_hard_regno_nregs): New function.
13465 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
13466 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
13467 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
13468 xtensa_hard_regno_nregs): New function.
13469 * system.h (HARD_REGNO_NREGS): Poison.
13471 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13473 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
13474 hard_regno_nregs instead of HARD_REGNO_NREGS.
13475 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
13476 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
13477 (c6x_expand_epilogue): Likewise.
13478 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
13479 (frv_read_iacc_argument): Likewise.
13480 * config/sh/sh.c: Include regs.h.
13481 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
13482 (regs_used): Likewise.
13483 (output_stack_adjust): Likewise.
13484 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
13485 * expmed.c: Include regs.h.
13486 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
13487 * ree.c: Include regs.h.
13488 (combine_reaching_defs): Use hard_regno_nregs instead of
13490 (add_removable_extension): Likewise.
13492 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13494 * regs.h (hard_regno_nregs): Turn into a function.
13495 (end_hard_regno): Update accordingly.
13496 * caller-save.c (setup_save_areas): Likewise.
13497 (save_call_clobbered_regs): Likewise.
13498 (replace_reg_with_saved_mem): Likewise.
13499 (insert_restore): Likewise.
13500 (insert_save): Likewise.
13501 * combine.c (can_change_dest_mode): Likewise.
13502 (move_deaths): Likewise.
13503 (distribute_notes): Likewise.
13504 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
13505 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
13506 (rs6000_split_multireg_move): Likewise.
13507 (rs6000_register_move_cost): Likewise.
13508 (rs6000_memory_move_cost): Likewise.
13509 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
13510 (rs6000_split_multireg_move): Likewise.
13511 (rs6000_register_move_cost): Likewise.
13512 (rs6000_memory_move_cost): Likewise.
13513 * cselib.c (cselib_reset_table): Likewise.
13514 (cselib_lookup_1): Likewise.
13515 * emit-rtl.c (set_mode_and_regno): Likewise.
13516 * function.c (aggregate_value_p): Likewise.
13517 * ira-color.c (setup_profitable_hard_regs): Likewise.
13518 (check_hard_reg_p): Likewise.
13519 (calculate_saved_nregs): Likewise.
13520 (assign_hard_reg): Likewise.
13521 (improve_allocation): Likewise.
13522 (calculate_spill_cost): Likewise.
13523 * ira-emit.c (modify_move_list): Likewise.
13524 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
13525 (ira_hard_reg_in_set_p): Likewise.
13526 * ira.c (setup_reg_mode_hard_regset): Likewise.
13527 (clarify_prohibited_class_mode_regs): Likewise.
13528 (check_allocation): Likewise.
13529 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13530 (lra_setup_reg_renumber): Likewise.
13531 (setup_try_hard_regno_pseudos): Likewise.
13532 (spill_for): Likewise.
13533 (assign_hard_regno): Likewise.
13534 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
13535 * lra-constraints.c (in_class_p): Likewise.
13536 (lra_constraint_offset): Likewise.
13537 (simplify_operand_subreg): Likewise.
13538 (lra_constraints): Likewise.
13539 (split_reg): Likewise.
13540 (split_if_necessary): Likewise.
13541 (invariant_p): Likewise.
13542 (inherit_in_ebb): Likewise.
13543 * lra-lives.c (process_bb_lives): Likewise.
13544 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
13545 (get_hard_regs): Likewise.
13546 (do_remat): Likewise.
13547 * lra-spills.c (assign_spill_hard_regs): Likewise.
13548 * mode-switching.c (create_pre_exit): Likewise.
13549 * postreload.c (reload_combine_recognize_pattern): Likewise.
13550 * recog.c (peep2_find_free_register): Likewise.
13551 * regcprop.c (kill_value_regno): Likewise.
13552 (set_value_regno): Likewise.
13553 (copy_value): Likewise.
13554 (maybe_mode_change): Likewise.
13555 (find_oldest_value_reg): Likewise.
13556 (copyprop_hardreg_forward_1): Likewise.
13557 * regrename.c (check_new_reg_p): Likewise.
13558 (regrename_do_replace): Likewise.
13559 * reload.c (push_reload): Likewise.
13560 (combine_reloads): Likewise.
13561 (find_dummy_reload): Likewise.
13562 (operands_match_p): Likewise.
13563 (find_reloads): Likewise.
13564 (find_equiv_reg): Likewise.
13565 (reload_adjust_reg_for_mode): Likewise.
13566 * reload1.c (count_pseudo): Likewise.
13567 (count_spilled_pseudo): Likewise.
13568 (find_reg): Likewise.
13569 (clear_reload_reg_in_use): Likewise.
13570 (free_for_value_p): Likewise.
13571 (allocate_reload_reg): Likewise.
13572 (choose_reload_regs): Likewise.
13573 (reload_adjust_reg_for_temp): Likewise.
13574 (emit_reload_insns): Likewise.
13575 (delete_output_reload): Likewise.
13576 * rtlanal.c (subreg_get_info): Likewise.
13577 * sched-deps.c (sched_analyze_reg): Likewise.
13578 * sel-sched.c (init_regs_for_mode): Likewise.
13579 (mark_unavailable_hard_regs): Likewise.
13580 (choose_best_reg_1): Likewise.
13581 (verify_target_availability): Likewise.
13582 * valtrack.c (dead_debug_insert_temp): Likewise.
13583 * var-tracking.c (track_loc_p): Likewise.
13584 (emit_note_insn_var_location): Likewise.
13585 * varasm.c (make_decl_rtl): Likewise.
13586 * reginfo.c (choose_hard_reg_mode): Likewise.
13587 (init_reg_modes_target): Refer directly to
13588 this_target_regs->x_hard_regno_nregs.
13590 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13592 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
13593 instead of hard_regno_nregs.
13595 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13597 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
13598 end_hard_regno instead of hard_regno_nregs.
13599 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
13600 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
13601 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
13602 * ira-color.c (improve_allocation): Likewise.
13603 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13604 * lra-lives.c (mark_regno_live): Likewise.
13605 (mark_regno_dead): Likewise.
13606 * lra-remat.c (operand_to_remat): Likewise.
13607 * lra.c (collect_non_operand_hard_regs): Likewise.
13608 * postreload.c (reload_combine_note_store): Likewise.
13609 (move2add_valid_value_p): Likewise.
13610 * reload.c (regno_clobbered_p): Likewise.
13612 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13614 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
13616 * config/v850/v850.c (v850_reorg): Likewise.
13617 * reload.c (refers_to_regno_for_reload_p): Likewise.
13618 (find_equiv_reg): Likewise.
13619 * reload1.c (reload_reg_reaches_end_p): Likewise.
13621 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13623 * caller-save.c (add_used_regs): Use REG_NREGS instead of
13625 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
13626 * config/arm/arm.c (output_move_neon): Likewise.
13627 (arm_attr_length_move_neon): Likewise.
13628 (neon_split_vcombine): Likewise.
13629 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
13630 (c6x_mark_reg_written): Likewise.
13631 (c6x_dwarf_register_span): Likewise.
13632 * config/i386/i386.c (ix86_save_reg): Likewise.
13633 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
13634 (rws_access_reg): Likewise.
13635 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
13636 * mode-switching.c (create_pre_exit): Likewise.
13637 * ree.c (combine_reaching_defs): Likewise.
13638 (add_removable_extension): Likewise.
13639 * regcprop.c (find_oldest_value_reg): Likewise.
13640 (copyprop_hardreg_forward_1): Likewise.
13641 * reload.c (reload_inner_reg_of_subreg): Likewise.
13642 (push_reload): Likewise.
13643 (combine_reloads): Likewise.
13644 (find_dummy_reload): Likewise.
13645 (reload_adjust_reg_for_mode): Likewise.
13646 * reload1.c (find_reload_regs): Likewise.
13647 (forget_old_reloads_1): Likewise.
13648 (reload_reg_free_for_value_p): Likewise.
13649 (reload_adjust_reg_for_temp): Likewise.
13650 (emit_reload_insns): Likewise.
13651 (delete_output_reload): Likewise.
13652 * sel-sched.c (choose_best_reg_1): Likewise.
13653 (choose_best_pseudo_reg): Likewise.
13655 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13656 Alan Hayward <alan.hayward@arm.com>
13657 David Sherwood <david.sherwood@arm.com>
13659 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
13660 * target.def (slow_unaligned_access): New hook.
13661 * targhooks.h (default_slow_unaligned_access): Declare.
13662 * targhooks.c (default_slow_unaligned_access): New function.
13663 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
13664 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
13665 * doc/tm.texi: Regenerate.
13666 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
13667 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
13668 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
13670 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
13671 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
13673 (rs6000_slow_unaligned_access): New function.
13674 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13675 (expand_block_compare): Likewise.
13676 (expand_strn_compare): Likewise.
13677 (rs6000_rtx_costs): Likewise.
13678 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
13679 (riscv_slow_unaligned_access): Likewise.
13680 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
13681 (riscv_slow_unaligned_access_p): ...this and make static.
13682 (riscv_option_override): Update accordingly.
13683 (riscv_slow_unaligned_access): New function.
13684 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13685 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
13686 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13687 (rs6000_slow_unaligned_access): New function.
13688 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13689 (rs6000_rtx_costs): Likewise.
13690 * config/rs6000/rs6000-string.c (expand_block_compare)
13691 (expand_strn_compare): Use targetm.slow_unaligned_access instead
13692 of SLOW_UNALIGNED_ACCESS.
13693 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
13694 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
13695 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
13696 of SLOW_UNALIGNED_ACCESS.
13697 * expmed.c (simple_mem_bitfield_p): Likewise.
13698 * expr.c (alignment_for_piecewise_move): Likewise.
13699 (emit_group_load_1): Likewise.
13700 (emit_group_store): Likewise.
13701 (copy_blkmode_from_reg): Likewise.
13702 (emit_push_insn): Likewise.
13703 (expand_assignment): Likewise.
13704 (store_field): Likewise.
13705 (expand_expr_real_1): Likewise.
13706 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
13707 * lra-constraints.c (simplify_operand_subreg): Likewise.
13708 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13709 * gimple-ssa-store-merging.c: Likewise in block comment at start
13711 * tree-ssa-strlen.c: Include target.h.
13712 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
13713 of SLOW_UNALIGNED_ACCESS.
13714 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
13716 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13718 PR rtl-optimization/82185
13719 * expmed.c (emit_store_flag_int): Only test tem if it has been
13722 2017-09-12 Richard Biener <rguenther@suse.de>
13724 PR middle-end/82149
13725 * match.pd ((FTYPE) N CMP CST): Fix typo.
13727 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
13729 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
13731 (mips_near_type_p): Add 'short_call' attribute as a synonym
13733 * doc/extend.texi (short_call): Document new function attribute.
13735 2017-09-12 Jakub Jelinek <jakub@redhat.com>
13738 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
13739 assertion check that in the condition.
13740 (get_atomic_generic_size): Likewise. Before testing if parameter
13741 has pointer type, if it has array type, call for C++
13742 default_conversion to perform array-to-pointer conversion.
13744 2017-09-12 Richard Biener <rguenther@suse.de>
13746 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
13747 for operations we cannot scalarize.
13749 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
13751 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
13752 vectors heap vectors. Clean up comments.
13753 Make visited_bbs a reference.
13754 (profitable_jump_thread_path): Make GC
13755 vectors heap vectors. Clean up comments.
13757 (convert_and_register_jump_thread_path): Make GC vectors heap
13759 (check_subpath_and_update_thread_path): Same. Clean up comments.
13760 Make visited_bbs a reference.
13761 (handle_phi): Abstract common code to to
13762 register_jump_thread_path_if_profitable.
13763 Rename VAR_BB to DEF_BB.
13765 Make GC vectors heap vectors.
13766 Make visited_bbs a reference.
13767 (handle_assignment): Same.
13768 (register_jump_thread_path_if_profitable): New.
13769 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
13771 Make GC vectors heap vectors. Clean up comments.
13772 Make visited_bbs a reference.
13773 (find_jump_threads_backwards): Make visited_bbs live in the stack.
13774 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
13777 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
13780 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
13781 words of E_DImode object are reachable by xtensa_uimm8x4 access.
13783 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
13785 Revert r251800 and r251799.
13787 2017-09-11 Martin Jambor <mjambor@suse.cz>
13790 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
13791 arguments in advance.
13792 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
13793 use it to find predecessor edges.
13794 (naive_outof_ssa): Collect vector of predecessors.
13796 2017-09-08 Jason Merrill <jason@redhat.com>
13798 PR c++/70029 - ICE with ref-qualifier and -flto
13799 * langhooks.h (struct lang_hooks_for_types): Add
13800 copy_lang_qualifiers.
13801 * attribs.c (build_type_attribute_qual_variant): Use it.
13802 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
13804 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
13805 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
13807 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
13810 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
13811 (*mulsi3_sp64): New instruction.
13812 (mulsi3): New expander.
13814 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
13816 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
13818 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13820 * sancov.c: Include memmodel.h.
13822 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
13825 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
13828 2017-09-07 Carl Love <cel@us.ibm.com>
13830 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
13831 the sldi instruction.
13833 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
13835 * sancov.c: Include tm_p.h.
13837 2017-09-07 Jakub Jelinek <jakub@redhat.com>
13840 * output.h (switch_to_other_text_partition): New declaration.
13841 * varasm.c (switch_to_other_text_partition): New function.
13842 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
13843 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
13844 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
13845 to the other text partition before emitting LCL label and switch back
13846 after emitting the word after it.
13848 2017-09-07 Richard Biener <rguenther@suse.de>
13850 * passes.def (pass_split_crit_edges): Remove instance before PRE.
13851 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
13852 critical edges here, after loop init.
13853 (pass_data_pre): Remove PROP_no_crit_edges flags.
13854 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
13855 for valueization of call args to avoid leaking VN_TOP.
13856 (visit_use): Assert we do not visit default defs.
13857 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
13858 Use error_mark_node to more easily detect leaking VN_TOP.
13859 All default-defs are varying, not VN_TOP. Mark them visited.
13860 (run_scc_vn): Make code match comment.
13862 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13864 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
13865 OPTION_MASK_FLOAT128_KEYWORD.
13866 (POWERPC_MASKS): Likewise.
13867 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
13868 support for the -mfloat128-type option, and make -mfloat128
13869 default on PowerPC Linux systems. Define or undefine
13870 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
13871 Define __float128 to be __ieee128 if IEEE 128-bit support is
13872 enabled, or undefine it.
13873 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
13874 Delete defining __FLOAT128_TYPE__.
13875 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
13876 -mfloat128-type option and make -mfloat128 default on PowerPC
13878 (TARGET_FLOAT128_TYPE): Likewise.
13879 (-mfloat128-type): Likewise.
13880 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13881 Delete the -mfloat128-type option and make -mfloat128 default on
13882 PowerPC Linux systems. Always use __ieee128 to be the keyword for
13883 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
13884 128-bit floating point is enabled. Change tests from using
13885 -mfloat128-type to -mfloat128.
13886 (rs6000_mangle_type): Use the correct mangling for the __float128
13887 type even if normal long double is restricted to 64-bits.
13888 (floatn_mode): Enable the _Float128 type by default on VSX Linux
13890 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
13891 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
13892 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
13894 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
13895 documentation for -mfloat128.
13897 2017-09-06 Olivier Hainque <hainque@adacore.com>
13899 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
13901 2017-09-06 Wish Wu <wishwu007@gmail.com>
13902 Jakub Jelinek <jakub@redhat.com>
13904 * asan.c (initialize_sanitizer_builtins): Add
13905 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
13906 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
13907 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
13908 BT_FN_VOID_UINT64_PTR variables.
13909 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
13910 (BT_FN_VOID_UINT16_UINT16): Likewise.
13911 (BT_FN_VOID_UINT32_UINT32): Likewise.
13912 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
13913 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
13914 (BT_FN_VOID_UINT64_PTR): Likewise.
13915 * common.opt (flag_sanitize_coverage): New variable.
13916 (fsanitize-coverage=trace-pc): Remove.
13917 (fsanitize-coverage=): Add.
13918 * flag-types.h (enum sanitize_coverage_code): New enum.
13919 * fold-const.c (fold_range_test): Disable non-short-circuit
13920 optimization if flag_sanitize_coverage.
13921 (fold_truth_andor): Likewise.
13922 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
13923 * opts.c (COVERAGE_SANITIZER_OPT): Define.
13924 (coverage_sanitizer_opts): New array.
13925 (get_closest_sanitizer_option): Add OPTS argument, handle also
13926 OPT_fsanitize_coverage_.
13927 (parse_sanitizer_options): Adjusted to also handle
13928 OPT_fsanitize_coverage_.
13929 (common_handle_option): Add OPT_fsanitize_coverage_.
13930 * sancov.c (instrument_comparison, instrument_switch): New function.
13931 (sancov_pass): Add trace-cmp support.
13932 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
13933 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
13934 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
13935 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
13936 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
13937 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
13938 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
13939 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
13940 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
13941 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
13943 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13945 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13946 error. Only quit immediately if parsing is complete.
13947 (BEGIN): Initialize fatal_err and parse_done.
13948 (begin fpu, end fpu): Check number of arguments.
13949 (begin arch, end arch): Likewise.
13950 (begin cpu, end cpu): Likewise.
13951 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13952 (optalias): Likewise.
13954 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13956 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13957 * config/arm/arm-isa.h: Delete. Move definitions to ...
13958 * arm-cpus.in: ... here. Use new feature and fgroup values.
13959 * config/arm/arm.c (arm_option_override): Use lower case for feature
13961 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13962 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13963 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13965 (print_isa_bits_for): New function.
13966 (gen_isa): New function.
13967 (gen_comm_data): Use print_isa_bits_for.
13968 (define feature): New keyword.
13969 (define fgroup): New keyword.
13970 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
13971 (arm-isa.h): Add rule to generate file.
13972 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13973 case for feature bit names.
13975 2017-09-06 Richard Biener <rguenther@suse.de>
13977 * tree-ssa-pre.c (NECESSARY): Remove.
13978 (create_expression_by_pieces): Do not touch pass-local flags.
13979 (insert_into_preds_of_block): Likewise.
13980 (do_pre_regular_insertion): Likewise.
13981 (eliminate_insert): Likewise.
13982 (eliminate_dom_walker::before_dom_children): Likewise.
13983 (fini_eliminate): Do not look at inserted_exprs.
13984 (mark_operand_necessary): Remove.
13985 (remove_dead_inserted_code): Replace with simple work-list
13986 algorithm based on inserted_exprs and SSA uses.
13987 (pass_pre::execute): Re-order fini_eliminate and
13988 remove_dead_inserted_code.
13990 2017-09-06 Olivier Hainque <hainque@adacore.com>
13992 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
13993 for VxWorks 7. Adjust surrounding comments.
13995 2017-09-06 Richard Biener <rguenther@suse.de>
13997 * gimple-ssa-strength-reduction.c
13998 (find_candidates_dom_walker::before_dom_children): Also allow
14001 2017-09-06 Richard Biener <rguenther@suse.de>
14003 PR tree-optimization/82108
14004 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
14005 for gap in the non-permutation SLP case.
14007 2017-09-06 Martin Jambor <mjambor@suse.cz>
14009 PR tree-optimization/82078
14010 * tree-sra.c (sort_and_splice_var_accesses): Move call to
14011 add_access_to_work_queue...
14012 (build_accesses_from_assign): ...here.
14013 (propagate_all_subaccesses): Make sure racc is the group
14014 representative, if there is one.
14016 2017-09-06 Jakub Jelinek <jakub@redhat.com>
14018 PR middle-end/82095
14019 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
14022 2017-09-06 Richard Biener <rguenther@suse.de>
14024 * gimple-ssa-strength-reduction.c
14025 (find_candidates_dom_walker::before_doom_children): Use a
14026 type and not a mode check.
14028 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
14031 * config/arm/predicates.md (arm_general_adddi_operand): Create new
14033 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
14035 2017-09-05 Jeff Law <law@redhat.com>
14037 PR tree-optimization/64910
14038 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
14039 cases where we have 3 or more operands.
14041 2017-09-05 Jakub Jelinek <jakub@redhat.com>
14043 PR middle-end/81768
14044 * omp-low.c (lower_omp_for): Recompute tree invariant if
14045 gimple_omp_for_initial/final is ADDR_EXPR.
14047 PR middle-end/81768
14048 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
14049 into gimple val before gimplification fo the COND_EXPR.
14051 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
14053 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
14054 REGION_COPY argument.
14055 (thread_through_all_blocks): Remove unused argument to
14056 duplicate_thread_path.
14058 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14059 Alan Hayward <alan.hayward@arm.com>
14060 David Sherwood <david.sherwood@arm.com>
14062 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
14063 Take a scalar_mode rather than a machine_mode.
14064 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14065 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
14066 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14067 (aarch64_gen_adjusted_ldpstp): Likewise.
14068 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
14070 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14071 Alan Hayward <alan.hayward@arm.com>
14072 David Sherwood <david.sherwood@arm.com>
14074 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
14075 Take a scalar_int_mode instead of a machine_mode.
14076 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
14077 (aarch64_output_scalar_simd_mov_immediate): Likewise.
14078 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
14079 (aarch64_simd_attr_length_rglist): Delete.
14080 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
14081 a scalar_int_mode instead of a machine_mode.
14082 (aarch64_add_offset): Likewise.
14083 (aarch64_internal_mov_immediate): Likewise
14084 (aarch64_add_constant_internal): Likewise.
14085 (aarch64_add_constant): Likewise.
14086 (aarch64_movw_imm): Likewise.
14087 (aarch64_rtx_arith_op_extract_p): Likewise.
14088 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
14089 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
14090 Remove assert that the mode isn't a vector.
14091 (aarch64_output_scalar_simd_mov_immediate): Likewise.
14092 (aarch64_expand_mov_immediate): Update calls after above changes.
14093 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
14094 (aarch64_and_bitmask_imm): Check for scalar integer modes.
14095 (aarch64_move_imm): Likewise.
14096 (aarch64_can_const_movi_rtx_p): Likewise.
14097 (aarch64_strip_extend): Likewise.
14098 (aarch64_extr_rtx_p): Likewise.
14099 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
14100 a CONST_INT when the mode parameter is VOIDmode.
14101 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
14103 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14105 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
14106 * stor-layout.c (bitwise_mode_for_mode): Likewise.
14107 (bitwise_type_for_mode): Update accordingly.
14109 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14111 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
14112 * stor-layout.c (mode_for_size_tree): Likewise.
14113 (mode_for_array): Update accordingly.
14114 (layout_decl): Likewise.
14115 (compute_record_mode): Likewise. Only set the mode once.
14117 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14119 * target.def (get_mask_mode): Change return type to opt_mode.
14121 * doc/tm.texi: Regenerate.
14122 * targhooks.h (default_get_mask_mode): Return an opt_mode.
14123 * targhooks.c (default_get_mask_mode): Likewise.
14124 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
14125 * optabs-query.c (can_vec_mask_load_store_p): Update use of
14126 targetm.get_mask_mode.
14127 * tree.c (build_truth_vector_type): Likewise.
14129 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14131 * machmode.h (mode_for_vector): Return an opt_mode.
14132 * stor-layout.c (mode_for_vector): Likewise.
14133 (mode_for_int_vector): Update accordingly.
14134 (layout_type): Likewise.
14135 * config/i386/i386.c (emit_memmov): Likewise.
14136 (ix86_expand_set_or_movmem): Likewise.
14137 (ix86_expand_vector_init): Likewise.
14138 (ix86_get_mask_mode): Likewise.
14139 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
14141 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
14142 * expmed.c (extract_bit_field_1): Likewise.
14143 * expr.c (expand_expr_real_2): Likewise.
14144 * optabs-query.c (can_vec_perm_p): Likewise.
14145 (can_vec_mask_load_store_p): Likewise.
14146 * optabs.c (expand_vec_perm): Likewise.
14147 * targhooks.c (default_get_mask_mode): Likewise.
14148 * tree-vect-stmts.c (vectorizable_store): Likewise.
14149 (vectorizable_load): Likewise.
14150 (get_vectype_for_scalar_type_and_size): Likewise.
14152 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14154 * machmode.h (mode_for_int_vector): New function.
14155 * stor-layout.c (mode_for_int_vector): Likewise.
14156 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
14157 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
14158 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
14159 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
14160 (s390_expand_vcond): Likewise.
14162 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14164 * machmode.h (opt_machine_mode): New type.
14165 (opt_mode<T>): Allow construction from anything that can be
14167 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
14168 (mode_for_size): Return an opt_machine_mode.
14169 * stor-layout.c (mode_for_size): Likewise.
14170 (mode_for_size_tree): Update call accordingly.
14171 (bitwise_mode_for_mode): Likewise.
14172 (make_fract_type): Likewise.
14173 (make_accum_type): Likewise.
14174 * caller-save.c (replace_reg_with_saved_mem): Update call
14176 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14177 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14178 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14179 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14180 * expmed.c (extract_bit_field_1): Likewise.
14181 * reload.c (get_secondary_mem): Likewise.
14182 * varasm.c (assemble_integer): Likewise.
14183 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
14186 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14188 * machmode.h (decimal_float_mode_for_size): New function.
14189 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
14190 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
14191 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
14192 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
14193 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
14194 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
14196 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14198 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
14199 (get_builtin_sync_mode): Likewise.
14200 (expand_ifn_atomic_compare_exchange): Likewise.
14201 (expand_builtin_atomic_clear): Likewise.
14202 (expand_builtin_atomic_test_and_set): Likewise.
14203 (fold_builtin_atomic_always_lock_free): Likewise.
14204 * calls.c (compute_argument_addresses): Likewise.
14205 (emit_library_call_value_1): Likewise.
14206 (store_one_arg): Likewise.
14207 * combine.c (combine_instructions): Likewise.
14208 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
14209 * config/arm/arm.c (arm_function_value): Likewise.
14210 (aapcs_allocate_return_reg): Likewise.
14211 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
14212 * config/i386/i386.c (construct_container): Likewise.
14213 (ix86_gimplify_va_arg): Likewise.
14214 (ix86_expand_sse_cmp): Likewise.
14215 (emit_memmov): Likewise.
14216 (emit_memset): Likewise.
14217 (expand_small_movmem_or_setmem): Likewise.
14218 (ix86_expand_pextr): Likewise.
14219 (ix86_expand_pinsr): Likewise.
14220 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
14221 * config/microblaze/microblaze.c (microblaze_block_move_straight):
14223 * config/mips/mips.c (mips_function_value_1) Likewise.
14224 (mips_block_move_straight): Likewise.
14225 (mips_expand_ins_as_unaligned_store): Likewise.
14226 * config/powerpcspe/powerpcspe.c
14227 (rs6000_darwin64_record_arg_advance_flush): Likewise.
14228 (rs6000_darwin64_record_arg_flush): Likewise.
14229 * config/rs6000/rs6000.c
14230 (rs6000_darwin64_record_arg_advance_flush): Likewise.
14231 (rs6000_darwin64_record_arg_flush): Likewise.
14232 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
14233 (sparc_function_value_1): Likewise.
14234 * config/spu/spu.c (adjust_operand): Likewise.
14235 (spu_emit_branch_or_set): Likewise.
14236 (arith_immediate_p): Likewise.
14237 * emit-rtl.c (gen_lowpart_common): Likewise.
14238 * expr.c (expand_expr_real_1): Likewise.
14239 * function.c (assign_parm_setup_block): Likewise.
14240 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
14241 * reload1.c (alter_reg): Likewise.
14242 * stor-layout.c (mode_for_vector): Likewise.
14243 (layout_type): Likewise.
14245 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14247 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
14248 (spu_convert_move): Likewise.
14249 * lower-subreg.c (resolve_simple_move): Likewise.
14251 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14254 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
14255 define_insn to a define_expand.
14256 (altivec_vsum2sws_direct): New define_insn.
14257 (altivec_vsumsws): Convert from a define_insn to a define_expand.
14259 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
14261 * config/arm/arm.c (arm_option_params_internal): Improve setting of
14264 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
14269 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
14270 realign stack if stack alignment needed is less than incoming
14273 2017-09-05 Marek Polacek <polacek@redhat.com>
14276 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
14279 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
14281 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
14283 2017-09-05 Richard Biener <rguenther@suse.de>
14285 PR tree-optimization/82084
14286 * fold-const.c (can_native_encode_string_p): Handle wide characters.
14288 2017-09-05 Richard Biener <rguenther@suse.de>
14290 PR tree-optimization/82102
14291 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
14293 2017-09-05 Martin Liska <mliska@suse.cz>
14295 PR tree-optimization/82032
14296 * tree-cfg.c (generate_range_test): New function.
14297 * tree-cfg.h (generate_range_test): Declared here.
14298 * tree-cfgcleanup.c (convert_single_case_switch): New function.
14299 (cleanup_control_expr_graph): Use it.
14300 * tree-switch-conversion.c (try_switch_expansion): Remove
14302 (emit_case_nodes): Use generate_range_test.
14304 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
14307 * config/i386/i386.md (*<btsc><mode>_mask): Add
14308 TARGET_USE_BT to insn constraint.
14309 (*btr<mode>_mask): Ditto.
14311 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
14313 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
14314 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
14316 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
14319 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
14320 TARGET_NEON and TARGET_IWMMXT.
14321 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
14322 TARGET_NEON and TARGET_IWMMXT.
14323 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
14325 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
14327 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
14328 (ix86_rewrite_tls_address): Ditto.
14329 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
14330 (ix86_rewrite_tls_address_1): Ditto.
14331 (ix86_rewrite_tls_address): Ditto.
14332 * config/i386/predicates.md (tls_address_pattern): New predicate.
14333 * config/i386/i386.md (TLS address splitter): New splitter.
14335 2017-09-04 Richard Biener <rguenther@suse.de>
14337 PR tree-optimization/82084
14338 * fold-const.h (can_native_encode_string_p): Declare.
14339 * fold-const.c (can_native_encode_string_p): Factor out from ...
14340 (native_encode_string): ... here.
14341 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
14342 vectorizing stores from constants we later cannot handle.
14344 2017-09-04 Marek Polacek <polacek@redhat.com>
14347 * doc/invoke.texi: Update -Wtautological-compare documentation.
14349 2017-09-04 Jeff Law <law@redhat.com>
14351 PR tree-optimization/64910
14352 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
14353 swap the first and last operand if the last is a constant.
14355 2017-09-04 Marek Polacek <polacek@redhat.com>
14358 * convert.c (do_narrow): When sanitizing signed integer overflows,
14359 bail out for signed types.
14360 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
14362 2017-09-04 Richard Biener <rguenther@suse.de>
14364 PR tree-optimization/82060
14365 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
14366 Move devirtualization after stmt folding and before EH/AB/noreturn
14367 cleanup to get the stmt refs canonicalized. Use a bool instead
14368 of gimple_modified_p since that doesn't work for NOPs. Schedule
14369 NOPs generated by folding for removal.
14371 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14372 Alan Hayward <alan.hayward@arm.com>
14373 David Sherwood <david.sherwood@arm.com>
14375 * coretypes.h (pad_direction): New enum.
14376 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
14377 (FUNCTION_ARG_PADDING): Likewise.
14378 * target.def (function_arg_padding): New hook.
14379 * targhooks.h (default_function_arg_padding): Declare.
14380 * targhooks.c (default_function_arg_padding): New function.
14381 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
14382 (TARGET_FUNCTION_ARG_PADDING): ...this.
14383 * doc/tm.texi: Regenerate.
14384 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
14385 instead of direction.
14386 (compute_argument_addresses): Likewise.
14387 (load_register_parameters): Likewise.
14388 (emit_library_call_value_1): Likewise.
14389 (store_one_arg): Use targetm.calls.function_arg_padding instead
14390 of FUNCTION_ARG_PADDING.
14391 (must_pass_in_stack_var_size_or_pad): Likewise.
14392 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
14393 (emit_group_store): Likewise.
14394 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
14395 instead of FUNCTION_ARG_PADDING.
14396 (emit_push_insn): Likewise, and propagate enum change throughout
14398 * function.h (direction): Delete.
14399 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
14401 * function.c (assign_parm_find_stack_rtl): Likewise.
14402 (assign_parm_setup_block_p): Likewise.
14403 (assign_parm_setup_block): Likewise.
14404 (gimplify_parameters): Likewise.
14405 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
14406 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
14408 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
14409 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
14410 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
14411 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
14412 (aarch64_function_arg_padding): ...this new function.
14413 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
14414 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14415 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
14416 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
14417 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
14418 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14419 (arm_pad_arg_upward): Replace with...
14420 (arm_function_arg_padding): ...this new function.
14421 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
14423 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
14424 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
14425 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14426 (ia64_hpux_function_arg_padding): Replace with...
14427 (ia64_function_arg_padding): ...this new function. Use pad_direction
14428 instead of direction. Check for TARGET_HPUX.
14429 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
14430 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14431 (iq2000_function_arg_padding): New function.
14432 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
14433 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
14434 (mips_function_arg_padding): ...this new function.
14435 (mips_pad_reg_upward): Update accordingly.
14436 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14437 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
14438 targetm.calls.function_arg_padding.
14439 (FUNCTION_ARG_PADDING): Delete.
14440 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
14441 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
14442 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
14443 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
14444 (nios2_block_reg_padding): Return pad_direction instead of direction.
14445 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
14446 instead of direction.
14447 (nios2_function_arg_padding): Likewise. Make static.
14448 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14449 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
14450 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
14451 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
14452 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14453 (pa_function_arg_padding): Make static. Return pad_direction instead
14455 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
14456 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
14457 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
14458 instead of direction. Use targetm.calls.function_arg_padding.
14459 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
14460 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
14461 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
14462 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
14463 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
14465 (function_arg_padding): Rename to...
14466 (rs6000_function_arg_padding): ...this. Make static. Return
14467 pad_direction instead of direction.
14468 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
14469 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
14470 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
14471 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
14472 instead of direction. Use targetm.calls.function_arg_padding.
14473 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
14474 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
14475 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
14476 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
14477 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14478 (function_arg_padding): Rename to...
14479 (rs6000_function_arg_padding): ...this. Make static. Return
14480 pad_direction instead of direction.
14481 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
14482 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
14483 * config/s390/s390.c (s390_function_arg_padding): New function.
14484 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14485 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
14486 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
14487 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14488 (function_arg_padding): Rename to...
14489 (sparc_function_arg_padding): ...this. Make static. Return
14490 pad_direction instead of direction.
14491 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
14492 * config/spu/spu.c (spu_function_arg_padding): New function.
14493 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14494 * system.h (FUNCTION_ARG_PADDING): Poison.
14496 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14497 Alan Hayward <alan.hayward@arm.com>
14498 David Sherwood <david.sherwood@arm.com>
14500 * target.def (modes_tieable_p): New hook.
14501 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
14502 (TARGET_MODES_TIEABLE_P): ...this.
14503 * doc/tm.texi.in: Regenerate.
14504 * hooks.h (hook_bool_mode_mode_true): Declare.
14505 * hooks.c (hook_bool_mode_mode_true): New function.
14506 * combine.c (subst): Use targetm.modes_tieable_p instead of
14508 * dse.c (find_shift_sequence): Likewise.
14509 * expmed.c (extract_low_bits): Likewise.
14510 * lower-subreg.c: Include target.h.
14511 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
14513 * rtlanal.c (rtx_cost): Likewise.
14514 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
14515 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
14516 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
14517 (TARGET_MODES_TIEABLE_P): Redefine.
14518 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
14519 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
14520 (TARGET_MODES_TIEABLE_P): Redefine.
14521 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
14522 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
14523 (arc_modes_tieable_p): New function.
14524 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
14525 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
14526 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
14527 (arm_modes_tieable_p): Make static.
14528 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
14529 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
14530 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
14531 (TARGET_MODES_TIEABLE_P): Redefine.
14532 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
14533 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
14534 (TARGET_MODES_TIEABLE_P): Redefine.
14535 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
14536 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
14537 (cr16_modes_tieable_p): New function.
14538 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
14539 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
14540 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
14541 (TRULY_NOOP_TRUNCATION): Update comment.
14542 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
14543 (TRULY_NOOP_TRUNCATION): Update comment.
14544 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
14545 (frv_modes_tieable_p): New function.
14546 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
14547 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
14548 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
14549 (TARGET_MODES_TIEABLE_P): Redefine.
14550 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
14551 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
14552 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
14553 (TARGET_MODES_TIEABLE_P): Redefine.
14554 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
14555 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
14556 (ia64_modes_tieable_p): New function.
14557 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
14558 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
14559 (iq2000_modes_tieable_p): New function.
14560 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
14561 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
14562 (lm32_modes_tieable_p): New function.
14563 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
14564 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
14565 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
14566 (TARGET_MODES_TIEABLE_P): Redefine.
14567 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
14568 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
14569 (m32r_modes_tieable_p): New function.
14570 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
14571 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
14572 (m68k_modes_tieable_p): New function.
14573 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
14574 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
14575 (mcore_modes_tieable_p): New function.
14576 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
14577 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
14579 (TARGET_MODES_TIEABLE_P): Redefine.
14580 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
14581 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
14582 * config/mips/mips.c (mips_modes_tieable_p): Make static.
14583 (TARGET_MODES_TIEABLE_P): Redefine.
14584 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
14585 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
14586 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
14587 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
14588 (mn10300_modes_tieable_p): ...this and make static.
14589 (TARGET_MODES_TIEABLE_P): Redefine.
14590 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
14591 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
14592 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
14593 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
14594 (msp430_modes_tieable_p): Make static.
14595 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
14596 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
14597 (TARGET_MODES_TIEABLE_P): Redefine.
14598 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
14599 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
14600 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
14601 (TARGET_MODES_TIEABLE_P): Redefine.
14602 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
14603 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
14604 * config/pa/pa.c (pa_modes_tieable_p): Make static.
14605 (TARGET_MODES_TIEABLE_P): Redefine.
14606 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
14607 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
14608 (pdp11_modes_tieable_p): New function.
14609 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
14610 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
14611 (rs6000_modes_tieable_p): New function.
14612 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
14613 * config/powerpcspe/powerpcspe.md: Update comment.
14614 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
14615 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
14616 (TARGET_MODES_TIEABLE_P): Redefine.
14617 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
14618 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
14619 (rl78_modes_tieable_p): New function.
14620 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
14621 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
14622 (rs6000_modes_tieable_p): New function.
14623 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
14624 * config/rs6000/rs6000.md: Update comment.
14625 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
14626 * config/rx/rx.c (rx_modes_tieable_p): New function.
14627 (TARGET_MODES_TIEABLE_P): Redefine.
14628 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
14629 * config/s390/s390.c (s390_modes_tieable_p): New function.
14630 (TARGET_MODES_TIEABLE_P): Redefine.
14631 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
14632 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
14633 (sh_modes_tieable_p): New function.
14634 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
14635 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
14636 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
14637 (sparc_modes_tieable_p): Make static.
14638 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
14639 * config/spu/spu.c (spu_modes_tieable_p): New function.
14640 (TARGET_MODES_TIEABLE_P): Redefine.
14641 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
14642 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
14643 (TARGET_MODES_TIEABLE_P): Redefine.
14644 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
14645 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
14646 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
14647 * config/v850/v850.c (v850_modes_tieable_p): New function.
14648 (TARGET_MODES_TIEABLE_P): Redefine.
14649 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
14650 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
14651 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
14652 (visium_modes_tieable_p): New function.
14653 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
14654 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
14655 (xtensa_modes_tieable_p): New function.
14656 * system.h (MODES_TIEABLE_P): Poison.
14658 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14659 Alan Hayward <alan.hayward@arm.com>
14660 David Sherwood <david.sherwood@arm.com>
14662 * target.def (hard_regno_mode_ok): New hook.
14663 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
14664 (TARGET_HARD_REGNO_MODE_OK): ...this.
14665 * doc/tm.texi.in: Regenerate.
14666 * hooks.h (hook_bool_uint_mode_true): Declare.
14667 * hooks.c (hook_bool_uint_mode_true): New function.
14668 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
14669 HARD_REGNO_MODE_OK.
14670 * genpreds.c (write_insn_preds_c): Add an include of target.h.
14671 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
14672 instead of HARD_REGNO_MODE_OK.
14673 * caller-save.c: Include target.h.
14674 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
14675 HARD_REGNO_MODE_OK.
14676 * combine.c (can_combine_p): Likewise.
14677 (combinable_i3pat): Likewise.
14678 (can_change_dest_mode): Likewise.
14679 * expr.c (init_expr_target): Likewise.
14680 (convert_move): Likewise.
14681 (convert_modes): Likewise.
14682 * ira.c (setup_prohibited_class_mode_regs): Likewise.
14683 (setup_prohibited_mode_move_regs): Likewise.
14684 * ira.h (target_ira): Likewise.
14685 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14686 * lra-constraints.c (process_alt_operands): Likewise.
14687 (split_reg): Likewise.
14688 * recog.c (peep2_find_free_register): Likewise.
14689 * ree.c (combine_reaching_defs): Likewise.
14690 * regcprop.c (maybe_mode_change): Likewise.
14691 * reginfo.c (init_reg_sets_1): Likewise.
14692 (choose_hard_reg_mode): Likewise.
14693 (simplifiable_subregs): Likewise.
14694 * regrename.c (check_new_reg_p): Likewise.
14695 * reload.c (find_valid_class): Likewise.
14696 (find_valid_class_1): Likewise.
14697 (reload_inner_reg_of_subreg): Likewise.
14698 (push_reload): Likewise.
14699 (combine_reloads): Likewise.
14700 (find_dummy_reload): Likewise.
14701 (find_reloads): Likewise.
14702 * reload1.c (find_reg): Likewise.
14703 (set_reload_reg): Likewise.
14704 (allocate_reload_reg): Likewise.
14705 (choose_reload_regs): Likewise.
14706 (reload_adjust_reg_for_temp): Likewise.
14707 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
14708 (simplify_subreg_regno): Likewise.
14709 * sel-sched.c (init_regs_for_mode): Likewise.
14710 * varasm.c (make_decl_rtl): Likewise.
14711 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
14712 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
14713 HARD_REGNO_MODE_OK.
14714 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
14715 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
14716 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14717 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
14718 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
14719 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14720 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
14721 (arc_mode_class): Delete.
14722 (HARD_REGNO_MODE_OK): Delete.
14723 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14724 (arc_hard_regno_mode_ok): Rename old array to...
14725 (arc_hard_regno_mode_ok_modes): ...this.
14726 (arc_conditional_register_usage): Update accordingly.
14727 (arc_mode_class): Make static.
14728 (arc_hard_regno_mode_ok): New function.
14729 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
14730 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
14731 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14732 (arm_hard_regno_mode_ok): Make static.
14733 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
14734 HARD_REGNO_MODE_OK.
14735 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
14736 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
14737 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
14739 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14740 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
14741 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
14742 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
14743 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
14744 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14745 * config/bfin/predicates.md (valid_reg_operand): Use
14746 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
14747 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
14748 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
14749 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14750 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
14751 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
14752 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14753 (cr16_hard_regno_mode_ok): Make static and return a bool.
14754 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
14755 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14756 (cris_hard_regno_mode_ok): New function.
14757 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
14758 (epiphany_mode_class): Delete.
14759 (HARD_REGNO_MODE_OK): Delete.
14760 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
14761 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14762 (hard_regno_mode_ok): Rename to...
14763 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
14764 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
14765 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
14766 HARD_REGNO_MODE_OK.
14767 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
14768 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
14769 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14770 (frv_hard_regno_mode_ok): Make static and return a bool.
14771 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
14772 HARD_REGNO_MODE_OK.
14773 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
14774 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
14775 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
14776 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
14778 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14779 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
14780 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
14781 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
14783 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14784 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
14785 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14786 (ia64_hard_regno_mode_ok): New function.
14787 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
14788 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14789 (iq2000_hard_regno_mode_ok): New function.
14790 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
14791 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14792 (lm32_hard_regno_mode_ok): New function.
14793 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
14794 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
14795 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
14796 instead of HARD_REGNO_MODE_OK.
14797 (m32c_hard_regno_ok): Rename to...
14798 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
14799 (m32c_cannot_change_mode_class): Update accordingly.
14800 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14801 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
14802 (m32r_mode_class): Delete.
14803 (HARD_REGNO_MODE_OK): Delete.
14804 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14805 (m32r_hard_regno_mode_ok): Rename to...
14806 (m32r_hard_regno_modes): ...this.
14807 (m32r_mode_class): Make static.
14808 (m32r_hard_regno_mode_ok): New function.
14809 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
14810 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
14811 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14812 (m68k_hard_regno_mode_ok): Make static.
14813 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
14814 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14815 (mcore_hard_regno_mode_ok): New function.
14816 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
14817 (HARD_REGNO_MODE_OK): Delete.
14818 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
14820 (microblaze_hard_regno_mode_ok_p): ...this and make static.
14821 (microblaze_hard_regno_mode_ok): New function.
14822 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14823 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
14824 (mips_hard_regno_mode_ok): Delete.
14825 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
14826 (mips_hard_regno_mode_ok_p): ...this and make static.
14827 (mips_hard_regno_mode_ok_p): Rename to...
14828 (mips_hard_regno_mode_ok_uncached): ...this.
14829 (mips_hard_regno_mode_ok): New function.
14830 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
14831 of HARD_REGNO_MODE_OK.
14832 (mips_option_override): Update after above name changes.
14833 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14834 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
14835 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
14836 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
14837 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
14838 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14839 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
14840 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
14841 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
14842 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14843 (msp430_hard_regno_mode_ok): Make static and return a bool.
14844 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
14845 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
14846 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
14848 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14849 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
14850 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
14851 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
14852 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
14853 (PA_HARD_REGNO_MODE_OK): ...this
14854 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
14855 (PA_HARD_REGNO_MODE_OK): ...this.
14856 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14857 (pa_hard_regno_mode_ok): New function.
14858 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
14859 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14860 (pdp11_hard_regno_mode_ok): New function.
14861 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
14862 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
14864 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
14866 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14867 (rs6000_hard_regno_mode_ok): Rename to...
14868 (rs6000_hard_regno_mode_ok_uncached): ...this.
14869 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14870 (rs6000_hard_regno_mode_ok): New function.
14871 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
14872 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
14873 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
14874 (riscv_hard_regno_mode_ok): ...this and make static.
14875 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14876 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
14877 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
14878 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14879 (rl78_hard_regno_mode_ok): Make static and return bool.
14880 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
14881 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
14883 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
14884 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14885 (rs6000_hard_regno_mode_ok): Rename to...
14886 (rs6000_hard_regno_mode_ok_uncached): ...this.
14887 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14888 (rs6000_hard_regno_mode_ok): New function.
14889 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
14890 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
14891 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14892 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
14893 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
14894 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
14895 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14896 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
14897 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
14898 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14899 (sh_hard_regno_mode_ok): Make static.
14900 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
14901 instead of HARD_REGNO_MODE_OK.
14902 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
14903 (sparc_mode_class): Delete.
14904 (HARD_REGNO_MODE_OK): Delete.
14905 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14906 (hard_regno_mode_classes): Make static.
14907 (sparc_mode_class): Likewise.
14908 (sparc_hard_regno_mode_ok): New function.
14909 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
14910 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
14911 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
14913 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14914 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
14915 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
14916 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
14917 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
14918 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14919 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
14920 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
14921 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14922 (visium_hard_regno_mode_ok): New function.
14923 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
14924 instead of HARD_REGNO_MODE_OK.
14925 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
14926 (HARD_REGNO_MODE_OK): Delete.
14927 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
14928 (xtensa_hard_regno_mode_ok_p): ...this and make static.
14929 (xtensa_option_override): Update accordingly.
14930 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14931 (xtensa_hard_regno_mode_ok): New function.
14932 * system.h (HARD_REGNO_MODE_OK): Poison.
14934 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14935 Alan Hayward <alan.hayward@arm.com>
14936 David Sherwood <david.sherwood@arm.com>
14938 * target.def (hard_regno_call_part_clobbered): New hook.
14939 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
14940 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
14941 * doc/tm.texi: Regenerate.
14942 * hooks.h (hook_bool_uint_mode_false): Declare.
14943 * hooks.c (hook_bool_uint_mode_false): New function.
14944 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14945 * cselib.c (cselib_process_insn): Use
14946 targetm.hard_regno_call_part_clobbered instead of
14947 HARD_REGNO_CALL_PART_CLOBBERED.
14948 * ira-conflicts.c (ira_build_conflicts): Likewise.
14949 * ira-costs.c (ira_tune_allocno_costs): Likewise.
14950 * lra-constraints.c (need_for_call_save_p): Likewise.
14951 * lra-lives.c: Include target.h.
14952 (check_pseudos_live_through_calls): Use
14953 targetm.hard_regno_call_part_clobbered instead of
14954 HARD_REGNO_CALL_PART_CLOBBERED.
14955 * regcprop.c: Include target.h.
14956 (copyprop_hardreg_forward_1): Use
14957 targetm.hard_regno_call_part_clobbered instead of
14958 HARD_REGNO_CALL_PART_CLOBBERED.
14959 * reginfo.c (choose_hard_reg_mode): Likewise.
14960 * regrename.c (check_new_reg_p): Likewise.
14961 * reload.c (find_equiv_reg): Likewise.
14962 * reload1.c (emit_reload_insns): Likewise.
14963 * sched-deps.c (deps_analyze_insn): Likewise.
14964 * sel-sched.c (init_regs_for_mode): Likewise.
14965 (mark_unavailable_hard_regs): Likewise.
14966 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
14967 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14968 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
14970 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14971 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14972 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
14974 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
14976 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14977 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14978 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
14980 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14981 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14982 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
14984 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14985 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
14987 * config/powerpcspe/powerpcspe.c
14988 (rs6000_hard_regno_call_part_clobbered): New function.
14989 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14990 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14991 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
14993 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14994 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14995 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
14997 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14998 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14999 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
15001 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15002 Alan Hayward <alan.hayward@arm.com>
15003 David Sherwood <david.sherwood@arm.com>
15005 * rtl.h (subreg_memory_offset): Declare.
15006 * emit-rtl.c (subreg_memory_offset): New function.
15007 * expmed.c (store_bit_field_1): Use it.
15008 * expr.c (undefined_operand_subword_p): Likewise.
15009 * simplify-rtx.c (simplify_subreg): Likewise.
15011 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
15013 PR rtl-optimization/57448
15016 * optabs.c (expand_atomic_load): Place compiler memory barriers if
15017 using atomic_load pattern.
15018 (expand_atomic_store): Likewise.
15020 2017-09-04 Jakub Jelinek <jakub@redhat.com>
15023 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
15024 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
15025 handling. Use replace_call_with_value with NULL instead of
15026 gsi_replace, unlink_stmt_vdef and release_defs.
15028 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
15031 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
15033 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15036 * rtl.h (emit_library_call_value_1): Declare.
15037 (emit_library_call): Replace declaration with a series of overloads.
15038 Remove the parameter count argument.
15039 (emit_library_call_value): Likewise.
15040 * calls.c (emit_library_call_value_1): Make global. Replace varargs
15041 with an "rtx_mode_t *".
15042 (emit_library_call_value): Delete.
15043 (emit_library_call): Likewise.
15044 * asan.c (asan_emit_stack_protection): Update calls accordingly.
15045 (asan_emit_allocas_unpoison): Likewise.
15046 * builtins.c (expand_builtin_powi): Likewise.
15047 (expand_asan_emit_allocas_unpoison): Likewise.
15048 * cfgexpand.c (expand_main_function): Likewise.
15049 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
15050 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
15051 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
15052 * config/arm/arm.c (arm_trampoline_init): Likewise.
15053 (arm_call_tls_get_addr): Likewise.
15054 (arm_expand_divmod_libfunc): Likewise.
15055 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
15056 (smulsi3_highpart): Likewise.
15057 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
15058 (c6x_expand_compare): Likewise.
15059 (c6x_expand_movmem): Likewise.
15060 * config/frv/frv.c (frv_trampoline_init): Likewise.
15061 * config/i386/i386.c (ix86_trampoline_init): Likewise.
15062 (ix86_expand_divmod_libfunc): Likewise.
15063 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
15064 (ia64_expand_compare): Likewise.
15065 (ia64_profile_hook): Likewise.
15066 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
15067 (nonlocal_goto): Likewise.
15068 (restore_stack_nonlocal): Likewise.
15069 * config/m32r/m32r.c (block_move_call): Likewise.
15070 (m32r_trampoline_init): Likewise.
15071 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
15072 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
15073 (m68k_call_m68k_read_tp): Likewise.
15074 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
15075 (microblaze_expand_divide): Likewise.
15076 * config/mips/mips.h (mips_args): Likewise.
15077 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
15078 (MIPS_ICACHE_SYNC): Likewise.
15079 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
15080 (nios2_trampoline_init): Likewise.
15081 * config/pa/pa.c (hppa_tls_call): Likewise.
15082 (pa_trampoline_init): Likewise.
15083 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
15084 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
15085 (expand_strn_compare): Likewise.
15086 (rs6000_generate_compare): Likewise.
15087 (rs6000_expand_float128_convert): Likewise.
15088 (output_profile_hook): Likewise.
15089 (rs6000_trampoline_init): Likewise.
15090 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
15091 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
15092 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
15093 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
15094 (rs6000_generate_compare): Likewise.
15095 (rs6000_expand_float128_convert): Likewise.
15096 (output_profile_hook): Likewise.
15097 (rs6000_trampoline_init): Likewise.
15098 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
15099 * config/sh/sh.c (sh_trampoline_init): Likewise.
15100 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
15101 (sparc_emit_float_lib_cmp): Likewise.
15102 (sparc32_initialize_trampoline): Likewise.
15103 (sparc64_initialize_trampoline): Likewise.
15104 (sparc_profile_hook): Likewise.
15105 * config/spu/spu.c (ea_load_store): Likewise.
15106 * config/spu/spu.md (floatunssidf2): Likewise.
15107 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
15108 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
15109 * config/visium/visium.c (expand_block_move_4): Likewise.
15110 (expand_block_move_2): Likewise.
15111 (expand_block_move_1): Likewise.
15112 (expand_block_set_4): Likewise.
15113 (expand_block_set_2): Likewise.
15114 (expand_block_set_1): Likewise.
15115 (visium_trampoline_init): Likewise.
15116 (visium_profile_hook): Likewise.
15117 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
15118 (xtensa_setup_frame_addresses): Likewise.
15119 (xtensa_trampoline_init): Likewise.
15120 * except.c (sjlj_emit_function_enter): Likewise.
15121 (sjlj_emit_function_exit): Likewise.
15122 * explow.c (allocate_dynamic_stack_space): Likewise.
15123 (probe_stack_range): Likewise.
15124 * expr.c (convert_mode_scalar): Likewise.
15125 * optabs.c (expand_binop): Likewise.
15126 (expand_twoval_binop_libfunc): Likewise.
15127 (expand_unop): Likewise.
15128 (prepare_cmp_insn): Likewise.
15129 (prepare_float_lib_cmp): Likewise.
15130 (expand_float): Likewise.
15131 (expand_fix): Likewise.
15132 (expand_fixed_convert): Likewise.
15133 (maybe_emit_sync_lock_test_and_set): Likewise.
15134 (expand_atomic_compare_and_swap): Likewise.
15135 (expand_mem_thread_fence): Likewise.
15136 (expand_atomic_fetch_op): Likewise.
15138 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
15140 * doc/generic.texi (OpenACC): Adjust URL.
15141 * doc/invoke.texi (C Dialect Options): Ditto.
15143 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
15145 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
15146 predicate for operand 1. Add (m,<S>) constraint.
15147 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
15148 Prevent memory operand 1 with register operand 2.
15150 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
15152 PR rtl-optimization/82024
15153 * combine.c (try_combine): If the combination result is a PARALLEL,
15154 and we only need to retain the SET in there that would be placed
15155 at I2, check that we can place that at I3 instead, before doing so.
15157 2017-09-01 Jakub Jelinek <jakub@redhat.com>
15160 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
15162 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
15163 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
15166 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
15167 Jeff Law <law@redhat.com>
15169 * varasm.c (bss_initializer_p): Do not put constants into .bss
15170 (categorize_decl_for_section): Handle bss_initializer_p returning
15171 false when DECL_INITIAL is NULL.
15173 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15176 * config/s390/s390.c (s390_can_inline_p): New function.
15178 2017-09-01 Jeff Law <law@redhat.com>
15180 PR tree-optimization/82052
15181 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
15182 Always initialize the returned slot after a hash table miss
15183 when INSERT is true.
15185 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
15187 * config/s390/s390.md (mem_signal_fence): Remove.
15188 * doc/md.texi (mem_signal_fence): Remove.
15189 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
15191 * target-insns.def (mem_signal_fence): Remove.
15193 2017-09-01 Jakub Jelinek <jakub@redhat.com>
15196 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
15199 * asan.c (create_odr_indicator): Strip name encoding from assembler
15200 name before appending it after __odr_asan_.
15202 2017-09-01 Martin Liska <mliska@suse.cz>
15204 PR tree-optimization/82059
15205 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
15206 frequency only when an edge is redirected.
15208 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15210 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
15211 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
15212 (arc_conditional_register_usage): Remove ARC600 lp_count
15214 (arc_file_start): Emit Tag_ARC_CPU_variation.
15215 (arc_can_use_doloop_p): New conditions to use ZOLs.
15216 (hwloop_fail): New function.
15217 (hwloop_optimize): Likewise.
15218 (hwloop_pattern_reg): Likewise.
15219 (arc_doloop_hooks): New struct, to be used with reorg_loops.
15220 (arc_reorg_loops): New function, calls reorg_loops.
15221 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
15222 (arc600_corereg_hazard): Remove ZOL checking, case handled by
15224 (arc_loop_hazard): Remove function, functionality moved into
15226 (arc_hazard): Remove arc_loop_hazard call.
15227 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
15228 into hwloop_optimize.
15229 (arc_label_align): Remove ZOL handling.
15230 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
15231 * config/arc/arc.md (doloop_begin): Remove pattern.
15232 (doloop_begin_i): Likewise.
15233 (doloop_end_i): Likewise.
15234 (doloop_fallback): Likewise.
15235 (doloop_fallback_m): Likewise.
15236 (doloop_end): Reimplement expand.
15237 (arc_lp): New pattern for LP instruction.
15238 (loop_end): New pattern.
15239 (loop_fail): Likewise.
15240 (decrement_and_branch_until_zero): Likewise.
15241 * config/arc/arc.opt (mlpc-width): New option.
15242 * doc/invoke.texi (mlpc-width): Document option.
15244 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15246 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
15247 (arc_ccfsm_advance): Fix checking for delay slots.
15248 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
15250 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15252 * config/arc/arc.md (movqi_insn): Add stores to save constant long
15254 (movhi_insn): Update store instruction constraint which are saving
15255 6-bit short immediates.
15256 (movsi_insn): Consider also short scaled load operations.
15257 (zero_extendhisi2_i): Use Usd constraint instead of T.
15258 (extendhisi2_i): Add q constraint.
15259 (arc_clzsi2): Add type and length attributes.
15260 (arc_ctzsi2): Likewise.
15261 * config/arc/constraints.md (Usc): Update constraint, the
15262 assembler can parse two relocations for a single instruction.
15264 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15266 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
15267 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
15269 2017-08-31 Olivier Hainque <hainque@adacore.com>
15271 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
15272 match as powerpc-wrs-vxworks*.
15274 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
15276 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
15277 register constraint for by-element operand.
15278 (aarch64_mls_elt_merge<mode>): Likewise.
15280 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
15282 * config/arc/arc.c (arc_can_follow_jump): Check for short
15285 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
15287 * config.gcc: Use g.opt for arc.
15288 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
15289 functionality moved to ...
15290 (legitimate_scaled_address_p): New function, ...here.
15291 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
15292 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
15293 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
15295 (arc_override_options): Handle G option.
15296 (arc_output_pic_addr_const): Correct function definition.
15297 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
15298 (arc_decl_anon_ns_mem_p): Delete.
15299 (arc_in_small_data_p): Overhaul this function to take into
15300 consideration the value given via G option.
15301 (arc_rewrite_small_data_1): Renamed and corrected old
15302 arc_rewrite_small_data function.
15303 (arc_rewrite_small_data): New function.
15304 (small_data_pattern): Don't use pic_offset_table_rtx.
15305 * config/arc/arc.h (CC1_SPEC): Recognize G option.
15306 * config/arc/simdext.md (movmisalignv2hi): Use
15307 prepare_move_operands function.
15309 (movmisalign*): Likewise.
15310 * doc/invoke.texi (ARC options): Document -G option.
15312 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
15314 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
15316 * config/arc/arc.c (arc_print_operand): Output scalled address for
15317 sdata whenever is possible.
15318 (arc_in_small_data_p): Allow sdata for 64bit datum when double
15319 load/stores are available.
15320 (compact_sda_memory_operand): Check for the alignment required by
15321 code density instructions.
15322 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
15324 * config/arc/constraints.md (Usd): Update constraint.
15325 (Us0): New constraint.
15326 (Usc): Update constraint.
15328 2017-08-31 Richard Biener <rguenther@suse.de>
15330 PR middle-end/82054
15331 * dwarf2out.c (dwarf2out_early_global_decl): Process each
15332 function only once.
15334 2017-08-31 Tamar Christina <tamar.christina@arm.com>
15336 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
15337 Resize type_signature.
15339 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
15340 Alan Hayward <alan.hayward@arm.com>
15341 David Sherwood <david.sherwood@arm.com>
15343 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
15344 subregs whose inner modes can be stored in GPRs.
15345 (aarch64_classify_index): Likewise.
15347 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
15348 Alan Hayward <alan.hayward@arm.com>
15349 David Sherwood <david.sherwood@arm.com>
15351 * config/aarch64/iterators.md (V_cmp_result): Rename to...
15352 (V_INT_EQUIV): ...this.
15353 (v_cmp_result): Rename to...
15354 (v_int_equiv): ...this.
15355 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
15356 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
15357 (copysign<mode>3): Likewise.
15358 (aarch64_simd_bsl<mode>_internal): Likewise.
15359 (aarch64_simd_bsl<mode>): Likewise.
15360 (vec_cmp<mode><mode>): Likewise.
15361 (vcond<mode><mode>): Likewise.
15362 (vcond<v_cmp_mixed><mode>): Likewise.
15363 (vcondu<mode><v_cmp_mixed>): Likewise.
15364 (aarch64_cm<optab><mode>): Likewise.
15365 (aarch64_cmtst<mode>): Likewise.
15366 (aarch64_fac<optab><mode>): Likewise.
15367 (vec_perm_const<mode>): Likewise.
15368 (vcond_mask_<mode><v_cmp_result>): Rename to...
15369 (vcond_mask_<mode><v_int_equiv>): ...this.
15370 (vec_cmp<mode><v_cmp_result>): Rename to...
15371 (vec_cmp<mode><v_int_equiv>): ...this.
15373 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
15374 Alan Hayward <alan.hayward@arm.com>
15375 David Sherwood <david.sherwood@arm.com>
15377 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
15379 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
15380 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
15381 (UNSPEC_LD4_DREG): New unspecs.
15382 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
15383 (aarch64_ld2<mode>_dreg_be): Replace with...
15384 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
15386 (aarch64_ld3<mode>_dreg_le)
15387 (aarch64_ld3<mode>_dreg_be): Replace with...
15388 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
15390 (aarch64_ld4<mode>_dreg_le)
15391 (aarch64_ld4<mode>_dreg_be): Replace with...
15392 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
15395 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15397 PR tree-optimization/81987
15398 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
15399 insert an initializer in a location not dominated by the stride
15402 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
15404 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
15405 on the entire header of the finally block in the fallthru case.
15407 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
15409 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
15411 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
15413 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
15414 rs6000_emit_move_from_cr and call renamed function.
15415 (rs6000_emit_prologue): Call renamed functions.
15416 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
15417 movesi_from_cr, remove volatile CRs.
15419 2017-08-30 Jon Beniston <jon@beniston.com>
15420 Richard Biener <rguenther@suse.de>
15422 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
15423 of VECTOR_MODE_P check.
15424 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
15425 element vector types.
15427 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15429 * df.h (df_read_modify_subreg_p): Remove in favor of...
15430 * rtl.h (read_modify_subreg_p): ...this new function. Take a
15431 const_rtx instead of an rtx.
15432 * cprop.c (local_cprop_find_used_regs): Update accordingly.
15433 * df-problems.c (df_word_lr_mark_ref): Likewise.
15434 * ira-lives.c (mark_pseudo_reg_live): Likewise.
15435 (mark_pseudo_reg_dead): Likewise.
15436 (mark_ref_dead): Likewise.
15437 * reginfo.c (init_subregs_of_mode): Likewise.
15438 * sched-deps.c (sched_analyze_1): Likewise.
15439 * df-scan.c (df_def_record_1): Likewise.
15440 (df_uses_record): Likewise.
15441 (df_read_modify_subreg_p): Remove in favor of...
15442 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
15443 const_rtx instead of an rtx.
15445 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15446 Alan Hayward <alan.hayward@arm.com>
15447 David Sherwood <david.sherwood@arm.com>
15449 * rtl.h (partial_subreg_p): New function.
15450 * caller-save.c (save_call_clobbered_regs): Use it.
15451 * calls.c (expand_call): Likewise.
15452 * combine.c (combinable_i3pat): Likewise.
15453 (simplify_set): Likewise.
15454 (make_extraction): Likewise.
15455 (make_compound_operation_int): Likewise.
15456 (gen_lowpart_or_truncate): Likewise.
15457 (force_to_mode): Likewise.
15458 (make_field_assignment): Likewise.
15459 (reg_truncated_to_mode): Likewise.
15460 (record_truncated_value): Likewise.
15461 (move_deaths): Likewise.
15462 * cse.c (record_jump_cond): Likewise.
15463 (cse_insn): Likewise.
15464 * cselib.c (cselib_lookup_1): Likewise.
15465 * expmed.c (extract_bit_field_using_extv): Likewise.
15466 * function.c (assign_parm_setup_reg): Likewise.
15467 * ifcvt.c (noce_convert_multiple_sets): Likewise.
15468 * ira-build.c (create_insn_allocnos): Likewise.
15469 * lra-coalesce.c (merge_pseudos): Likewise.
15470 * lra-constraints.c (match_reload): Likewise.
15471 (simplify_operand_subreg): Likewise.
15472 (curr_insn_transform): Likewise.
15473 * lra-lives.c (process_bb_lives): Likewise.
15474 * lra.c (new_insn_reg): Likewise.
15475 (lra_substitute_pseudo): Likewise.
15476 * regcprop.c (mode_change_ok): Likewise.
15477 (maybe_mode_change): Likewise.
15478 (copyprop_hardreg_forward_1): Likewise.
15479 * reload.c (push_reload): Likewise.
15480 (find_reloads): Likewise.
15481 (find_reloads_subreg_address): Likewise.
15482 * reload1.c (alter_reg): Likewise.
15483 (eliminate_regs_1): Likewise.
15484 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15486 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
15488 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
15489 back to if statements, including unpack.
15491 2017-08-30 Martin Liska <mliska@suse.cz>
15493 PR inline-asm/82001
15494 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
15496 (func_checker::compare_asm_inputs_outputs): ... this function.
15497 (func_checker::compare_gimple_asm): Use the function to compare
15498 also ASM constrains.
15499 * ipa-icf-gimple.h: Rename the function.
15501 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15502 Alan Hayward <alan.hayward@arm.com>
15503 David Sherwood <david.sherwood@arm.com>
15505 * coretypes.h (complex_mode): New type.
15506 * gdbhooks.py (build_pretty_printer): Handle it.
15507 * machmode.h (complex_mode): New class.
15508 (complex_mode::includes_p): New function.
15509 (is_complex_int_mode): Likewise.
15510 (is_complex_float_mode): Likewise.
15511 * genmodes.c (get_mode_class): Handle complex mode classes.
15512 * function.c (expand_function_end): Use is_complex_int_mode.
15514 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15515 Alan Hayward <alan.hayward@arm.com>
15516 David Sherwood <david.sherwood@arm.com>
15518 * coretypes.h (scalar_mode_pod): New typedef.
15519 * gdbhooks.py (build_pretty_printer): Handle it.
15520 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
15521 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
15522 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
15523 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
15524 as_a <scalar_mode>.
15526 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15527 Alan Hayward <alan.hayward@arm.com>
15528 David Sherwood <david.sherwood@arm.com>
15530 * machmode.h (mode_for_vector): Take a scalar_mode instead
15532 * stor-layout.c (mode_for_vector): Likewise.
15533 * explow.c (promote_mode): Use as_a <scalar_mode>.
15534 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
15536 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15537 Alan Hayward <alan.hayward@arm.com>
15538 David Sherwood <david.sherwood@arm.com>
15540 * target.def (preferred_simd_mode): Take a scalar_mode
15541 instead of a machine_mode.
15542 * targhooks.h (default_preferred_simd_mode): Likewise.
15543 * targhooks.c (default_preferred_simd_mode): Likewise.
15544 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
15545 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
15546 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
15547 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
15548 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
15549 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
15550 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
15551 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
15553 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
15554 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
15555 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
15556 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
15557 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
15558 * doc/tm.texi: Regenerate.
15559 * optabs-query.c (can_vec_mask_load_store_p): Return false for
15562 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15563 Alan Hayward <alan.hayward@arm.com>
15564 David Sherwood <david.sherwood@arm.com>
15566 * target.def (scalar_mode_supported_p): Take a scalar_mode
15567 instead of a machine_mode.
15568 * targhooks.h (default_scalar_mode_supported_p): Likewise.
15569 * targhooks.c (default_scalar_mode_supported_p): Likewise.
15570 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
15571 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
15572 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
15573 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
15574 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
15575 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
15576 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
15577 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
15578 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
15579 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
15580 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
15581 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
15583 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
15584 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
15585 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
15586 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
15587 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
15589 * doc/tm.texi: Regenerate.
15591 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15592 Alan Hayward <alan.hayward@arm.com>
15593 David Sherwood <david.sherwood@arm.com>
15595 * coretypes.h (opt_scalar_mode): New typedef.
15596 * gdbhooks.py (build_pretty_printers): Handle it.
15597 * machmode.h (mode_iterator::get_2xwider): Add overload for
15599 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
15601 * expr.c (convert_mode_scalar): Likewise.
15602 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15603 * optabs.c (expand_float): Likewise.
15604 (expand_fix): Likewise.
15605 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15607 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15608 Alan Hayward <alan.hayward@arm.com>
15609 David Sherwood <david.sherwood@arm.com>
15611 * optabs.c (expand_float): Explicitly check for scalars before
15612 using a branching expansion.
15613 (expand_fix): Likewise.
15615 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15616 Alan Hayward <alan.hayward@arm.com>
15617 David Sherwood <david.sherwood@arm.com>
15619 * expr.c (convert_mode): Split scalar handling out into...
15620 (convert_mode_scalar): ...this new function. Treat the modes
15623 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15624 Alan Hayward <alan.hayward@arm.com>
15625 David Sherwood <david.sherwood@arm.com>
15627 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
15629 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
15631 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15632 Alan Hayward <alan.hayward@arm.com>
15633 David Sherwood <david.sherwood@arm.com>
15635 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
15636 rather than a machine_mode.
15637 (fixed_from_string): Likewise.
15638 (fixed_convert): Likewise.
15639 (fixed_convert_from_int): Likewise.
15640 (fixed_convert_from_real): Likewise.
15641 (real_convert_from_fixed): Likewise.
15642 * fixed-value.c (fixed_from_double_int): Likewise.
15643 (fixed_from_string): Likewise.
15644 (fixed_convert): Likewise.
15645 (fixed_convert_from_int): Likewise.
15646 (fixed_convert_from_real): Likewise.
15647 (real_convert_from_fixed): Likewise.
15648 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
15650 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15651 Alan Hayward <alan.hayward@arm.com>
15652 David Sherwood <david.sherwood@arm.com>
15654 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
15655 of separate mode class checks. Do not allow vector modes here.
15656 (immed_wide_int_const): Use as_a <scalar_mode>.
15657 * explow.c (trunc_int_for_mode): Likewise.
15658 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
15659 (wi::shwi): Likewise.
15660 (wi::min_value): Likewise.
15661 (wi::max_value): Likewise.
15662 * dwarf2out.c (loc_descriptor): Likewise.
15663 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
15664 for CONST_WIDE_INT.
15666 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15667 Alan Hayward <alan.hayward@arm.com>
15668 David Sherwood <david.sherwood@arm.com>
15670 * tree.h (SCALAR_TYPE_MODE): New macro.
15671 * expr.c (expand_expr_addr_expr_1): Use it.
15672 (expand_expr_real_2): Likewise.
15673 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
15674 (fold_convert_const_fixed_from_int): Likewise.
15675 (fold_convert_const_fixed_from_real): Likewise.
15676 (native_encode_fixed): Likewise
15677 (native_encode_complex): Likewise
15678 (native_encode_vector): Likewise.
15679 (native_interpret_fixed): Likewise.
15680 (native_interpret_real): Likewise.
15681 (native_interpret_complex): Likewise.
15682 (native_interpret_vector): Likewise.
15683 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
15684 (simd_clone_adjust_argument_types): Likewise.
15685 (simd_clone_init_simd_arrays): Likewise.
15686 (simd_clone_adjust): Likewise.
15687 * stor-layout.c (layout_type): Likewise.
15688 * tree.c (build_minus_one_cst): Likewise.
15689 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15690 * tree-inline.c (estimate_move_cost): Likewise.
15691 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
15692 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
15693 (vectorizable_reduction): Likewise.
15694 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
15695 (vect_recog_mixed_size_cond_pattern): Likewise.
15696 (check_bool_pattern): Likewise.
15697 (adjust_bool_pattern): Likewise.
15698 (search_type_for_mask_1): Likewise.
15699 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
15700 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15701 (vectorizable_load): Likewise.
15702 (vectorizable_store): Likewise.
15703 * ubsan.c (ubsan_encode_value): Likewise.
15704 * varasm.c (output_constant): Likewise.
15706 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15707 Alan Hayward <alan.hayward@arm.com>
15708 David Sherwood <david.sherwood@arm.com>
15710 * coretypes.h (scalar_mode): New class.
15711 * machmode.h (scalar_mode): Likewise.
15712 (scalar_mode::includes_p): New function.
15713 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
15714 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
15715 * genmodes.c (get_mode_class): Handle remaining scalar modes.
15716 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
15717 * expmed.c (store_bit_field_1): Likewise.
15718 (extract_bit_field_1): Likewise.
15719 * expr.c (write_complex_part): Likewise.
15720 (read_complex_part): Likewise.
15721 (emit_move_complex_push): Likewise.
15722 (expand_expr_real_2): Likewise.
15723 * function.c (assign_parm_setup_reg): Likewise.
15724 (assign_parms_unsplit_complex): Likewise.
15725 * optabs.c (expand_binop): Likewise.
15726 * rtlanal.c (subreg_get_info): Likewise.
15727 * simplify-rtx.c (simplify_immed_subreg): Likewise.
15728 * varasm.c (output_constant_pool_2): Likewise.
15730 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15731 Alan Hayward <alan.hayward@arm.com>
15732 David Sherwood <david.sherwood@arm.com>
15734 * expmed.c (extract_high_half): Use scalar_int_mode and remove
15736 (expmed_mult_highpart_optab): Likewise.
15737 (expmed_mult_highpart): Likewise.
15739 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15740 Alan Hayward <alan.hayward@arm.com>
15741 David Sherwood <david.sherwood@arm.com>
15743 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
15744 instead of a machine_mode.
15745 (builtin_memset_read_str): Likewise.
15746 * builtins.c (c_readstr): Likewise.
15747 (builtin_memcpy_read_str): Likewise.
15748 (builtin_strncpy_read_str): Likewise.
15749 (builtin_memset_read_str): Likewise.
15750 (builtin_memset_gen_str): Likewise.
15751 (expand_builtin_signbit): Use scalar_int_mode for local variables.
15752 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
15753 instead of a machine_mode.
15754 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
15756 (make_extraction): Likewise.
15757 (try_widen_shift_mode): Take and return scalar_int_modes instead
15759 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
15760 a scalar_int_mode instead of a machine_mode.
15761 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
15762 (avr_addr_space_pointer_mode): Likewise.
15763 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
15764 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
15765 (msp430_unwind_word_mode): Likewise.
15766 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
15767 (spu_addr_space_pointer_mode): Likewise.
15768 (spu_addr_space_address_mode): Likewise.
15769 (spu_libgcc_cmp_return_mode): Likewise.
15770 (spu_libgcc_shift_count_mode): Likewise.
15771 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
15772 (rl78_addr_space_pointer_mode): Likewise.
15773 (fl78_unwind_word_mode): Likewise.
15774 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
15776 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
15777 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
15778 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
15779 (mips_valid_pointer_mode): Likewise.
15780 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
15781 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
15782 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
15784 (ft32_addr_space_address_mode): Likewise.
15785 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
15786 scalar_int_mode instead of a machine_mode.
15787 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
15789 (m32c_addr_space_address_mode): Likewise.
15790 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
15791 (rs6000_eh_return_filter_mode): Likewise.
15792 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
15793 (rs6000_eh_return_filter_mode): Likewise.
15794 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
15795 (s390_libgcc_shift_count_mode): Likewise.
15796 (s390_unwind_word_mode): Likewise.
15797 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
15799 * target.def (mode_rep_extended): Likewise.
15800 (valid_pointer_mode): Likewise.
15801 (addr_space.valid_pointer_mode): Likewise.
15802 (eh_return_filter_mode): Return a scalar_int_mode rather than
15804 (libgcc_cmp_return_mode): Likewise.
15805 (libgcc_shift_count_mode): Likewise.
15806 (unwind_word_mode): Likewise.
15807 (addr_space.pointer_mode): Likewise.
15808 (addr_space.address_mode): Likewise.
15809 * doc/tm.texi: Regenerate.
15810 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
15812 (do_jump): Use scalar_int_mode for local variables.
15813 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
15814 rather than a machine_mode.
15815 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
15816 (scompare_loc_descriptor_wide): Likewise.
15817 (scompare_loc_descriptor_narrow): Likewise.
15818 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
15820 * except.c (sjlj_emit_dispatch_table): Likewise.
15821 (expand_builtin_eh_copy_values): Likewise.
15822 * explow.c (convert_memory_address_addr_space_1): Likewise.
15823 Take a scalar_int_mode rather than a machine_mode.
15824 (convert_memory_address_addr_space): Take a scalar_int_mode rather
15825 than a machine_mode.
15826 (memory_address_addr_space): Use scalar_int_mode for local variables.
15827 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
15828 rather than a machine_mode.
15829 * expmed.c (mask_rtx): Likewise.
15830 (init_expmed_one_conv): Likewise.
15831 (expand_mult_highpart_adjust): Likewise.
15832 (extract_high_half): Likewise.
15833 (expmed_mult_highpart_optab): Likewise.
15834 (expmed_mult_highpart): Likewise.
15835 (expand_smod_pow2): Likewise.
15836 (expand_sdiv_pow2): Likewise.
15837 (emit_store_flag_int): Likewise.
15838 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
15840 (extract_low_bits): Likewise.
15841 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
15843 * expr.c (pieces_addr::adjust): Likewise.
15844 (can_store_by_pieces): Likewise.
15845 (store_by_pieces): Likewise.
15846 (clear_by_pieces_1): Likewise.
15847 (expand_expr_addr_expr_1): Likewise.
15848 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
15849 (expand_expr_real_1): Likewise.
15850 (try_casesi): Likewise.
15851 * final.c (shorten_branches): Likewise.
15852 * fold-const.c (fold_convert_const_int_from_fixed): Change the
15853 type of "mode" to machine_mode.
15854 * internal-fn.c (expand_arith_overflow_result_store): Take a
15855 scalar_int_mode rather than a machine_mode.
15856 (expand_mul_overflow): Use scalar_int_mode for local variables.
15857 * loop-doloop.c (doloop_modify): Likewise.
15858 (doloop_optimize): Likewise.
15859 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
15860 than a machine_mode.
15861 (expand_doubleword_shift_condmove): Likewise.
15862 (expand_doubleword_shift): Likewise.
15863 (expand_doubleword_clz): Likewise.
15864 (expand_doubleword_popcount): Likewise.
15865 (expand_doubleword_parity): Likewise.
15866 (expand_absneg_bit): Use scalar_int_mode for local variables.
15867 (prepare_float_lib_cmp): Likewise.
15868 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
15869 rather than a machine_mode.
15870 (convert_memory_address_addr_space): Likewise.
15871 (get_mode_bounds): Likewise.
15872 (get_address_mode): Return a scalar_int_mode rather than a
15874 * rtlanal.c (get_address_mode): Likewise.
15875 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
15876 than a machine_mode.
15877 * targhooks.c (default_mode_rep_extended): Likewise.
15878 (default_valid_pointer_mode): Likewise.
15879 (default_addr_space_valid_pointer_mode): Likewise.
15880 (default_eh_return_filter_mode): Return a scalar_int_mode rather
15881 than a machine_mode.
15882 (default_libgcc_cmp_return_mode): Likewise.
15883 (default_libgcc_shift_count_mode): Likewise.
15884 (default_unwind_word_mode): Likewise.
15885 (default_addr_space_pointer_mode): Likewise.
15886 (default_addr_space_address_mode): Likewise.
15887 * targhooks.h (default_eh_return_filter_mode): Likewise.
15888 (default_libgcc_cmp_return_mode): Likewise.
15889 (default_libgcc_shift_count_mode): Likewise.
15890 (default_unwind_word_mode): Likewise.
15891 (default_addr_space_pointer_mode): Likewise.
15892 (default_addr_space_address_mode): Likewise.
15893 (default_mode_rep_extended): Take a scalar_int_mode rather than
15895 (default_valid_pointer_mode): Likewise.
15896 (default_addr_space_valid_pointer_mode): Likewise.
15897 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
15899 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
15900 rather than a machine_mode.
15901 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
15902 for local variables.
15903 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15904 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
15905 than a machine_mode.
15907 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15908 Alan Hayward <alan.hayward@arm.com>
15909 David Sherwood <david.sherwood@arm.com>
15911 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
15912 the mode argument to scalar_int_mode.
15913 (do_jump_by_parts_zero_rtx): Likewise.
15914 (do_jump_by_parts_equality_rtx): Likewise.
15915 (do_jump_by_parts_greater): Take a mode argument.
15916 (do_jump_by_parts_equality): Likewise.
15917 (do_jump_1): Update calls accordingly.
15919 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15920 Alan Hayward <alan.hayward@arm.com>
15921 David Sherwood <david.sherwood@arm.com>
15923 * is-a.h (safe_dyn_cast): New function.
15924 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
15925 (jump_table_for_label): Likewise.
15926 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
15927 instead of an rtx_insn *.
15928 (shorten_branches): Use dyn_cast instead of LABEL_P and
15929 JUMP_TABLE_DATA_P. Use jump_table_for_label and
15930 rtx_jump_table_data::get_data_mode.
15931 (final_scan_insn): Likewise.
15933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15934 Alan Hayward <alan.hayward@arm.com>
15935 David Sherwood <david.sherwood@arm.com>
15937 * combine.c (try_combine): Use is_a <scalar_int_mode> when
15938 trying to combine a full-register integer set with a subreg
15941 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15942 Alan Hayward <alan.hayward@arm.com>
15943 David Sherwood <david.sherwood@arm.com>
15945 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
15946 that is always either address_mode or pointer_mode.
15948 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15949 Alan Hayward <alan.hayward@arm.com>
15950 David Sherwood <david.sherwood@arm.com>
15952 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
15953 when the two are known to be equal.
15955 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15956 Alan Hayward <alan.hayward@arm.com>
15957 David Sherwood <david.sherwood@arm.com>
15959 * simplify-rtx.c (simplify_const_unary_operation): Use
15960 is_a <scalar_int_mode> instead of checking for a nonzero
15961 precision. Forcibly convert op_mode to a scalar_int_mode
15962 in that case. More clearly differentiate the operand and
15963 result modes and use the former when deciding what the value
15964 of a count-bits operation should be. Use is_int_mode instead
15965 of checking for a MODE_INT. Remove redundant check for whether
15966 this mode has a zero precision.
15968 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15969 Alan Hayward <alan.hayward@arm.com>
15970 David Sherwood <david.sherwood@arm.com>
15972 * optabs.c (widen_leading): Change the type of the mode argument
15973 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
15974 (widen_bswap): Likewise.
15975 (expand_parity): Likewise.
15976 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
15977 (expand_ffs): Likewise.
15978 (epand_unop): Check for scalar integer modes before calling the
15981 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15982 Alan Hayward <alan.hayward@arm.com>
15983 David Sherwood <david.sherwood@arm.com>
15985 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
15987 (expand_expr_real_1): Update call accordingly.
15989 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15990 Alan Hayward <alan.hayward@arm.com>
15991 David Sherwood <david.sherwood@arm.com>
15993 * expmed.c (store_bit_field_using_insv): Add op0_mode and
15994 value_mode arguments. Use scalar_int_mode internally.
15995 (store_bit_field_1): Rename the new integer mode from imode
15996 to op0_mode and use it instead of GET_MODE (op0). Update calls
15997 to store_split_bit_field, store_bit_field_using_insv and
15998 store_fixed_bit_field.
15999 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
16000 Use scalar_int_mode internally. Use a bit count rather than a mode
16001 when calculating the largest bit size for get_best_mode.
16002 Update calls to store_split_bit_field and store_fixed_bit_field_1.
16003 (store_fixed_bit_field_1): Add mode and value_mode arguments.
16004 Remove assertion that OP0 has a scalar integer mode.
16005 (store_split_bit_field): Add op0_mode and value_mode arguments.
16006 Update calls to extract_fixed_bit_field.
16007 (extract_bit_field_using_extv): Add an op0_mode argument.
16008 Use scalar_int_mode internally.
16009 (extract_bit_field_1): Rename the new integer mode from imode to
16010 op0_mode and use it instead of GET_MODE (op0). Update calls to
16011 extract_split_bit_field, extract_bit_field_using_extv and
16012 extract_fixed_bit_field.
16013 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
16014 to extract_split_bit_field and extract_fixed_bit_field_1.
16015 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
16016 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
16017 on the target mode.
16018 (extract_split_bit_field): Add an op0_mode argument. Update call
16019 to extract_fixed_bit_field.
16021 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16022 Alan Hayward <alan.hayward@arm.com>
16023 David Sherwood <david.sherwood@arm.com>
16025 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
16026 * explow.c (hard_function_value): Likewise.
16027 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
16028 convert_to_mode call outside the loop.
16029 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
16030 for the mode iterator. Require the mode specified by max_pieces
16032 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
16034 (copy_blkmode_to_reg): Likewise.
16035 (set_storage_via_setmem): Likewise.
16036 * optabs.c (prepare_cmp_insn): Likewise.
16037 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16038 * stor-layout.c (finish_bitfield_representative): Likewise.
16040 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16041 Alan Hayward <alan.hayward@arm.com>
16042 David Sherwood <david.sherwood@arm.com>
16044 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
16045 * expr.c (convert_move): Use them.
16046 (convert_modes): Likewise.
16047 (store_expr_with_bounds): Likewise.
16049 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16050 Alan Hayward <alan.hayward@arm.com>
16051 David Sherwood <david.sherwood@arm.com>
16053 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
16054 parameter for the mode of "x". Remove the "known_x", "known_mode"
16055 and "known_ret" arguments. Change the type of the mode argument
16056 to scalar_int_mode.
16057 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
16058 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
16059 (reg_num_sign_bit_copies_for_combine): Likewise.
16060 * rtlanal.c (nonzero_bits1): Likewise.
16061 (num_sign_bit_copies1): Likewise.
16062 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
16063 (reg_num_sign_bit_copies_general): Likewise.
16064 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
16065 (reg_nonzero_bits_general): Likewise.
16067 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16068 Alan Hayward <alan.hayward@arm.com>
16069 David Sherwood <david.sherwood@arm.com>
16071 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
16072 than in subroutines. Return 1 for non-integer modes.
16073 (cached_num_sign_bit_copies): Change the type of the mode parameter
16074 to scalar_int_mode.
16075 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
16076 classes. Handle CONST_INT_P first and then check whether X also
16077 has a scalar integer mode. Check the same thing for inner registers
16078 of a SUBREG and for values that are being extended or truncated.
16080 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16081 Alan Hayward <alan.hayward@arm.com>
16082 David Sherwood <david.sherwood@arm.com>
16084 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
16085 in subroutines. Return the mode mask for non-integer modes.
16086 (cached_nonzero_bits): Change the type of the mode parameter
16087 to scalar_int_mode.
16088 (nonzero_bits1): Likewise. Remove early exit for other mode
16089 classes. Handle CONST_INT_P first and then check whether X
16090 also has a scalar integer mode.
16092 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16093 Alan Hayward <alan.hayward@arm.com>
16094 David Sherwood <david.sherwood@arm.com>
16096 * expr.c (widest_int_mode_for_size): Make the comment match the code.
16097 Return a scalar_int_mode and assert that the size is greater than
16099 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
16100 (op_by_pieces_d::op_by_pieces_d): Likewise.
16101 (op_by_pieces_d::run): Likewise.
16102 (can_store_by_pieces): Likewise.
16104 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16105 Alan Hayward <alan.hayward@arm.com>
16106 David Sherwood <david.sherwood@arm.com>
16108 * combine.c (extract_left_shift): Add a mode argument and update
16110 (make_compound_operation_int): Change the type of the mode parameter
16111 to scalar_int_mode and update the call to extract_left_shift.
16113 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16114 Alan Hayward <alan.hayward@arm.com>
16115 David Sherwood <david.sherwood@arm.com>
16117 * combine.c (simplify_and_const_int): Change the type of the mode
16118 parameter to scalar_int_mode.
16119 (simplify_and_const_int_1): Likewise. Update recursive call.
16121 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16122 Alan Hayward <alan.hayward@arm.com>
16123 David Sherwood <david.sherwood@arm.com>
16125 * combine.c (simplify_compare_const): Check that the mode is a
16126 scalar_int_mode (rather than VOIDmode) before testing its
16128 (simplify_comparison): Move COMPARISON_P handling out of the
16129 loop and restrict the latter part of the loop to scalar_int_modes.
16130 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
16131 and when considering SUBREG_REGs. Use is_int_mode instead of
16132 checking GET_MODE_CLASS against MODE_INT.
16134 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16135 Alan Hayward <alan.hayward@arm.com>
16136 David Sherwood <david.sherwood@arm.com>
16138 * combine.c (try_widen_shift_mode): Move check for equal modes to...
16139 (simplify_shift_const_1): ...here. Use scalar_int_mode for
16140 shift_unit_mode and for modes involved in scalar shifts.
16142 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16143 Alan Hayward <alan.hayward@arm.com>
16144 David Sherwood <david.sherwood@arm.com>
16146 * combine.c (force_int_to_mode): New function, split out from...
16147 (force_to_mode): ...here. Keep xmode up-to-date and use it
16148 instead of GET_MODE (x).
16150 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16151 Alan Hayward <alan.hayward@arm.com>
16152 David Sherwood <david.sherwood@arm.com>
16154 * optabs-query.h (extraction_insn::struct_mode): Change type to
16155 opt_scalar_int_mode and update comment.
16156 (extraction_insn::field_mode): Change type to scalar_int_mode.
16157 (extraction_insn::pos_mode): Likewise.
16158 * combine.c (make_extraction): Update accordingly.
16159 * optabs-query.c (get_traditional_extraction_insn): Likewise.
16160 (get_optab_extraction_insn): Likewise.
16161 * recog.c (simplify_while_replacing): Likewise.
16162 * expmed.c (narrow_bit_field_mem): Change the type of the mode
16163 parameter to opt_scalar_int_mode.
16165 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16166 Alan Hayward <alan.hayward@arm.com>
16167 David Sherwood <david.sherwood@arm.com>
16169 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
16170 to a scalar_int_mode instead of a machine_mode.
16171 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
16172 (get_best_mode): Return a boolean and use a pointer argument to store
16173 the selected mode. Replace the limit mode parameter with a bit limit.
16174 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
16175 for the values returned by bit_field_mode_iterator::next_mode.
16176 (store_bit_field): Update call to get_best_mode.
16177 (store_fixed_bit_field): Likewise.
16178 (extract_fixed_bit_field): Likewise.
16179 * expr.c (optimize_bitfield_assignment_op): Likewise.
16180 * fold-const.c (optimize_bit_field_compare): Likewise.
16181 (fold_truth_andor_1): Likewise.
16182 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
16183 Update for new type of m_mode.
16184 (get_best_mode): As above.
16186 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16187 Alan Hayward <alan.hayward@arm.com>
16188 David Sherwood <david.sherwood@arm.com>
16190 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
16191 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
16192 (store_bit_field): Check is_a <scalar_int_mode> before calling
16193 strict_volatile_bitfield_p.
16194 (extract_bit_field): Likewise.
16196 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16197 Alan Hayward <alan.hayward@arm.com>
16198 David Sherwood <david.sherwood@arm.com>
16200 * target.def (cstore_mode): Return a scalar_int_mode.
16201 * doc/tm.texi: Regenerate.
16202 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
16203 * targhooks.h (default_cstore_mode): Likewise.
16204 * targhooks.c (default_cstore_mode): Likewise, using a forced
16206 * expmed.c (emit_cstore): Expect the target of the cstore to be
16209 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16210 Alan Hayward <alan.hayward@arm.com>
16211 David Sherwood <david.sherwood@arm.com>
16213 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
16215 (niter_desc): Likewise mode.
16216 (iv_analyze): Add a mode parameter.
16218 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
16219 and change its type to scalar_int_mode.
16220 * loop-iv.c: Update commentary at head of file.
16221 (iv_constant): Pass the mode paraeter before the rtx it describes
16222 and change its type to scalar_int_mode. Remove VOIDmode handling.
16223 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
16224 (iv_extend): Likewise.
16225 (shorten_into_mode): Likewise.
16226 (iv_add): Use scalar_int_mode.
16227 (iv_mult): Likewise.
16228 (iv_shift): Likewise.
16229 (canonicalize_iv_subregs): Likewise.
16230 (get_biv_step_1): Pass the outer_mode parameter before the rtx
16231 it describes and change its mode to scalar_int_mode. Also change
16232 the type of the returned inner_mode to scalar_int_mode.
16233 (get_biv_step): Likewise, turning outer_mode from a pointer
16234 into a direct parameter. Update call to get_biv_step_1.
16235 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
16236 iv_constant and get_biv_step.
16237 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
16238 and change its type to scalar_int_mode. Don't initialise iv->mode
16239 to VOIDmode and remove later checks for its still being VOIDmode.
16240 Update calls to iv_analyze_op and iv_analyze_expr. Check
16241 is_a <scalar_int_mode> when changing the mode under consideration.
16242 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
16243 Update call to iv_analyze_expr.
16244 (iv_analyze_op): Add a mode parameter. Reject subregs whose
16245 inner register is not also a scalar_int_mode. Update call to
16247 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
16248 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
16249 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
16250 separate mode class checks. Update calls to iv_analyze. Remove
16251 fix-up of VOIDmodes after iv_analyze_biv.
16252 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
16253 don't have a scalar_int_mode. Update call to biv_p.
16255 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16256 Alan Hayward <alan.hayward@arm.com>
16257 David Sherwood <david.sherwood@arm.com>
16259 * cfgexpand.c (convert_debug_memory_address): Use
16260 as_a <scalar_int_mode>.
16261 * combine.c (expand_compound_operation): Likewise.
16262 (make_extraction): Likewise.
16263 (change_zero_ext): Likewise.
16264 (simplify_comparison): Likewise.
16265 * cse.c (cse_insn): Likewise.
16266 * dwarf2out.c (minmax_loc_descriptor): Likewise.
16267 (mem_loc_descriptor): Likewise.
16268 (loc_descriptor): Likewise.
16269 * expmed.c (init_expmed_one_mode): Likewise.
16270 (synth_mult): Likewise.
16271 (emit_store_flag_1): Likewise.
16272 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
16273 of a comparison with size.
16274 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
16275 (reduce_to_bit_field_precision): Likewise.
16276 * function.c (expand_function_end): Likewise.
16277 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
16278 * loop-doloop.c (doloop_modify): Likewise.
16279 * optabs.c (expand_binop): Likewise.
16280 (expand_unop): Likewise.
16281 (expand_copysign_absneg): Likewise.
16282 (prepare_cmp_insn): Likewise.
16283 (maybe_legitimize_operand): Likewise.
16284 * recog.c (const_scalar_int_operand): Likewise.
16285 * rtlanal.c (get_address_mode): Likewise.
16286 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16287 (simplify_cond_clz_ctz): Likewise.
16288 * tree-nested.c (get_nl_goto_field): Likewise.
16289 * tree.c (build_vector_type_for_mode): Likewise.
16290 * var-tracking.c (use_narrower_mode): Likewise.
16292 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16293 Alan Hayward <alan.hayward@arm.com>
16294 David Sherwood <david.sherwood@arm.com>
16296 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
16297 * builtins.c (expand_builtin_signbit): Use it.
16298 * cfgexpand.c (expand_debug_expr): Likewise.
16299 * dojump.c (do_jump): Likewise.
16300 (do_compare_and_jump): Likewise.
16301 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
16302 * expmed.c (make_tree): Likewise.
16303 * expr.c (expand_expr_real_2): Likewise.
16304 (expand_expr_real_1): Likewise.
16305 (try_casesi): Likewise.
16306 * fold-const-call.c (fold_const_call_ss): Likewise.
16307 * fold-const.c (unextend): Likewise.
16308 (extract_muldiv_1): Likewise.
16309 (fold_single_bit_test): Likewise.
16310 (native_encode_int): Likewise.
16311 (native_encode_string): Likewise.
16312 (native_interpret_int): Likewise.
16313 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
16314 * internal-fn.c (expand_addsub_overflow): Likewise.
16315 (expand_neg_overflow): Likewise.
16316 (expand_mul_overflow): Likewise.
16317 (expand_arith_overflow): Likewise.
16318 * match.pd: Likewise.
16319 * stor-layout.c (layout_type): Likewise.
16320 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
16321 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
16322 * tree-ssanames.c (get_range_info): Likewise.
16323 * tree-switch-conversion.c (array_value_type) Likewise.
16324 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
16325 (vect_recog_divmod_pattern): Likewise.
16326 (vect_recog_mixed_size_cond_pattern): Likewise.
16327 * tree-vrp.c (extract_range_basic): Likewise.
16328 (simplify_float_conversion_using_ranges): Likewise.
16329 * tree.c (int_fits_type_p): Likewise.
16330 * ubsan.c (instrument_bool_enum_load): Likewise.
16331 * varasm.c (mergeable_string_section): Likewise.
16332 (narrowing_initializer_constant_valid_p): Likewise.
16333 (output_constant): Likewise.
16335 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16336 Alan Hayward <alan.hayward@arm.com>
16337 David Sherwood <david.sherwood@arm.com>
16339 * machmode.h (NARROWEST_INT_MODE): New macro.
16340 * expr.c (alignment_for_piecewise_move): Use it instead of
16341 GET_CLASS_NARROWEST_MODE (MODE_INT).
16342 (push_block): Likewise.
16343 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
16345 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16347 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16348 Alan Hayward <alan.hayward@arm.com>
16349 David Sherwood <david.sherwood@arm.com>
16351 * postreload.c (move2add_valid_value_p): Change the type of the
16352 mode parameter to scalar_int_mode.
16353 (move2add_use_add2_insn): Add a mode parameter and use it instead
16355 (move2add_use_add3_insn): Likewise.
16356 (reload_cse_move2add): Update accordingly.
16358 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16359 Alan Hayward <alan.hayward@arm.com>
16360 David Sherwood <david.sherwood@arm.com>
16362 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
16364 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
16365 * optabs.c (expand_unop): Likewise.
16367 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16368 Alan Hayward <alan.hayward@arm.com>
16369 David Sherwood <david.sherwood@arm.com>
16371 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
16372 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
16373 (popcount_loc_descriptor): Likewise.
16374 (bswap_loc_descriptor): Likewise.
16375 (rotate_loc_descriptor): Likewise.
16376 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
16377 calling the functions above.
16379 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16380 Alan Hayward <alan.hayward@arm.com>
16381 David Sherwood <david.sherwood@arm.com>
16383 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
16385 (try_combine): Likewise.
16386 (simplify_if_then_else): Likewise.
16387 * cse.c (cse_insn): Likewise.
16388 * dwarf2out.c (mem_loc_descriptor): Likewise.
16389 * emit-rtl.c (gen_lowpart_common): Likewise.
16390 * simplify-rtx.c (simplify_truncation): Likewise.
16391 (simplify_binary_operation_1): Likewise.
16392 (simplify_const_relational_operation): Likewise.
16393 (simplify_ternary_operation): Likewise.
16394 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
16396 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16397 Alan Hayward <alan.hayward@arm.com>
16398 David Sherwood <david.sherwood@arm.com>
16400 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
16401 * reload.c (push_reload): Likewise.
16402 (find_reloads): Likewise.
16404 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16405 Alan Hayward <alan.hayward@arm.com>
16406 David Sherwood <david.sherwood@arm.com>
16408 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
16409 (make_compound_operation_int): Likewise.
16410 (change_zero_ext): Likewise.
16411 * expr.c (convert_move): Likewise.
16412 (convert_modes): Likewise.
16413 * fwprop.c (forward_propagate_subreg): Likewise.
16414 * loop-iv.c (get_biv_step_1): Likewise.
16415 * optabs.c (widen_operand): Likewise.
16416 * postreload.c (move2add_valid_value_p): Likewise.
16417 * recog.c (simplify_while_replacing): Likewise.
16418 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16419 (simplify_binary_operation_1): Likewise. Remove redundant
16420 mode equality check.
16422 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16423 Alan Hayward <alan.hayward@arm.com>
16424 David Sherwood <david.sherwood@arm.com>
16426 * combine.c (combine_simplify_rtx): Add checks for
16427 is_a <scalar_int_mode>.
16428 (simplify_if_then_else): Likewise.
16429 (make_field_assignment): Likewise.
16430 (simplify_comparison): Likewise.
16431 * ifcvt.c (noce_try_bitop): Likewise.
16432 * loop-invariant.c (canonicalize_address_mult): Likewise.
16433 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16436 Alan Hayward <alan.hayward@arm.com>
16437 David Sherwood <david.sherwood@arm.com>
16439 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
16440 is_a <scalar_int_mode> instead of != BLKmode.
16442 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16443 Alan Hayward <alan.hayward@arm.com>
16444 David Sherwood <david.sherwood@arm.com>
16446 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
16447 instead of != VOIDmode.
16448 * combine.c (if_then_else_cond): Likewise.
16449 (change_zero_ext): Likewise.
16450 * dwarf2out.c (mem_loc_descriptor): Likewise.
16451 (loc_descriptor): Likewise.
16452 * rtlanal.c (canonicalize_condition): Likewise.
16453 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
16455 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16456 Alan Hayward <alan.hayward@arm.com>
16457 David Sherwood <david.sherwood@arm.com>
16459 * simplify-rtx.c (simplify_binary_operation_1): Use
16460 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
16462 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16463 Alan Hayward <alan.hayward@arm.com>
16464 David Sherwood <david.sherwood@arm.com>
16466 * wide-int.h (int_traits<unsigned char>) New class.
16467 (int_traits<unsigned short>) Likewise.
16468 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
16469 Use GET_MODE_UNIT_PRECISION and remove redundant test for
16471 * combine.c (set_nonzero_bits_and_sign_copies): Use
16472 is_a <scalar_int_mode>.
16473 (find_split_point): Likewise.
16474 (combine_simplify_rtx): Likewise.
16475 (simplify_logical): Likewise.
16476 (expand_compound_operation): Likewise.
16477 (expand_field_assignment): Likewise.
16478 (make_compound_operation): Likewise.
16479 (extended_count): Likewise.
16480 (change_zero_ext): Likewise.
16481 (simplify_comparison): Likewise.
16482 * dwarf2out.c (scompare_loc_descriptor): Likewise.
16483 (ucompare_loc_descriptor): Likewise.
16484 (minmax_loc_descriptor): Likewise.
16485 (mem_loc_descriptor): Likewise.
16486 (loc_descriptor): Likewise.
16487 * expmed.c (init_expmed_one_mode): Likewise.
16488 * lra-constraints.c (lra_constraint_offset): Likewise.
16489 * optabs.c (prepare_libcall_arg): Likewise.
16490 * postreload.c (move2add_note_store): Likewise.
16491 * reload.c (operands_match_p): Likewise.
16492 * rtl.h (load_extend_op): Likewise.
16493 * rtlhooks.c (gen_lowpart_general): Likewise.
16494 * simplify-rtx.c (simplify_truncation): Likewise.
16495 (simplify_unary_operation_1): Likewise.
16496 (simplify_binary_operation_1): Likewise.
16497 (simplify_const_binary_operation): Likewise.
16498 (simplify_const_relational_operation): Likewise.
16499 (simplify_subreg): Likewise.
16500 * stor-layout.c (bitwise_mode_for_mode): Likewise.
16501 * var-tracking.c (adjust_mems): Likewise.
16502 (prepare_call_arguments): Likewise.
16504 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16505 Alan Hayward <alan.hayward@arm.com>
16506 David Sherwood <david.sherwood@arm.com>
16508 * machmode.h (is_int_mode): New fuction.
16509 * combine.c (find_split_point): Use it.
16510 (combine_simplify_rtx): Likewise.
16511 (simplify_if_then_else): Likewise.
16512 (simplify_set): Likewise.
16513 (simplify_shift_const_1): Likewise.
16514 (simplify_comparison): Likewise.
16515 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
16516 * cse.c (notreg_cost): Likewise.
16517 (cse_insn): Likewise.
16518 * cselib.c (cselib_lookup_1): Likewise.
16519 * dojump.c (do_jump_1): Likewise.
16520 (do_compare_rtx_and_jump): Likewise.
16521 * dse.c (get_call_args): Likewise.
16522 * dwarf2out.c (rtl_for_decl_init): Likewise.
16523 (native_encode_initializer): Likewise.
16524 * expmed.c (emit_store_flag_1): Likewise.
16525 (emit_store_flag): Likewise.
16526 * expr.c (convert_modes): Likewise.
16527 (store_field): Likewise.
16528 (expand_expr_real_1): Likewise.
16529 * fold-const.c (fold_read_from_constant_string): Likewise.
16530 * gimple-ssa-sprintf.c (get_format_string): Likewise.
16531 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
16532 * optabs.c (expand_binop): Likewise.
16533 (expand_unop): Likewise.
16534 (expand_abs_nojump): Likewise.
16535 (expand_one_cmpl_abs_nojump): Likewise.
16536 * simplify-rtx.c (mode_signbit_p): Likewise.
16537 (val_signbit_p): Likewise.
16538 (val_signbit_known_set_p): Likewise.
16539 (val_signbit_known_clear_p): Likewise.
16540 (simplify_relational_operation_1): Likewise.
16541 * tree.c (vector_type_mode): Likewise.
16543 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16544 Alan Hayward <alan.hayward@arm.com>
16545 David Sherwood <david.sherwood@arm.com>
16547 * machmode.h (smallest_mode_for_size): Fix formatting.
16548 (smallest_int_mode_for_size): New function.
16549 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
16550 instead of smallest_mode_for_size.
16551 * combine.c (make_extraction): Likewise.
16552 * config/arc/arc.c (arc_expand_movmem): Likewise.
16553 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
16554 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
16555 * config/s390/s390.c (s390_expand_insv): Likewise.
16556 * config/sparc/sparc.c (assign_int_registers): Likewise.
16557 * config/spu/spu.c (spu_function_value): Likewise.
16558 (spu_function_arg): Likewise.
16559 * coverage.c (get_gcov_type): Likewise.
16560 (get_gcov_unsigned_t): Likewise.
16561 * dse.c (find_shift_sequence): Likewise.
16562 * expmed.c (store_bit_field_1): Likewise.
16563 * expr.c (convert_move): Likewise.
16564 (store_field): Likewise.
16565 * internal-fn.c (expand_arith_overflow): Likewise.
16566 * optabs-query.c (get_best_extraction_insn): Likewise.
16567 * optabs.c (expand_twoval_binop_libfunc): Likewise.
16568 * stor-layout.c (layout_type): Likewise.
16569 (initialize_sizetypes): Likewise.
16570 * targhooks.c (default_get_mask_mode): Likewise.
16571 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
16573 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16574 Alan Hayward <alan.hayward@arm.com>
16575 David Sherwood <david.sherwood@arm.com>
16577 * machmode.h (opt_mode::else_blk): New function.
16578 (int_mode_for_mode): Declare.
16579 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
16580 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
16582 * cfgexpand.c (expand_debug_expr): Likewise.
16583 * combine.c (gen_lowpart_or_truncate): Likewise.
16584 (gen_lowpart_for_combine): Likewise.
16585 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
16586 * config/avr/avr.c (avr_to_int_mode): Likewise.
16587 (avr_out_plus_1): Likewise.
16588 (avr_out_plus): Likewise.
16589 (avr_out_round): Likewise.
16590 * config/i386/i386.c (ix86_split_to_parts): Likewise.
16591 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
16592 (s390_expand_vcond): Likewise.
16593 * config/spu/spu.c (spu_split_immediate): Likewise.
16594 (spu_expand_mov): Likewise.
16595 * dse.c (get_stored_val): Likewise.
16596 * expmed.c (store_bit_field_1): Likewise.
16597 (convert_extracted_bit_field): Use int_mode_for_mode instead of
16599 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
16600 (extract_low_bits): Likewise.
16601 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
16602 handling rather than repeating the check.
16603 (emit_group_store): Likewise.
16604 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
16605 * optabs.c (expand_absneg_bit): Likewise.
16606 (expand_copysign_absneg): Likewise.
16607 (expand_copysign_bit): Likewise.
16608 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
16609 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
16610 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
16611 * var-tracking.c (prepare_call_arguments): Likewise.
16612 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
16613 int_mode_for_mode instead of mode_for_size.
16614 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
16616 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16617 Alan Hayward <alan.hayward@arm.com>
16618 David Sherwood <david.sherwood@arm.com>
16620 * machmode.h (int_mode_for_size): New function.
16621 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
16622 instead of mode_for_size.
16623 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
16625 * combine.c (expand_field_assignment): Use int_mode_for_size
16626 instead of mode_for_size.
16627 (make_extraction): Likewise.
16628 (simplify_shift_const_1): Likewise.
16629 (simplify_comparison): Likewise.
16630 * dojump.c (do_jump): Likewise.
16631 * dwarf2out.c (mem_loc_descriptor): Likewise.
16632 * emit-rtl.c (init_derived_machine_modes): Likewise.
16633 * expmed.c (flip_storage_order): Likewise.
16634 (convert_extracted_bit_field): Likewise.
16635 * expr.c (copy_blkmode_from_reg): Likewise.
16636 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
16637 * internal-fn.c (expand_mul_overflow): Likewise.
16638 * lower-subreg.c (simple_move): Likewise.
16639 * optabs-libfuncs.c (init_optabs): Likewise.
16640 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16641 * tree.c (vector_type_mode): Likewise.
16642 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
16643 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
16644 * tree-vect-generic.c (expand_vector_parallel): Likewise.
16645 * tree-vect-stmts.c (vectorizable_load): Likewise.
16646 (vectorizable_store): Likewise.
16648 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16649 Alan Hayward <alan.hayward@arm.com>
16650 David Sherwood <david.sherwood@arm.com>
16652 * coretypes.h (pod_mode): New type.
16653 (scalar_int_mode_pod): New typedef.
16654 * machmode.h (pod_mode): New class.
16655 (int_n_data_t::m): Change type to scalar_int_mode_pod.
16656 * genmodes.c (emit_mode_int_n): Update accordingly.
16657 * lower-subreg.h (target_lower_subreg): Change type to
16658 scalar_int_mode_pod.
16659 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
16660 scalar_int_mode_pod.
16662 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16663 Alan Hayward <alan.hayward@arm.com>
16664 David Sherwood <david.sherwood@arm.com>
16666 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
16667 machine_mode to scalar_int_mode.
16668 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
16669 (rs6000_option_override_internal): Remove cast to int.
16670 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
16671 machine_mode to scalar_int_mode.
16672 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
16673 (rs6000_option_override_internal): Remove cast to int.
16674 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
16675 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
16677 * config/s390/s390.c (s390_expand_builtin): Likewise.
16678 * coretypes.h (scalar_int_mode): New type.
16679 (opt_scalar_int_mode): New typedef.
16680 * machmode.h (scalar_int_mode): New class.
16681 (scalar_int_mode::includes_p): New function.
16682 (byte_mode): Change type to scalar_int_mode.
16683 (word_mode): Likewise.
16684 (ptr_mode): Likewise.
16685 * emit-rtl.c (byte_mode): Likewise.
16686 (word_mode): Likewise.
16687 (ptr_mode): Likewise.
16688 (init_derived_machine_modes): Update accordingly.
16689 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
16690 and MODE_PARTIAL_INT.
16691 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
16692 opt_scalar_int_mode.
16694 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16695 Alan Hayward <alan.hayward@arm.com>
16696 David Sherwood <david.sherwood@arm.com>
16698 * target.def (libgcc_floating_mode_supported_p): Take a
16700 * doc/tm.texi: Regenerate.
16701 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
16703 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16704 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
16707 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16708 Alan Hayward <alan.hayward@arm.com>
16709 David Sherwood <david.sherwood@arm.com>
16711 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
16712 * doc/tm.texi: Regenerate.
16713 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
16714 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
16715 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
16716 * targhooks.h (default_floatn_mode): Likewise.
16717 * targhooks.c (default_floatn_mode): Likewise.
16718 * tree.c (build_common_tree_nodes): Update accordingly.
16720 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16721 Alan Hayward <alan.hayward@arm.com>
16722 David Sherwood <david.sherwood@arm.com>
16724 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
16725 (mode_iterator::iterate_p): Likewise.
16726 (mode_iterator::get_wider): Likewise.
16727 * expr.c (init_expr_target): Use opt_scalar_float_mode.
16729 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16730 Alan Hayward <alan.hayward@arm.com>
16731 David Sherwood <david.sherwood@arm.com>
16733 * coretypes.h (opt_scalar_float_mode): New typedef.
16734 * machmode.h (float_mode_for_size): New function.
16735 * emit-rtl.c (double_mode): Delete.
16736 (init_emit_once): Use float_mode_for_size.
16737 * stor-layout.c (layout_type): Likewise.
16738 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
16740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16741 Alan Hayward <alan.hayward@arm.com>
16742 David Sherwood <david.sherwood@arm.com>
16744 * output.h (assemble_real): Take a scalar_float_mode.
16745 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
16746 * config/arm/arm.md (consttable_4): Likewise.
16747 (consttable_8): Likewise.
16748 (consttable_16): Likewise.
16749 * config/mips/mips.md (consttable_float): Likewise.
16750 * config/s390/s390.c (s390_output_pool_entry): Likewise.
16751 * varasm.c (assemble_real): Take a scalar_float_mode.
16752 (output_constant_pool_2): Update accordingly.
16753 (output_constant): Likewise.
16755 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16756 Alan Hayward <alan.hayward@arm.com>
16757 David Sherwood <david.sherwood@arm.com>
16759 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
16760 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
16761 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
16762 (native_encode_real): Likewise.
16763 (native_interpret_real): Likewise.
16764 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
16765 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16768 Alan Hayward <alan.hayward@arm.com>
16769 David Sherwood <david.sherwood@arm.com>
16771 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
16772 <scalar_float_mode>. Simplify.
16773 (gen_extend_conv_libfunc): Likewise.
16775 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16776 Alan Hayward <alan.hayward@arm.com>
16777 David Sherwood <david.sherwood@arm.com>
16779 * coretypes.h (scalar_float_mode): New type.
16780 * machmode.h (mode_traits::from_int): Use machine_mode if
16781 USE_ENUM_MODES is defined.
16782 (is_a): New function.
16784 (dyn_cast): Likewise.
16785 (scalar_float_mode): New class.
16786 (scalar_float_mode::includes_p): New function.
16787 (is_float_mode): Likewise.
16788 * gdbhooks.py (MachineModePrinter): New class.
16789 (build_pretty_printer): Use it for scalar_float_mode.
16790 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
16791 (format_helper::format_helper): Turn into a template.
16792 * genmodes.c (get_mode_class): New function.
16793 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
16794 or machine_mode if none.
16795 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
16796 as_a <scalar_float_mode>.
16797 * dwarf2out.c (mem_loc_descriptor): Likewise.
16798 (insert_float): Likewise.
16799 (add_const_value_attribute): Likewise.
16800 * simplify-rtx.c (simplify_immed_subreg): Likewise.
16801 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
16802 (expand_unop): Update accordingly.
16803 (expand_abs_nojump): Likewise.
16804 (expand_copysign_absneg): Take a scalar_float_mode.
16805 (expand_copysign_bit): Likewise.
16806 (expand_copysign): Update accordingly.
16808 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16809 Alan Hayward <alan.hayward@arm.com>
16810 David Sherwood <david.sherwood@arm.com>
16812 * coretypes.h (opt_mode): New class.
16813 * machmode.h (opt_mode): Likewise.
16814 (opt_mode::else_void): New function.
16815 (opt_mode::require): Likewise.
16816 (opt_mode::exists): Likewise.
16817 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
16818 (GET_MODE_2XWIDER_MODE): Likewise.
16819 (mode_iterator::get_wider): Update accordingly.
16820 (mode_iterator::get_2xwider): Likewise.
16821 (mode_iterator::get_known_wider): Likewise, turning into a template.
16822 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
16823 forcing a wider mode to exist.
16824 * config/cr16/cr16.h (LONG_REG_P): Likewise.
16825 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16826 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
16827 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
16828 * lower-subreg.c (init_lower_subreg): Likewise.
16829 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
16830 on the final iteration.
16831 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
16832 a wider mode exists before asking for a move pattern.
16833 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
16834 forcing a wider mode to exist.
16835 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
16836 returning false if no such mode exists.
16837 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
16838 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
16839 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
16840 Avoid checking for a MODE_INT if we already know the mode is not a
16842 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
16843 forcing a wider mode to exist.
16844 (expmed_mult_highpart_optab): Likewise.
16845 (expmed_mult_highpart): Likewise.
16846 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
16848 * lto-streamer-in.c (lto_input_mode_table): Likewise.
16849 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
16850 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16851 * internal-fn.c (expand_mul_overflow): Update use of
16852 GET_MODE_2XWIDER_MODE.
16853 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16854 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
16855 GET_MODE_WIDER_MODE.
16856 (convert_plusminus_to_widen): Likewise.
16857 * tree-switch-conversion.c (array_value_type): Likewise.
16858 * var-tracking.c (emit_note_insn_var_location): Likewise.
16859 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16860 Return false inside rather than outside the loop if no wider mode
16862 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
16863 and GET_MODE_2XWIDER_MODE
16864 (can_compare_p): Use else_void.
16865 * gdbhooks.py (OptMachineModePrinter): New class.
16866 (build_pretty_printer): Use it for opt_mode.
16868 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16869 Alan Hayward <alan.hayward@arm.com>
16870 David Sherwood <david.sherwood@arm.com>
16872 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
16873 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
16875 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16876 Alan Hayward <alan.hayward@arm.com>
16877 David Sherwood <david.sherwood@arm.com>
16879 * machmode.h (mode_traits): New structure.
16880 (get_narrowest_mode): New function.
16881 (mode_iterator::start): Likewise.
16882 (mode_iterator::iterate_p): Likewise.
16883 (mode_iterator::get_wider): Likewise.
16884 (mode_iterator::get_known_wider): Likewise.
16885 (mode_iterator::get_2xwider): Likewise.
16886 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
16887 (FOR_EACH_MODE): Likewise.
16888 (FOR_EACH_MODE_FROM): Likewise.
16889 (FOR_EACH_MODE_UNTIL): Likewise.
16890 (FOR_EACH_WIDER_MODE): Likewise.
16891 (FOR_EACH_2XWIDER_MODE): Likewise.
16892 * builtins.c (expand_builtin_strlen): Use new mode iterators.
16893 * combine.c (simplify_comparison): Likewise
16894 * config/i386/i386.c (type_natural_mode): Likewise.
16895 * cse.c (cse_insn): Likewise.
16896 * dse.c (find_shift_sequence): Likewise.
16897 * emit-rtl.c (init_derived_machine_modes): Likewise.
16898 (init_emit_once): Likewise.
16899 * explow.c (hard_function_value): Likewise.
16900 * expmed.c (extract_fixed_bit_field_1): Likewise.
16901 (extract_bit_field_1): Likewise.
16902 (expand_divmod): Likewise.
16903 (emit_store_flag_1): Likewise.
16904 * expr.c (init_expr_target): Likewise.
16905 (convert_move): Likewise.
16906 (alignment_for_piecewise_move): Likewise.
16907 (widest_int_mode_for_size): Likewise.
16908 (emit_block_move_via_movmem): Likewise.
16909 (copy_blkmode_to_reg): Likewise.
16910 (set_storage_via_setmem): Likewise.
16911 (compress_float_constant): Likewise.
16912 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16913 * optabs-query.c (get_best_extraction_insn): Likewise.
16914 * optabs.c (expand_binop): Likewise.
16915 (expand_twoval_unop): Likewise.
16916 (expand_twoval_binop): Likewise.
16917 (widen_leading): Likewise.
16918 (widen_bswap): Likewise.
16919 (expand_parity): Likewise.
16920 (expand_unop): Likewise.
16921 (prepare_cmp_insn): Likewise.
16922 (prepare_float_lib_cmp): Likewise.
16923 (expand_float): Likewise.
16924 (expand_fix): Likewise.
16925 (expand_sfix_optab): Likewise.
16926 * postreload.c (move2add_use_add2_insn): Likewise.
16927 * reg-stack.c (reg_to_stack): Likewise.
16928 * reginfo.c (choose_hard_reg_mode): Likewise.
16929 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16930 * stor-layout.c (mode_for_size): Likewise.
16931 (smallest_mode_for_size): Likewise.
16932 (mode_for_vector): Likewise.
16933 (finish_bitfield_representative): Likewise.
16934 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
16935 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
16936 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16937 * var-tracking.c (prepare_call_arguments): Likewise.
16939 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16940 Alan Hayward <alan.hayward@arm.com>
16941 David Sherwood <david.sherwood@arm.com>
16943 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
16944 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
16945 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
16946 * machmode.h (mode_size): Move earlier in file.
16947 (mode_precision): Likewise.
16948 (mode_inner): Likewise.
16949 (mode_nunits): Likewise.
16950 (mode_unit_size): Likewise.
16951 (unit_unit_precision): Likewise.
16952 (mode_wider): Likewise.
16953 (mode_2xwider): Likewise.
16954 (machine_mode): New class.
16955 (mode_to_bytes): New function.
16956 (mode_to_bits): Likewise.
16957 (mode_to_precision): Likewise.
16958 (mode_to_inner): Likewise.
16959 (mode_to_unit_size): Likewise.
16960 (mode_to_unit_precision): Likewise.
16961 (mode_to_nunits): Likewise.
16962 (GET_MODE_SIZE): Use mode_to_bytes.
16963 (GET_MODE_BITSIZE): Use mode_to_bits.
16964 (GET_MODE_PRECISION): Use mode_to_precision.
16965 (GET_MODE_INNER): Use mode_to_inner.
16966 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
16967 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
16968 (GET_MODE_NUNITS): Use mode_to_nunits.
16969 * system.h (ALWAYS_INLINE): New macro.
16970 * config/powerpcspe/powerpcspe-c.c
16971 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
16972 int for arg1_mode and arg2_mode.
16974 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16975 Alan Hayward <alan.hayward@arm.com>
16976 David Sherwood <david.sherwood@arm.com>
16978 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16979 Prefix mode names with E_ in case statements.
16980 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16981 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
16982 (aarch64_split_simd_move): Likewise.
16983 (aarch64_gen_storewb_pair): Likewise.
16984 (aarch64_gen_loadwb_pair): Likewise.
16985 (aarch64_gen_store_pair): Likewise.
16986 (aarch64_gen_load_pair): Likewise.
16987 (aarch64_get_condition_code_1): Likewise.
16988 (aarch64_constant_pool_reload_icode): Likewise.
16989 (get_rsqrte_type): Likewise.
16990 (get_rsqrts_type): Likewise.
16991 (get_recpe_type): Likewise.
16992 (get_recps_type): Likewise.
16993 (aarch64_gimplify_va_arg_expr): Likewise.
16994 (aarch64_simd_container_mode): Likewise.
16995 (aarch64_emit_load_exclusive): Likewise.
16996 (aarch64_emit_store_exclusive): Likewise.
16997 (aarch64_expand_compare_and_swap): Likewise.
16998 (aarch64_gen_atomic_cas): Likewise.
16999 (aarch64_emit_bic): Likewise.
17000 (aarch64_emit_atomic_swap): Likewise.
17001 (aarch64_emit_atomic_load_op): Likewise.
17002 (aarch64_evpc_trn): Likewise.
17003 (aarch64_evpc_uzp): Likewise.
17004 (aarch64_evpc_zip): Likewise.
17005 (aarch64_evpc_ext): Likewise.
17006 (aarch64_evpc_rev): Likewise.
17007 (aarch64_evpc_dup): Likewise.
17008 (aarch64_gen_ccmp_first): Likewise.
17009 (aarch64_gen_ccmp_next): Likewise.
17010 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17011 (alpha_emit_xfloating_libcall): Likewise.
17012 (emit_insxl): Likewise.
17013 (alpha_arg_type): Likewise.
17014 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
17015 (arc_preferred_simd_mode): Likewise.
17016 (arc_secondary_reload): Likewise.
17017 (get_arc_condition_code): Likewise.
17018 (arc_print_operand): Likewise.
17019 (arc_legitimate_constant_p): Likewise.
17020 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17021 * config/arc/arc.md (casesi_load): Likewise.
17022 (casesi_compact_jump): Likewise.
17023 * config/arc/predicates.md (proper_comparison_operator): Likewise.
17024 (cc_use_register): Likewise.
17025 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17026 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17027 (arm_init_iwmmxt_builtins): Likewise.
17028 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
17029 (neon_expand_vector_init): Likewise.
17030 (arm_attr_length_move_neon): Likewise.
17031 (maybe_get_arm_condition_code): Likewise.
17032 (arm_emit_vector_const): Likewise.
17033 (arm_preferred_simd_mode): Likewise.
17034 (arm_output_iwmmxt_tinsr): Likewise.
17035 (thumb1_output_casesi): Likewise.
17036 (thumb2_output_casesi): Likewise.
17037 (arm_emit_load_exclusive): Likewise.
17038 (arm_emit_store_exclusive): Likewise.
17039 (arm_expand_compare_and_swap): Likewise.
17040 (arm_evpc_neon_vuzp): Likewise.
17041 (arm_evpc_neon_vzip): Likewise.
17042 (arm_evpc_neon_vrev): Likewise.
17043 (arm_evpc_neon_vtrn): Likewise.
17044 (arm_evpc_neon_vext): Likewise.
17045 (arm_validize_comparison): Likewise.
17046 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
17047 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
17048 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
17049 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
17050 (c6x_preferred_simd_mode): Likewise.
17051 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
17052 (epiphany_rtx_costs): Likewise.
17053 * config/epiphany/predicates.md (proper_comparison_operator):
17055 * config/frv/frv.c (condexec_memory_operand): Likewise.
17056 (frv_emit_move): Likewise.
17057 (output_move_single): Likewise.
17058 (output_condmove_single): Likewise.
17059 (frv_hard_regno_mode_ok): Likewise.
17060 (frv_matching_accg_mode): Likewise.
17061 * config/h8300/h8300.c (split_adds_subs): Likewise.
17062 (h8300_rtx_costs): Likewise.
17063 (h8300_print_operand): Likewise.
17064 (compute_mov_length): Likewise.
17065 (output_logical_op): Likewise.
17066 (compute_logical_op_length): Likewise.
17067 (compute_logical_op_cc): Likewise.
17068 (h8300_shift_needs_scratch_p): Likewise.
17069 (output_a_shift): Likewise.
17070 (compute_a_shift_length): Likewise.
17071 (compute_a_shift_cc): Likewise.
17072 (expand_a_rotate): Likewise.
17073 (output_a_rotate): Likewise.
17074 * config/i386/i386.c (classify_argument): Likewise.
17075 (function_arg_advance_32): Likewise.
17076 (function_arg_32): Likewise.
17077 (function_arg_64): Likewise.
17078 (function_value_64): Likewise.
17079 (ix86_gimplify_va_arg): Likewise.
17080 (ix86_legitimate_constant_p): Likewise.
17081 (put_condition_code): Likewise.
17082 (split_double_mode): Likewise.
17083 (ix86_avx256_split_vector_move_misalign): Likewise.
17084 (ix86_expand_vector_logical_operator): Likewise.
17085 (ix86_split_idivmod): Likewise.
17086 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
17087 (ix86_build_const_vector): Likewise.
17088 (ix86_build_signbit_mask): Likewise.
17089 (ix86_match_ccmode): Likewise.
17090 (ix86_cc_modes_compatible): Likewise.
17091 (ix86_expand_branch): Likewise.
17092 (ix86_expand_sse_cmp): Likewise.
17093 (ix86_expand_sse_movcc): Likewise.
17094 (ix86_expand_int_sse_cmp): Likewise.
17095 (ix86_expand_vec_perm_vpermi2): Likewise.
17096 (ix86_expand_vec_perm): Likewise.
17097 (ix86_expand_sse_unpack): Likewise.
17098 (ix86_expand_int_addcc): Likewise.
17099 (ix86_split_to_parts): Likewise.
17100 (ix86_vectorize_builtin_gather): Likewise.
17101 (ix86_vectorize_builtin_scatter): Likewise.
17102 (avx_vpermilp_parallel): Likewise.
17103 (inline_memory_move_cost): Likewise.
17104 (ix86_tieable_integer_mode_p): Likewise.
17105 (x86_maybe_negate_const_int): Likewise.
17106 (ix86_expand_vector_init_duplicate): Likewise.
17107 (ix86_expand_vector_init_one_nonzero): Likewise.
17108 (ix86_expand_vector_init_one_var): Likewise.
17109 (ix86_expand_vector_init_concat): Likewise.
17110 (ix86_expand_vector_init_interleave): Likewise.
17111 (ix86_expand_vector_init_general): Likewise.
17112 (ix86_expand_vector_set): Likewise.
17113 (ix86_expand_vector_extract): Likewise.
17114 (emit_reduc_half): Likewise.
17115 (ix86_emit_i387_round): Likewise.
17116 (ix86_mangle_type): Likewise.
17117 (ix86_expand_round_sse4): Likewise.
17118 (expand_vec_perm_blend): Likewise.
17119 (canonicalize_vector_int_perm): Likewise.
17120 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
17121 (expand_vec_perm_1): Likewise.
17122 (expand_vec_perm_interleave3): Likewise.
17123 (expand_vec_perm_even_odd_pack): Likewise.
17124 (expand_vec_perm_even_odd_1): Likewise.
17125 (expand_vec_perm_broadcast_1): Likewise.
17126 (ix86_vectorize_vec_perm_const_ok): Likewise.
17127 (ix86_expand_vecop_qihi): Likewise.
17128 (ix86_expand_mul_widen_hilo): Likewise.
17129 (ix86_expand_sse2_abs): Likewise.
17130 (ix86_expand_pextr): Likewise.
17131 (ix86_expand_pinsr): Likewise.
17132 (ix86_preferred_simd_mode): Likewise.
17133 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
17134 * config/i386/sse.md (*andnot<mode>3): Likewise.
17135 (<mask_codefor><code><mode>3<mask_name>): Likewise.
17136 (*<code><mode>3): Likewise.
17137 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
17138 (ia64_expand_atomic_op): Likewise.
17139 (ia64_arg_type): Likewise.
17140 (ia64_mode_to_int): Likewise.
17141 (ia64_scalar_mode_supported_p): Likewise.
17142 (ia64_vector_mode_supported_p): Likewise.
17143 (expand_vec_perm_broadcast): Likewise.
17144 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
17145 (iq2000_function_arg_advance): Likewise.
17146 (iq2000_function_arg): Likewise.
17147 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
17148 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
17149 (m68k_libcall_value): Likewise.
17150 (m68k_function_value): Likewise.
17151 (sched_attr_op_type): Likewise.
17152 * config/mcore/mcore.c (mcore_output_move): Likewise.
17153 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
17155 (microblaze_function_arg): Likewise.
17156 * config/mips/mips.c (mips16_build_call_stub): Likewise.
17157 (mips_print_operand): Likewise.
17158 (mips_mode_ok_for_mov_fmt_p): Likewise.
17159 (mips_vector_mode_supported_p): Likewise.
17160 (mips_preferred_simd_mode): Likewise.
17161 (mips_expand_vpc_loongson_even_odd): Likewise.
17162 (mips_expand_vec_unpack): Likewise.
17163 (mips_expand_vi_broadcast): Likewise.
17164 (mips_expand_vector_init): Likewise.
17165 (mips_expand_vec_reduc): Likewise.
17166 (mips_expand_msa_cmp): Likewise.
17167 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
17168 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
17169 (cc_flags_for_mode): Likewise.
17170 * config/msp430/msp430.c (msp430_print_operand): Likewise.
17171 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
17172 (nds32_output_casesi_pc_relative): Likewise.
17173 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17174 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
17175 (nvptx_gen_unpack): Likewise.
17176 (nvptx_gen_pack): Likewise.
17177 (nvptx_gen_shuffle): Likewise.
17178 (nvptx_gen_wcast): Likewise.
17179 (nvptx_preferred_simd_mode): Likewise.
17180 * config/pa/pa.c (pa_secondary_reload): Likewise.
17181 * config/pa/predicates.md (base14_operand): Likewise.
17182 * config/powerpcspe/powerpcspe-c.c
17183 (altivec_resolve_overloaded_builtin): Likewise.
17184 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
17186 (rs6000_preferred_simd_mode): Likewise.
17187 (output_vec_const_move): Likewise.
17188 (rs6000_expand_vector_extract): Likewise.
17189 (rs6000_split_vec_extract_var): Likewise.
17190 (reg_offset_addressing_ok_p): Likewise.
17191 (rs6000_legitimate_offset_address_p): Likewise.
17192 (rs6000_legitimize_address): Likewise.
17193 (rs6000_emit_set_const): Likewise.
17194 (rs6000_const_vec): Likewise.
17195 (rs6000_emit_move): Likewise.
17196 (spe_build_register_parallel): Likewise.
17197 (rs6000_darwin64_record_arg_recurse): Likewise.
17198 (swap_selector_for_mode): Likewise.
17199 (spe_init_builtins): Likewise.
17200 (paired_init_builtins): Likewise.
17201 (altivec_init_builtins): Likewise.
17202 (do_load_for_compare): Likewise.
17203 (rs6000_generate_compare): Likewise.
17204 (rs6000_expand_float128_convert): Likewise.
17205 (emit_load_locked): Likewise.
17206 (emit_store_conditional): Likewise.
17207 (rs6000_output_function_epilogue): Likewise.
17208 (rs6000_handle_altivec_attribute): Likewise.
17209 (rs6000_function_value): Likewise.
17210 (emit_fusion_gpr_load): Likewise.
17211 (emit_fusion_p9_load): Likewise.
17212 (emit_fusion_p9_store): Likewise.
17213 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
17214 (fusion_gpr_mem_load): Likewise.
17215 (fusion_addis_mem_combo_load): Likewise.
17216 (fusion_addis_mem_combo_store): Likewise.
17217 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
17218 (fusion_gpr_mem_load): Likewise.
17219 (fusion_addis_mem_combo_load): Likewise.
17220 (fusion_addis_mem_combo_store): Likewise.
17221 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17223 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
17224 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
17225 (rs6000_preferred_simd_mode): Likewise.
17226 (output_vec_const_move): Likewise.
17227 (rs6000_expand_vector_extract): Likewise.
17228 (rs6000_split_vec_extract_var): Likewise.
17229 (reg_offset_addressing_ok_p): Likewise.
17230 (rs6000_legitimate_offset_address_p): Likewise.
17231 (rs6000_legitimize_address): Likewise.
17232 (rs6000_emit_set_const): Likewise.
17233 (rs6000_const_vec): Likewise.
17234 (rs6000_emit_move): Likewise.
17235 (rs6000_darwin64_record_arg_recurse): Likewise.
17236 (swap_selector_for_mode): Likewise.
17237 (paired_init_builtins): Likewise.
17238 (altivec_init_builtins): Likewise.
17239 (rs6000_expand_float128_convert): Likewise.
17240 (emit_load_locked): Likewise.
17241 (emit_store_conditional): Likewise.
17242 (rs6000_output_function_epilogue): Likewise.
17243 (rs6000_handle_altivec_attribute): Likewise.
17244 (rs6000_function_value): Likewise.
17245 (emit_fusion_gpr_load): Likewise.
17246 (emit_fusion_p9_load): Likewise.
17247 (emit_fusion_p9_store): Likewise.
17248 * config/rx/rx.c (rx_gen_move_template): Likewise.
17249 (flags_from_mode): Likewise.
17250 * config/s390/predicates.md (s390_alc_comparison): Likewise.
17251 (s390_slb_comparison): Likewise.
17252 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
17253 (s390_vector_mode_supported_p): Likewise.
17254 (s390_cc_modes_compatible): Likewise.
17255 (s390_match_ccmode_set): Likewise.
17256 (s390_canonicalize_comparison): Likewise.
17257 (s390_emit_compare_and_swap): Likewise.
17258 (s390_branch_condition_mask): Likewise.
17259 (s390_rtx_costs): Likewise.
17260 (s390_secondary_reload): Likewise.
17261 (__SECONDARY_RELOAD_CASE): Likewise.
17262 (s390_expand_cs): Likewise.
17263 (s390_preferred_simd_mode): Likewise.
17264 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
17265 * config/sh/sh.c (sh_print_operand): Likewise.
17266 (dump_table): Likewise.
17267 (sh_secondary_reload): Likewise.
17268 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17269 * config/sh/sh.md (casesi_worker_1): Likewise.
17270 (casesi_worker_2): Likewise.
17271 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
17272 (fcc_comparison_operator): Likewise.
17273 * config/sparc/sparc.c (sparc_expand_move): Likewise.
17274 (emit_soft_tfmode_cvt): Likewise.
17275 (sparc_preferred_simd_mode): Likewise.
17276 (output_cbranch): Likewise.
17277 (sparc_print_operand): Likewise.
17278 (sparc_expand_vec_perm_bmask): Likewise.
17279 (vector_init_bshuffle): Likewise.
17280 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17281 (spu_vector_mode_supported_p): Likewise.
17282 (spu_expand_insv): Likewise.
17283 (spu_emit_branch_or_set): Likewise.
17284 (spu_handle_vector_attribute): Likewise.
17285 (spu_builtin_splats): Likewise.
17286 (spu_builtin_extract): Likewise.
17287 (spu_builtin_promote): Likewise.
17288 (spu_expand_sign_extend): Likewise.
17289 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17290 (tilegx_simd_int): Likewise.
17291 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
17292 (tilepro_simd_int): Likewise.
17293 * config/v850/v850.c (const_double_split): Likewise.
17294 (v850_print_operand): Likewise.
17295 (ep_memory_offset): Likewise.
17296 * config/vax/vax.c (vax_rtx_costs): Likewise.
17297 (vax_output_int_move): Likewise.
17298 (vax_output_int_add): Likewise.
17299 (vax_output_int_subtract): Likewise.
17300 * config/visium/predicates.md (visium_branch_operator): Likewise.
17301 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
17302 (visium_print_operand_address): Likewise.
17303 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17304 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
17305 (xtensa_expand_conditional_branch): Likewise.
17306 (xtensa_copy_incoming_a7): Likewise.
17307 (xtensa_output_literal): Likewise.
17308 * dfp.c (decimal_real_maxval): Likewise.
17309 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
17311 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17312 Alan Hayward <alan.hayward@arm.com>
17313 David Sherwood <david.sherwood@arm.com>
17315 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
17316 (mode_nunits_inline): Likewise.
17317 (mode_inner_inline): Likewise.
17318 (mode_unit_size_inline): Likewise.
17319 (mode_unit_precision_inline): Likewise.
17320 (emit_insn_modes_h): Likewise. Also emit a #define of the
17322 (emit_mode_wider): Add an E_ prefix to mode names.
17323 (emit_mode_complex): Likewise.
17324 (emit_mode_inner): Likewise.
17325 (emit_mode_adjustments): Likewise.
17326 (emit_mode_int_n): Likewise.
17327 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
17328 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
17329 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
17330 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
17331 (CRC32_BUILTIN, ENTRY): Likewise.
17332 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
17333 (aarch64_pop_regs): Likewise.
17334 (aarch64_process_components): Likewise.
17335 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
17336 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
17337 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
17338 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
17339 * config/arm/arm.c (arm_init_libfuncs): Likewise.
17340 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
17342 * config/i386/i386-builtin.def (pcmpestr): Likewise.
17343 (pcmpistr): Likewise.
17344 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
17345 * config/mmix/mmix.c (mmix_output_condition): Likewise.
17346 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17348 * config/rl78/rl78.c (mduc_regs): Likewise.
17349 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17350 (htm_expand_builtin): Likewise.
17351 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
17352 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
17353 * config/xtensa/xtensa.c (print_operand): Likewise.
17354 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
17355 (NUM_MODE_VECTOR_INT): Likewise.
17356 * genoutput.c (null_operand): Likewise.
17357 (output_operand_data): Likewise.
17358 * genrecog.c (print_parameter_value): Likewise.
17359 * lra.c (debug_operand_data): Likewise.
17361 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17363 * dwarf2out.c (scompare_loc_descriptor_wide)
17364 (scompare_loc_descriptor_narrow): New functions, split out from...
17365 (scompare_loc_descriptor): ...here.
17366 * expmed.c (emit_store_flag_int): New function, split out from...
17367 (emit_store_flag): ...here.
17369 2017-08-30 Richard Biener <rguenther@suse.de>
17371 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
17372 (dwarf2out_early_finish): Move setting of AT_pubnames from
17373 early debug output to early finish.
17375 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17377 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
17378 and -mdata-region to the assembler.
17380 2017-08-30 Richard Biener <rguenther@suse.de>
17382 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
17384 (gen_subprogram_die): Add DW_AT_object_pointer only early.
17385 (dwarf2out_early_global_decl): Only generate a DIE for the
17386 abstract origin if it doesn't already exist or is a declaration DIE.
17387 (resolve_addr): Do not add the linkage name twice when
17388 generating a stub DIE for the DW_TAG_GNU_call_site target.
17390 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17392 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17393 Use machine_mode rather than int for arg1_mode.
17395 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
17398 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
17399 that the second argument of the built-in functions to unpack
17400 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
17401 switch statement instead a lot of if statements.
17402 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
17403 Allow 64-bit values to be in Altivec registers as well as
17404 traditional floating point registers.
17405 (pack<mode>, FMOVE128_VSX iterator): Likewise.
17407 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
17409 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
17410 MAX_REGS_PER_ADDRESS == 1.
17412 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
17414 * config/i386/i386.opt (flag_fentry): Do not init to -1.
17415 * config/i386/i386.c (ix86_option_override_internal): Simplify
17416 setting of opts->x_flag_entry.
17418 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17419 Jakub Jelinek <jakub@redhat.com>
17420 Richard Biener <rguenther@suse.de>
17422 PR tree-optimization/81503
17423 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
17424 folded constant fits in the target type; reorder tests for clarity.
17426 2017-08-29 Martin Liska <mliska@suse.cz>
17428 * passes.def: Include pass_lower_switch.
17429 * stmt.c (dump_case_nodes): Remove and move to
17430 tree-switch-conversion.
17431 (case_values_threshold): Likewise.
17432 (expand_switch_as_decision_tree_p): Likewise.
17433 (emit_case_decision_tree): Likewise.
17434 (expand_case): Likewise.
17435 (balance_case_nodes): Likewise.
17436 (node_has_low_bound): Likewise.
17437 (node_has_high_bound): Likewise.
17438 (node_is_bounded): Likewise.
17439 (emit_case_nodes): Likewise.
17440 (struct simple_case_node): New struct.
17441 (add_case_node): Remove.
17442 (emit_case_dispatch_table): Use vector instead of case_list.
17443 (reset_out_edges_aux): Remove.
17444 (compute_cases_per_edge): Likewise.
17445 (expand_case): Build list of simple_case_node.
17446 (expand_sjlj_dispatch_table): Use it.
17447 * tree-switch-conversion.c (struct case_node): Moved from
17448 stmt.c and adjusted.
17449 (emit_case_nodes): Likewise.
17450 (node_has_low_bound): Likewise.
17451 (node_has_high_bound): Likewise.
17452 (node_is_bounded): Likewise.
17453 (case_values_threshold): Likewise.
17454 (reset_out_edges_aux): Likewise.
17455 (compute_cases_per_edge): Likewise.
17456 (add_case_node): Likewise.
17457 (dump_case_nodes): Likewise.
17458 (balance_case_nodes): Likewise.
17459 (expand_switch_as_decision_tree_p): Likewise.
17460 (emit_jump): Likewise.
17461 (emit_case_decision_tree): Likewise.
17462 (try_switch_expansion): Likewise.
17463 (do_jump_if_equal): Likewise.
17464 (emit_cmp_and_jump_insns): Likewise.
17465 (fix_phi_operands_for_edge): New function.
17466 (record_phi_operand_mapping): Likewise.
17467 (class pass_lower_switch): New pass.
17468 (pass_lower_switch::execute): New function.
17469 (make_pass_lower_switch): Likewise.
17470 (conditional_probability):
17471 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
17472 * tree-pass.h: Add make_pass_lower_switch.
17474 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17477 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
17480 2017-08-29 Richard Biener <rguenther@suse.de>
17482 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
17483 we do not add a DW_AT_inline attribute twice.
17484 (gen_subprogram_die): Remove code setting DW_AT_inline on
17485 DECL_ABSTRACT_P nodes.
17487 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
17489 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
17490 calls to internal functions.
17491 (gimplify_modify_expr): Likewise.
17492 * tree-call-cdce.c (use_internal_fn): Likewise.
17493 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
17494 (convert_to_divmod): Set the nothrow flag.
17495 * tree-if-conv.c (predicate_mem_writes): Likewise.
17496 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
17497 (vectorizable_call): Likewise.
17498 (vectorizable_store): Likewise.
17499 (vectorizable_load): Likewise.
17500 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
17501 (vect_recog_mask_conversion_pattern): Likewise.
17503 2017-08-29 Martin Liska <mliska@suse.cz>
17506 * gcc.c (driver_handle_option): Add new argument.
17507 * opts-common.c (handle_option): Pass
17508 target_option_override_hook.
17509 * opts-global.c (lang_handle_option): Add new option.
17510 (set_default_handlers): Add new argument.
17511 (decode_options): Likewise.
17512 * opts.c (target_handle_option): Likewise.
17513 (common_handle_option): Call target_option_override_hook.
17514 * opts.h (struct cl_option_handler_func): Add hook for
17515 target option override.
17516 (struct cl_option_handlers): Likewise.
17517 (set_default_handlers): Add new argument.
17518 (decode_options): Likewise.
17519 (common_handle_option): Likewise.
17520 (target_handle_option): Likewise.
17521 * toplev.c (toplev::main): Pass targetm.target_option.override
17524 2017-08-29 Richard Biener <rguenther@suse.de>
17525 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
17527 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
17528 life to the active subtree.
17530 2017-08-28 Jeff Law <law@redhat.com>
17532 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
17533 derive_equivalences.
17534 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
17535 Code moved into....
17536 (edge_info::derive_equivalences): New private member function
17538 * tree-ssa-dom.c (class edge_info): Changed from a struct
17539 to a class. Add ctor/dtor, methods and data members.
17540 (edge_info::edge_info): Renamed from allocate_edge_info.
17541 Initialize additional members.
17542 (edge_info::~edge_info): New.
17543 (free_dom_edge_info): Delete the edge info.
17544 (record_edge_info): Use new class & associated member functions.
17545 Tighten forms for testing for edge equivalences.
17546 (record_temporary_equivalences): Iterate over the simple
17547 equivalences rather than assuming there's only one per edge.
17548 (cprop_into_successor_phis): Iterate over the simple
17549 equivalences rather than assuming there's only one per edge.
17550 (optimize_stmt): Use operand_equal_p rather than pointer
17551 equality for mini-DSE code.
17553 2017-08-28 Nathan Sidwell <nathan@acm.org>
17555 * gcc.c (execute): Fold SIGPIPE handling into switch
17556 statement. Adjust internal error message.
17558 2017-08-28 Richard Biener <rguenther@suse.de>
17561 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
17562 Do nothing for removed DIEs.
17564 2017-08-28 Richard Biener <rguenther@suse.de>
17566 PR tree-optimization/81977
17567 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
17570 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
17573 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
17574 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
17575 using targetm.gen_mem_thread_fence.
17577 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
17580 * config/i386/i386.md (*<btsc><mode>): Change operand 2
17581 predicate to register_operand. Reorder operands.
17582 (*btr<mode>): Ditto.
17583 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
17584 (*btr<mode>_mask): Ditto.
17586 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
17588 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
17589 * config/rs6000/xmmintrin.h: New file.
17590 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
17592 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17595 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
17596 parameter and_insn and return it.
17597 (recombine_lvx_pattern): Insert a copy to ensure availability of
17598 the base register of the copied masking operation at the point of
17599 the instruction replacement.
17600 (recombine_stvx_pattern): Likewise.
17602 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
17604 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
17605 undocumented switches.
17606 (-mpower9-dform-vector): Likewise.
17607 (-mpower9-dform): Likewise.
17608 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
17609 comments to delete references to -mpower9-dform* switches.
17610 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
17611 Delete reference to -mpower9-dform* switches, test for
17612 -mpower9-vector instead.
17613 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
17614 (OTHER_P9_VECTOR_MASKS): Likewise.
17615 (POWERPC_MASKS): Likewise.
17616 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
17617 tests against -mpower9-dform* to -mpower9-vector. Delete code
17618 that checked for -mpower9-dform* consistancy with other options.
17619 Add test for -mpower9-misc to enable other power9 switches.
17620 (rs6000_init_hard_regno_mode_ok): Likewise.
17621 (rs6000_option_override_internal): Likewise.
17622 (rs6000_emit_prologue): Likewise.
17623 (rs6000_emit_epilogue): Likewise.
17624 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
17625 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
17626 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
17628 (emit_fusion_p9_store): Likewise.
17629 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
17630 resetting these macros if the assembler does not support ISA 3.0
17632 (TARGET_P9_DFORM_VECTOR): Likewise.
17633 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
17634 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
17636 2017-08-25 Alan Modra <amodra@gmail.com>
17638 PR rtl-optimization/81747
17639 * cse.c (cse_extended_basic_block): Don't attempt to record
17640 equivalences for degenerate conditional branches that branch
17641 to their fall-through.
17643 2017-08-24 Martin Sebor <msebor@redhat.com>
17645 PR middle-end/81908
17646 * gimple-fold.c (size_must_be_zero_p): New function.
17647 (gimple_fold_builtin_memory_op): Call it.
17649 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
17651 * config/rs6000/mm_malloc.h: New file.
17653 2017-08-24 Bin Cheng <bin.cheng@arm.com>
17655 PR tree-optimization/81913
17656 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
17657 analysis when either IVs in condition can wrap.
17659 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
17661 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
17662 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
17664 2017-08-24 Richard Biener <rguenther@suse.de>
17667 * targhooks.c (default_target_can_inline_p): Properly
17668 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
17669 is present and always compare.
17670 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
17671 infer -mfpmath=sse from TARGET_SSE_P.
17672 (ix86_can_inline_p): Properly use target_option_default_node when
17673 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
17675 2017-08-24 Richard Biener <rguenther@suse.de>
17678 * dwarf2out.c (output_die): Handle flag_generate_offload like
17680 (output_comp_unit): Likewise.
17681 (gen_array_type_die): Likewise.
17682 (dwarf2out_early_finish): Likewise.
17683 (note_variable_value_in_expr): Likewise.
17684 (dwarf2out_finish): Likewise. Adjust assert.
17685 * cgraphunit.c (symbol_table::compile): Move setting of
17686 flag_generate_offload earlier ...
17687 (symbol_table::finalize_compilation_unit): ... here, before
17688 early debug finalization.
17690 2017-08-24 Richard Biener <rguenther@suse.de>
17692 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
17693 and ipa-fnsummary.h.
17694 (ix86_can_inline_p): When ix86_fpmath flags do not match
17695 check whether the callee uses FP math at all.
17697 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
17699 PR middle-end/81931
17700 * tree-ssanames.c (get_nonzero_bits): Use element_precision
17701 instead of TYPE_PRECISION.
17703 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
17704 Alan Hayward <alan.hayward@arm.com>
17705 David Sherwood <david.sherwood@arm.com>
17707 * combine.c (make_extraction): Use subreg_offset_from_lsb.
17709 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
17711 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
17713 (ix86_frame::stack_realign_allocate): New field.
17714 (struct machine_frame_state): Modify comments.
17715 (machine_frame_state::sp_realigned_fp_end): New field.
17716 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
17717 layout calculation.
17718 (sp_valid_at): Add assertion to assure no attempt to access invalid
17719 offset of a realigned stack.
17720 (fp_valid_at): Likewise.
17721 (choose_baseaddr): Modify comments.
17722 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
17723 ix86_expand_prologue.
17724 (ix86_expand_prologue): Modify stack realignment and allocation.
17725 (ix86_expand_epilogue): Modify comments.
17726 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
17727 avx2_runtime, avx512f, and avx512f_runtime.
17729 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
17731 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
17732 (mstackrealign): Do not init to -1.
17733 * config/i386/i386.c (ix86_option_override_internal):
17734 Check opts_set, not opts when setting default value of
17735 opts->x_ix86_force_align_arg_pointer.
17737 2017-08-23 Richard Biener <rguenther@suse.de>
17739 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
17740 lang_hooks.decl_printable_name.
17741 * print-rtl-function.c (print_rtx_function): Likewise.
17742 * tree-pretty-print.c (dump_function_header): Likewise.
17744 2017-08-23 Richard Biener <rguenther@suse.de>
17747 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
17748 -g0 at compile-time.
17750 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17752 PR middle-end/19706
17753 * doc/sourcebuild.texi (Other hardware attributes):
17756 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17758 PR middle-end/19706
17759 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
17760 Fix single-use check.
17762 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
17764 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
17766 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17768 * doc/install.texi: Modify to add more details on running selected
17771 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17773 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
17774 is combined with -mabi=ms.
17775 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
17778 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17780 PR tree-optimization/81488
17781 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
17782 and cached_basis fields.
17783 (MAX_SPREAD): New constant.
17784 (alloc_cand_and_find_basis): Initialize new fields.
17785 (clear_visited): New function.
17786 (create_phi_basis_1): Rename from create_phi_basis, set visited
17787 and cached_basis fields.
17788 (create_phi_basis): New wrapper function.
17789 (phi_add_costs_1): Rename from phi_add_costs, add spread
17790 parameter, set visited field, short-circuit when limits reached.
17791 (phi_add_costs): New wrapper function.
17792 (record_phi_increments_1): Rename from record_phi_increments, set
17794 (record_phi_increments): New wrapper function.
17795 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
17796 (phi_incr_cost): New wrapper function.
17797 (all_phi_incrs_profitable_1): Rename from
17798 all_phi_incrs_profitable, set visited field.
17799 (all_phi_incrs_profitable): New wrapper function.
17801 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17802 Alan Hayward <alan.hayward@arm.com>
17803 David Sherwood <david.sherwood@arm.com>
17805 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
17806 that takes the outer and inner modes.
17807 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
17808 comparison as the canonical test for a paradoxical subreg.
17809 * combine.c (simplify_set): Use paradoxical_subreg_p.
17810 (make_extraction): Likewise.
17811 (force_to_mode): Likewise.
17812 (rtx_equal_for_field_assignment_p): Likewise.
17813 (gen_lowpart_for_combine): Likewise.
17814 (simplify_comparison): Likewise.
17815 * cse.c (equiv_constant): Likewise.
17816 * expmed.c (store_bit_field_1): Likewise.
17817 * final.c (alter_subreg): Likewise.
17818 * fwprop.c (propagate_rtx): Likewise.
17819 (forward_propagate_subreg): Likewise.
17820 * ira-conflicts.c (ira_build_conflicts): Likewise.
17821 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
17822 * lra-constraints.c (curr_insn_transform): Likewise.
17823 (split_reg): Likewise.
17824 * lra-eliminations.c (move_plus_up): Likewise.
17825 (lra_eliminate_regs_1): Likewise.
17826 * recog.c (general_operand): Likewise.
17827 * ree.c (combine_reaching_defs): Likewise.
17828 * reload.c (push_reload): Likewise.
17829 (find_reloads): Likewise.
17830 * reload1.c (elimination_effects): Likewise.
17831 (compute_reload_subreg_offset): Likewise.
17832 (choose_reload_regs): Likewise.
17833 * rtlanal.c (subreg_lsb_1): Likewise.
17834 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17835 (simplify_subreg): Likewise.
17836 * var-tracking.c (track_loc_p): Likewise.
17837 * emit-rtl.c (byte_lowpart_offset): Likewise.
17838 (paradoxical_subreg_p): Delete out-of-line definition.
17840 2017-08-22 Jeff Law <law@redhat.com>
17842 PR tree-optimization/81741
17843 PR tree-optimization/71947
17844 * tree-ssa-dom.c: Include tree-inline.h.
17845 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
17846 equivalences if one is more expensive to compute than the other.
17847 * tree-ssa-scopedtables.h (class const_or_copies): Make
17848 record_const_or_copy_raw method private.
17849 (class avail_exprs_stack): New method simplify_binary_operation.
17850 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
17851 avail_exprs_stack::simplify_binary_operation as needed.
17852 (avail_exprs_stack::simplify_binary_operation): New function.
17854 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
17856 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
17857 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
17858 (DOT_SYMBOLS): Likewise.
17859 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
17860 (RELOCATABLE_NEEDS_FIXUP): Likewise.
17861 (RS6000_ABI_NAME): Likewise.
17862 (TARGET_CMODEL): Likewise.
17863 (TOC_SECTION_ASM_OP): Likewise.
17864 (SET_CMODEL): New macro.
17865 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
17867 2017-08-22 Richard Biener <rguenther@suse.de>
17869 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
17870 to allow for free-lang-data replacements similar to verify_type_variant.
17872 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17873 Alan Hayward <alan.hayward@arm.com>
17874 David Sherwood <david.sherwood@arm.com>
17876 * config/aarch64/aarch64.md (casesi): Use DImode rather than
17877 VOIDmode for the LABEL_REF.
17879 2017-08-22 Richard Biener <rguenther@suse.de>
17881 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
17883 2017-08-22 Richard Biener <rguenther@suse.de>
17885 * common.opt (feliminate-dwarf2-dups): Ignore.
17886 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
17887 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
17888 same_die_p_wrap, compute_section_prefix,
17889 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
17890 (comdat_symbol_id, comdat_symbol_number): Likewise.
17891 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
17893 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
17894 (output_die): Mark unreachable path unreachable.
17895 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
17896 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
17897 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
17898 (dwarf2out_early_finish): Likewise.
17900 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
17902 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
17904 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
17907 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
17908 not VAR_P. Filter attribute warnings with OPT_Wattributes.
17909 (avr_attribute_table) <io, io_low, address>: Initialize
17910 .decl_required with true.
17912 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
17914 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
17915 undocumented debugging options.
17916 (-mvsx-scalar-double): Likewise.
17917 (-mallow-df-permute): Likewise.
17918 (-mvectorize-builtins): Likewise.
17919 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17920 (rs6000_builtin_vectorized_function): Likewise.
17921 (rs6000_builtin_md_vectorized_function): Likewise.
17922 (rs6000_opt_vars): Likewise.
17924 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
17927 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
17928 (*btrq_imm): Rename from *btrq.
17929 (*btcq_imm): Rename from *btcq.
17930 (btsc): New code attribute.
17931 (*<btsc><mode>): New insn pattern.
17932 (*btr<mode>): Ditto.
17933 (*<btsc><mode>_mask): New insn_and_split pattern.
17934 (*btr<mode>_mask): Ditto.
17936 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17937 Alan Hayward <alan.hayward@arm.com>
17938 David Sherwood <david.sherwood@arm.com>
17940 * function.c (pad_below): Simplify padding calculation.
17942 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17943 Alan Hayward <alan.hayward@arm.com>
17944 David Sherwood <david.sherwood@arm.com>
17946 * target.def (function_prologue): Remove frame size argument.
17947 (function_epilogue): Likewise.
17948 * doc/tm.texi: Regenerate.
17949 * final.c (final_start_function): Update call to function_prologue.
17950 (final_end_function): Update call to function_epilogue.
17951 (default_function_pro_epilogue): Remove frame size argument.
17952 * output.h (default_function_pro_epilogue): Likewise.
17953 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
17954 (arm_output_function_prologue): Likewise.
17955 * config/frv/frv.c (frv_function_prologue): Likewise.
17956 (frv_function_epilogue): Likewise.
17957 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
17958 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
17959 (ia64_output_function_epilogue): Likewise.
17960 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
17961 (m32r_output_function_epilogue): Likewise.
17962 * config/microblaze/microblaze.c (microblaze_function_prologue)
17963 (microblaze_function_epilogue): Likewise.
17964 * config/mips/mips.c (mips_output_function_prologue): Likewise.
17965 (mips_output_function_epilogue): Likewise.
17966 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
17967 (mmix_target_asm_function_epilogue): Likewise.
17968 * config/msp430/msp430.c (msp430_start_function): Likewise.
17969 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
17970 (nds32_asm_function_epilogue): Likewise.
17971 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
17972 * config/pa/pa.c (pa_output_function_prologue): Likewise.
17973 (pa_output_function_epilogue): Likewise.
17974 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
17975 (rs6000_output_function_epilogue): Likewise.
17976 * config/rl78/rl78.c (rl78_start_function): Likewise.
17977 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
17978 (rs6000_output_function_epilogue): Likewise.
17979 * config/rx/rx.c (rx_output_function_prologue): Likewise.
17980 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
17981 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
17982 (sparc_asm_function_epilogue): Likewise.
17984 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17986 * tree.h (type_has_mode_precision_p): New function.
17987 * convert.c (convert_to_integer_1): Use it.
17988 * expr.c (expand_expr_real_2): Likewise.
17989 (expand_expr_real_1): Likewise.
17990 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
17991 * match.pd: Likewise.
17992 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
17993 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
17994 * tree-tailcall.c (process_assignment): Likewise.
17995 * tree-vect-loop.c (vectorizable_reduction): Likewise.
17996 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
17997 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
17998 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17999 (vectorizable_assignment): Likewise.
18000 (vectorizable_shift): Likewise.
18001 (vectorizable_operation): Likewise.
18002 * tree-vrp.c (register_edge_assert_for_2): Likewise.
18004 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
18006 * match.pd: Add pow (C, x) simplification.
18008 2017-08-21 Richard Biener <rguenther@suse.de>
18010 PR tree-optimization/81900
18011 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
18012 for blocks with abnormal predecessors.
18013 (compute_antic): Do not set visited flag prematurely.
18015 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
18018 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
18020 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18022 * stor-layout.h (vector_type_mode): Move to...
18023 * tree.h (vector_type_mode): ...here.
18024 * stor-layout.c (vector_type_mode): Move to...
18025 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
18027 2017-08-21 Richard Biener <rguenther@suse.de>
18029 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
18030 register_external_die hooks.
18031 (debug_false_tree_charstarstar_uhwistar): Declare.
18032 (debug_nothing_tree_charstar_uhwi): Likewise.
18033 * debug.c (do_nothing_debug_hooks): Adjust.
18034 (debug_false_tree_charstarstar_uhwistar): New do nothing.
18035 (debug_nothing_tree_charstar_uhwi): Likewise.
18036 * dbxout.c (dbx_debug_hooks): Adjust.
18037 (xcoff_debug_hooks): Likewise.
18038 * sdbout.c (sdb_debug_hooks): Likewise.
18039 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
18040 * dwarf2out.c (macinfo_label_base): New global.
18041 (dwarf2out_register_external_die): New function for the
18042 register_external_die hook.
18043 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
18044 (dwarf2_debug_hooks): Use them.
18045 (dwarf2_lineno_debug_hooks): Adjust.
18046 (struct die_struct): Add with_offset flag.
18047 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
18048 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
18049 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
18050 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
18051 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
18052 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
18053 defining section names for the early LTO debug variants.
18054 (reset_indirect_string): New helper.
18055 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
18056 (print_dw_val): Add support for offsetted symbol references.
18057 (get_ultimate_context): Split out from is_cxx.
18058 (is_cxx): Use get_ultimate_context.
18059 (is_fortran): Add decl overload.
18060 (compute_comp_unit_symbol): Split out worker from
18061 compute_section_prefix.
18062 (compute_section_prefix): Call compute_comp_unit_symbol and
18063 set comdat_type_p here.
18064 (output_die): Skip DIE symbol output for the LTO added one.
18065 Handle DIE symbol references with offset.
18066 (output_comp_unit): Guard section name mangling properly.
18067 For LTO debug sections emit a symbol at the section beginning
18068 which we use to refer to its DIEs.
18069 (add_abstract_origin_attribute): For DIEs registered via
18070 dwarf2out_register_external_die directly refer to the early
18071 DIE rather than indirectly through the shadow one we created.
18072 Remove obsolete call to dwarf2out_abstract_function for
18073 non-function/block origins.
18074 (gen_array_type_die): When generating early LTO debug do
18075 not emit DW_AT_string_length.
18076 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
18077 late when in LTO. As suggested place a gcc_unreachable for
18078 the DECL_ABSTRACT_P case.
18079 (gen_subprogram_die): Avoid another specification DIE
18080 for early built declarations/definitions for the late LTO case.
18081 (gen_variable_die): Add type references for late duplicated VLA dies
18083 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
18084 we have the abstract instance already.
18085 (process_scope_var): Adjust decl DIE contexts in LTO which
18086 first puts them in limbo.
18087 (gen_decl_die): Do not generate type DIEs late apart from
18088 types for VLAs or for decls we do not yet have a DIE. Do not
18089 call dwarf2out_abstract_function late.
18090 (dwarf2out_early_global_decl): Make sure to create DIEs
18091 for abstract instances of a decl first.
18092 (dwarf2out_late_global_decl): Adjust comment.
18093 (output_macinfo_op): With multiple macro sections use
18094 macinfo_label_base to distinguish labels.
18095 (output_macinfo): Likewise. Update macinfo_label_base.
18096 Pass in the line info label.
18097 (note_variable_value_in_expr): When generating LTO resolve
18098 all variable values here by generating DIEs as needed.
18099 (init_sections_and_labels): Add early LTO debug flag parameter
18100 and generate different sections and names if set. Add generation
18101 counter for the labels so we can have multiple of them.
18102 (reset_dies): Helper to allow DIEs to be output multiple times.
18103 (dwarf2out_finish): When outputting DIEs to the fat part of an
18104 LTO object first reset DIEs.
18105 (dwarf2out_early_finish): Output early DIEs when generating LTO.
18106 (modified_type_die): Check for decl_ultimate_origin being self
18108 (gen_type_die_with_usage): Likewise.
18109 (gen_typedef_die): Allow decl_ultimate_origin being self.
18110 (set_decl_abstract_flags): Remove.
18111 (set_block_abstract_flags): Likewise.
18112 (dwarf2out_abstract_function): Treat the early generated DIEs
18113 as the abstract copy and only add DW_AT_inline and
18114 DW_AT_artificial here and call set_decl_origin_self.
18115 If the DIE has an abstract origin don't do anything.
18116 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
18117 if we have none yet (Go fails to build one, PR78628).
18118 (variably_modified_type_p): Prevent endless recursion for Ada
18119 cyclic pointer types.
18120 * lto-streamer-in.c: Include debug.h.
18121 (dref_queue): New global.
18122 (lto_read_tree_1): Stream in DIE references.
18123 (lto_input_tree): Register DIE references.
18124 (input_function): Stream DECL_DEBUG_ARGS.
18125 * lto-streamer-out.c: Include debug.h.
18126 (lto_write_tree_1): Output DIE references.
18127 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
18128 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
18129 (output_function): Stream DECL_DEBUG_ARGS.
18130 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
18131 Stream DECL_ABSTRACT_ORIGIN.
18132 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
18133 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
18134 DECL_CONTEXT for file-scope decls.
18135 * lto-streamer.h (struct dref_entry): Declare.
18136 (dref_queue): Likewise.
18137 * cfgexpand.c (pass_expand::execute): Do not call the
18138 outlining_inline_function hook here.
18139 * lto-wrapper.c (debug_obj): New global.
18140 (tool_cleanup): Unlink it if required.
18141 (debug_objcopy): New function.
18142 (run_gcc): Handle early debug sections in the IL files by
18143 extracting them to separate files, partially linkin them and
18144 feeding the result back as result to the linker.
18145 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
18146 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
18147 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
18148 sections into a separate segment.
18149 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
18151 (darwin_asm_dwarf_section): Likewise.
18152 (darwin_asm_output_dwarf_offset): Likewise.
18153 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
18155 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18156 Alan Hayward <alan.hayward@arm.com>
18157 David Sherwood <david.sherwood@arm.com>
18159 * read-md.h (md_reader::record_potential_iterator_use): Replace
18160 pointer argument with an rtx and an index.
18161 * read-rtl.c (iterator_group::apply_iterator): Likewise.
18162 (apply_mode_iterator): Likewise.
18163 (apply_code_iterator): Likewise.
18164 (apply_int_iterator): Likewise.
18165 (apply_subst_iterator): Likewise.
18166 (record_iterator_use): Likewise.
18167 (record_attribute_use): Likewise.
18168 (md_reader::record_potential_iterator_use): Likewise. Update calls
18169 to record_iterator_use and apply_iterator.
18170 (iterator_use): Replace ptr with x and index.
18171 (attribute_use): Likewise.
18172 (apply_attribute_uses): Update calls to apply_iterator.
18173 (apply_iterators): Likewise. Update initialization of iterator_use.
18174 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
18175 and record_potential_iterator_use.
18176 (rtx_reader::read_rtx_operand): Likewise.
18178 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18179 Alan Hayward <alan.hayward@arm.com>
18180 David Sherwood <david.sherwood@arm.com>
18182 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
18185 2017-08-21 Richard Biener <rguenther@suse.de>
18187 PR middle-end/81884
18188 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
18189 at struct end conservatively when comparing common bases.
18191 2017-08-21 Richard Biener <rguenther@suse.de>
18193 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
18194 (mem_ref_in_stmt): Remove.
18195 (determine_max_movement): Use ref index to get at the reference.
18196 (invariantness_dom_walker::before_dom_children): Deal with
18197 lim data already initialized.
18198 (gather_mem_refs_stmt): Initialize lim data and record ref index.
18200 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
18202 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
18203 (TARGET_ISA_ROUND): Ditto.
18204 (TARGET_ROUND): Ditto.
18205 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
18206 * config/i386/i386.md: Ditto.
18207 * config/i386/sse.md: Ditto.
18208 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
18209 with OPTION_MASK_ISA_SSE4_1.
18211 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
18214 * doc/extend.texi (x86 Built-in Functions): Correct the name of
18215 __builtin_ia32_lzcnt_u16.
18217 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
18220 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
18221 (rs6000_set_current_function): Rewrite function to use it.
18223 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
18226 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
18227 and TYPE_WARN_IF_NOT_ALIGN.
18228 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
18229 (handle_warn_if_not_align): New.
18230 (place_union_field): Call handle_warn_if_not_align.
18231 (place_field): Call handle_warn_if_not_align.
18232 Copy TYPE_WARN_IF_NOT_ALIGN.
18233 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
18234 (layout_type): Likewise.
18235 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
18237 (tree_decl_common): Add warn_if_not_align.
18238 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
18239 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
18240 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
18241 (DECL_WARN_IF_NOT_ALIGN): Likewise.
18242 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
18243 * doc/extend.texi: Document warn_if_not_aligned attribute.
18244 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
18246 2017-08-17 Martin Liska <mliska@suse.cz>
18249 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
18250 (get_data_dependence): Use it as pointer type.
18251 (distribute_loop): Likewise.
18253 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18255 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
18256 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
18257 (p8_vmrgow_<mode>_direct): New define_insn.
18258 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
18259 handle endianness for vmrgew and vmrgow permute patterns.
18261 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
18263 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
18264 * config/rs6000/rs6000-cpus.def: Remove comment.
18265 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
18266 (POWERPC_MASKS): Likewise.
18267 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
18268 use of TARGET_VSX_TIMODE.
18269 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
18270 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
18271 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
18272 (rs6000_option_override_internal): Remove dead code.
18273 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
18274 (rs6000_legitimize_reload_address): Likewise.
18275 (rs6000_legitimate_address_p): Likewise.
18276 (rs6000_opt_masks): Delete "vsx-timode".
18277 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
18278 from function comment.
18279 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
18280 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
18281 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
18283 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
18284 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
18285 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
18286 (VSX_TI): Likewise.
18288 (define_peephole2): Likewise.
18290 2017-08-17 Martin Sebor <msebor@redhat.com>
18293 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
18294 past the end of an array.
18295 (test_pp_format): Add test cases.
18297 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
18299 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
18300 missing ECF_NOTHROW flags.
18302 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
18305 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
18306 operands residing in integer registers.
18307 (*vsx_le_perm_load_<mode>): Likewise.
18308 (*vsx_le_perm_store_<mode>): Likewise.
18309 (define_peephole2): Add peepholes to optimize the above.
18311 2017-08-17 Marek Polacek <polacek@redhat.com>
18313 PR middle-end/81814
18314 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
18315 to mimic what shorten_compare did. Change the return type to bool.
18316 (fold_cond_expr_with_comparison): Update call to
18317 operand_equal_for_comparison_p.
18318 (fold_ternary_loc): Likewise.
18320 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
18322 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
18324 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
18326 2017-08-17 Richard Biener <rguenther@suse.de>
18328 * tree-ssa-structalias.c (solve_graph): When propagating
18329 to successors update the graphs succ edges and avoid duplicate work.
18331 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
18334 * config/i386/i386.c (ix86_option_override_internal): Save target
18335 specific options after ix86_stack_protector_guard_reg was changed.
18337 2017-08-17 Richard Biener <rguenther@suse.de>
18339 PR tree-optimization/81827
18340 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
18341 (new_var_info): Initialize it conservatively.
18342 (get_call_vi): Mark register vars.
18343 (new_scalar_tmp_constraint_exp): Likewise.
18344 (handle_rhs_call): Likewise.
18345 (handle_const_call): Likewise.
18346 (create_function_info_for): Likewise.
18347 (solve_constraints): Sort varinfos to separate register from
18348 non-register vars to pack points-to solution bitmaps during
18351 2017-08-17 Marek Polacek <polacek@redhat.com>
18353 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
18355 2017-08-17 Richard Biener <rguenther@suse.de>
18357 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
18358 to true when overflow is undefined and we saturated the result.
18360 2017-08-17 Alan Modra <amodra@gmail.com>
18363 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
18364 Don't use store multiple if only one reg needs saving.
18365 (interesting_frame_related_regno): New function.
18366 (rs6000_frame_related): Don't emit frame info for regs that
18368 (rs6000_emit_epilogue): Likewise.
18370 2017-08-16 Nathan Sidwell <nathan@acm.org>
18372 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
18373 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
18374 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
18375 * tree.c (free_lang_data_in_type): Use else-if chain. Always
18376 clear TYPE_LANG_1. Remove obsolete member-function stripping.
18377 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
18378 (verify_type): Adjust for TYPE_BINFO move.
18379 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
18380 process TYPE_BINFO directly.
18381 (hash_tree): Likewise.
18382 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
18384 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
18387 2017-08-16 David Malcolm <dmalcolm@redhat.com>
18389 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
18391 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
18394 * config/i386/i386.md (*anddi_1_btr): Change predicates of
18395 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
18396 Add ix86_binary_operator_ok to insn constraint.
18397 (*iordi_1_bts): Ditto.
18398 (*xordi_1_btc): Ditto.
18399 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
18400 Update corresponding peephole2 pattern.
18404 2017-08-16 Bin Cheng <bin.cheng@arm.com>
18406 PR tree-optimization/81832
18407 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
18408 copy loop header which has IFN_LOOP_DIST_ALIAS call.
18410 2017-08-16 Marek Polacek <polacek@redhat.com>
18413 * fold-const.c (fold_indirect_ref_1): Restore original behavior
18414 regarding size_zero_node.
18416 2017-08-16 Martin Liska <mliska@suse.cz>
18419 * config.gcc: Respect previously set extra_objs in case
18422 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
18424 PR tree-optimization/81835
18425 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
18426 the phi SSA_NAME. Check that the condition in a COND_EXPR does
18427 not depend on the phi.
18429 2017-08-16 Alan Modra <amodra@gmail.com>
18431 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
18434 2017-08-16 Alan Modra <amodra@gmail.com>
18436 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
18437 (save_reg_p): ..into this. Update all callers.
18438 (first_reg_to_save): Simplify.
18440 2017-08-16 Alan Modra <amodra@gmail.com>
18442 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
18445 2017-08-15 Joseph Myers <joseph@codesourcery.com>
18449 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
18450 constant count if that count is less than 32.
18452 2017-08-15 Nathan Sidwell <nathan@acm.org>
18454 * gcc.c (execute): Emit friendlier message if inferior is killed
18455 by an external cause.
18457 2017-08-15 Richard Biener <rguenther@suse.de>
18459 PR tree-optimization/81790
18460 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
18461 CONSTRUCTORs from simplifying and VN.
18463 2017-08-14 Martin Sebor <msebor@redhat.com>
18465 * builtin-attrs.def: Add comments.
18467 2017-08-14 Martin Sebor <msebor@redhat.com>
18470 * doc/extend.texi (attribute nonstring): Document new attribute.
18472 2017-08-14 Martin Sebor <msebor@redhat.com>
18475 * tree-diagnostic.c (default_tree_printer): Handle %G.
18476 * gimple-pretty-print.h (percent_G_format): Declare new function.
18477 * gimple-pretty-print.c (percent_G_format): Define.
18478 * tree-pretty-print.c (percent_K_format): Add argument.
18480 2017-08-14 Martin Sebor <msebor@redhat.com>
18482 PR translation/79998
18483 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
18484 Remove a stray space.
18486 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
18489 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
18490 (*iordi_1_bts): Ditto.
18491 (*xordi_1_btc): Ditto.
18493 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18496 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
18497 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18499 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
18500 quoted strings, and make more translator-friendly.
18501 (darwin_rs6000_override_options): Likewise.
18502 (rs6000_option_override_internal): Likewise.
18503 (rs6000_return_in_memory): Fix overlong line.
18504 (init_cmulative_args): Use quoted strings, and make more
18505 translator-friendly.
18506 (rs6000_pass_by_reference): Fix overlong line.
18507 (def_builtin): Use quoted strings.
18508 (altivec_expand_predicate_builtin): Use quoted strings, and make
18509 more translator-friendly.
18510 (htm_expand_builtin): Use quoted strings.
18511 (cpu_expand_builtin): Use quoted strings, and make more
18512 translator-friendly.
18513 (altivec_expand_builtin): Likewise.
18514 (paired_expand_predicate_builtin): Likewise.
18515 (rs6000_invalid_builtin): Likewise.
18516 (builtin_function_type): Use quoted strings.
18517 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
18518 more translator-friendly.
18519 (rs6000_trampoline_init): Likewise.
18520 (rs6000_handle_altivec_attribute): Likewise.
18521 (rs6000_inner_target_options): Use quoted strings.
18522 (rs6000_disable_incompatible_switches): Likewise.
18523 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
18524 strings, and make more translator-friendly.
18525 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
18527 2017-08-14 Bin Cheng <bin.cheng@arm.com>
18529 PR tree-optimization/81799
18530 * tree-loop-distribution.c (version_loop_by_alias_check): Force
18531 cond_expr to simple gimple operand.
18533 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
18535 PR middle-end/46932
18536 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
18538 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
18542 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
18543 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
18544 TARGET_GASISR_PROLOGUES.
18545 * config/avr/avr.c (avr_option_override): Same.
18546 (avr_pass_pre_proep::execute): Same.
18548 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
18551 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
18552 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
18554 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
18556 * config/i386/i386.md (*load_tp_<mode>): Redefine as
18557 define_insn_and_split. Split to a memory load from 0 in
18558 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
18559 using PTR mode iterator.
18560 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
18561 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
18562 (*add_tp_<mode>): Redefine as define_insn_and_split.
18563 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
18564 address space. Merge with *add_tp_x32 using PTR mode iterator.
18565 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
18566 Split to an add with a memory load from 0 in
18567 DEFAULT_TLS_SEG_REG address space.
18569 2017-08-12 Andrew Pinski <apinski@cavium.com>
18571 * config/aarch64/aarch64-option-extensions.def (rdma):
18572 Fix feature string to what Linux prints out in /proc/cpuinfo.
18574 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
18577 * dwarf2out.c (modified_type_die): For C typedef types that have
18578 an ultimate origin, process the ultimate origin instead of the
18580 (gen_typedef_die): Assert that input DECLs have no ultimate
18582 (gen_type_die_with_usage): For typedef variants that have an
18583 ultimate origin, just call gen_decl_die on the original DECL.
18584 (process_scope_var): Avoid creating DIEs for local typedefs and
18585 concrete static variables.
18587 2017-08-12 Alan Modra <amodra@gmail.com>
18591 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
18592 match gnu-user.h startfile.
18593 (ENDFILE_LINUX_SPEC): Similarly.
18595 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
18598 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
18600 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
18602 2017-08-11 Tamar Christina <tamar.christina@arm.com>
18603 * config/aarch64/aarch64.md (mov<mode>): Change.
18604 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
18605 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
18606 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
18608 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
18610 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
18611 for storage order barriers.
18613 2017-08-11 Martin Liska <mliska@suse.cz>
18616 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
18617 variables of void type.
18619 2017-08-11 Martin Liska <mliska@suse.cz>
18621 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
18622 TARGET_SUPPORTS_ALIASES.
18623 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
18624 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
18625 (optimize_weakref): Likewise.
18626 * symtab.c (symtab_node::noninterposable_alias): Likewise.
18627 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
18628 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
18630 2017-08-11 Martin Liska <mliska@suse.cz>
18633 * config/i386/i386.c (make_resolver_func): Do complete
18634 refactoring of the function.
18636 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
18639 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
18640 * config/i386/i386.c (ix86_stack_protect_guard): Use
18641 ix86_stack_protect_guard_symbol_str to generate varible declaration.
18642 * doc/invoke.texi (x86 Options): Document
18643 -mstack-protector-guard-symbol= option.
18645 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
18647 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
18648 * config/i386/i386.c (ix86_split_stack_guard): New function.
18649 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
18650 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
18651 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
18652 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
18653 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
18654 (split_stack_space_check): Call ix86_split_stack_guard.
18656 2017-08-10 Martin Sebor <msebor@redhat.com>
18658 * print-tree.c (print_node): Print location using the established
18660 Replace spaces with colons.
18661 (debug_raw, debug): Ditto.
18663 2017-08-10 Martin Sebor <msebor@redhat.com>
18666 * pretty-print.c (pp_format): Correct the handling of %s precision.
18668 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
18671 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18673 (ix86_finalize_stack_frame_flags): This. Also clear
18674 frame_pointer_needed if -fno-omit-frame-pointer is used without
18676 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18677 with ix86_finalize_stack_frame_flags.
18678 (ix86_expand_epilogue): Likewise.
18679 (ix86_expand_split_stack_prologue): Likewise.
18680 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
18682 2017-08-10 Martin Liska <mliska@suse.cz>
18685 * c-attribs.c (handle_target_attribute):
18686 Report warning for an empty string argument of target attribute.
18688 2017-08-09 Jakub Jelinek <jakub@redhat.com>
18691 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
18693 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
18694 or DECL_NONLOCAL labels.
18695 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
18696 or DECL_NONLOCAL labels here.
18698 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
18700 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
18701 to indicate when early gimple folding has been disabled.
18702 (rs6000_gimple_fold_builtin): Add debug content.
18703 (rs6000_invalid_builtin): Fix whitespace.
18704 (rs6000_expand_builtin): Fix whitespace.
18705 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
18707 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
18710 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
18711 SAVE_MULTIPLE if not all the registers that saves, should be saved.
18713 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
18715 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
18716 (qdf24xx): Likewise.
18717 * config/aarch64/aarch64.md: Include falkor.md.
18718 * config/aarch64/falkor.md: New.
18720 2017-08-09 Marek Polacek <polacek@redhat.com>
18723 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
18724 * diagnostic.c (emit_diagnostic): Add a comment.
18725 (emit_diagnostic_valist): New function.
18727 2017-08-09 Marek Polacek <polacek@redhat.com>
18730 * input.c (make_location): New overload.
18731 * input.h (make_location): Declare.
18733 2017-08-08 Alan Modra <amodra@gmail.com>
18734 H.J. Lu <hongjiu.lu@intel.com>
18737 * gcc.c (NO_PIE_SPEC): Delete.
18738 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
18740 (LINK_PIE_SPEC): ..to here.
18741 (LINK_COMMAND_SPEC): Support -no-pie.
18742 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
18743 chain of crtbegin*.o selection, update for PIE_SPEC changes and
18745 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
18746 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
18747 (ENDFILE_CRTEND_SPEC): Similarly.
18749 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
18752 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
18753 (mstack-protector-guard-offset=): Ditto.
18754 * config/i386/i386.c (ix86_option_override): Handle
18755 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
18757 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
18758 ix86_stack_protect_guard_offset variables.
18759 (TARGET_STACK_PROTECT_GUARD): Always define.
18760 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
18761 and -mstack-protector-guard-offset= options.
18763 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18765 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
18766 boundary case for the last candidate.
18768 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18770 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
18771 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
18773 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18775 PR middle-end/19706
18776 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
18777 * config/aarch64/aarch64-builtins.c
18778 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
18779 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
18780 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
18782 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18783 Andrew Pinski <pinskia@gmail.com>
18785 PR middle-end/19706
18786 * internal-fn.def (XORSIGN): New.
18787 * optabs.def (xorsign_optab): New.
18788 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
18789 (convert_expand_mult_copysign): New.
18790 (pass_optimize_widening_mul::execute): Call
18791 convert_expand_mult_copysign.
18793 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18795 PR tree-optimization/81354
18796 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
18797 Insert on edges rather than explicitly creating landing pads.
18798 (analyze_candidates_and_replace): Commit edge inserts.
18800 2017-08-08 Richard Biener <rguenther@suse.de>
18802 PR middle-end/81719
18803 * tree-ssa-loop-niter.c: Include tree-dfa.h.
18804 (expand_simple_operations): Also look through ADDR_EXPRs with
18805 MEM_REF bases treating them as POINTER_PLUS_EXPR.
18807 2017-08-08 Richard Biener <rguenther@suse.de>
18809 PR tree-optimization/81723
18810 * tree-vect-slp.c (struct bst_traits): New hash traits.
18811 (bst_fail): New global.
18812 (vect_build_slp_tree_2): New worker, split out from ...
18813 (vect_build_slp_tree): ... this now wrapping it with using
18814 bst_fail set to cache SLP tree build fails. Properly handle
18816 (vect_analyze_slp_instance): Allocate and free bst_fail.
18818 2017-08-08 Martin Liska <mliska@suse.cz>
18821 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
18822 LABEL_DECLs that can be from a different function.
18824 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18826 PR tree-optimization/81744
18827 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
18828 loop's number of iterations.
18830 2017-08-08 Martin Liska <mliska@suse.cz>
18832 * asan.c: Include header files.
18833 * attribs.c (build_decl_attribute_variant): New function moved
18835 (build_type_attribute_qual_variant): Likewise.
18836 (cmp_attrib_identifiers): Likewise.
18837 (simple_cst_list_equal): Likewise.
18838 (omp_declare_simd_clauses_equal): Likewise.
18839 (attribute_value_equal): Likewise.
18840 (comp_type_attributes): Likewise.
18841 (build_type_attribute_variant): Likewise.
18842 (lookup_ident_attribute): Likewise.
18843 (remove_attribute): Likewise.
18844 (merge_attributes): Likewise.
18845 (merge_type_attributes): Likewise.
18846 (merge_decl_attributes): Likewise.
18847 (merge_dllimport_decl_attributes): Likewise.
18848 (handle_dll_attribute): Likewise.
18849 (attribute_list_equal): Likewise.
18850 (attribute_list_contained): Likewise.
18851 * attribs.h (lookup_attribute): New function moved from tree.[ch].
18852 (lookup_attribute_by_prefix): Likewise.
18853 * bb-reorder.c: Include header files.
18854 * builtins.c: Likewise.
18855 * calls.c: Likewise.
18856 * cfgexpand.c: Likewise.
18857 * cgraph.c: Likewise.
18858 * cgraphunit.c: Likewise.
18859 * convert.c: Likewise.
18860 * dwarf2out.c: Likewise.
18861 * final.c: Likewise.
18862 * fold-const.c: Likewise.
18863 * function.c: Likewise.
18864 * gimple-expr.c: Likewise.
18865 * gimple-fold.c: Likewise.
18866 * gimple-pretty-print.c: Likewise.
18867 * gimple.c: Likewise.
18868 * gimplify.c: Likewise.
18869 * hsa-common.c: Likewise.
18870 * hsa-gen.c: Likewise.
18871 * internal-fn.c: Likewise.
18872 * ipa-chkp.c: Likewise.
18873 * ipa-cp.c: Likewise.
18874 * ipa-devirt.c: Likewise.
18875 * ipa-fnsummary.c: Likewise.
18876 * ipa-inline.c: Likewise.
18877 * ipa-visibility.c: Likewise.
18879 * lto-cgraph.c: Likewise.
18880 * omp-expand.c: Likewise.
18881 * omp-general.c: Likewise.
18882 * omp-low.c: Likewise.
18883 * omp-offload.c: Likewise.
18884 * omp-simd-clone.c: Likewise.
18885 * opts-global.c: Likewise.
18886 * passes.c: Likewise.
18887 * predict.c: Likewise.
18888 * sancov.c: Likewise.
18889 * sanopt.c: Likewise.
18890 * symtab.c: Likewise.
18891 * toplev.c: Likewise.
18892 * trans-mem.c: Likewise.
18893 * tree-chkp.c: Likewise.
18894 * tree-eh.c: Likewise.
18895 * tree-into-ssa.c: Likewise.
18896 * tree-object-size.c: Likewise.
18897 * tree-parloops.c: Likewise.
18898 * tree-profile.c: Likewise.
18899 * tree-ssa-ccp.c: Likewise.
18900 * tree-ssa-live.c: Likewise.
18901 * tree-ssa-loop.c: Likewise.
18902 * tree-ssa-sccvn.c: Likewise.
18903 * tree-ssa-structalias.c: Likewise.
18904 * tree-ssa.c: Likewise.
18905 * tree-streamer-in.c: Likewise.
18906 * tree-vectorizer.c: Likewise.
18907 * tree-vrp.c: Likewise.
18908 * tsan.c: Likewise.
18909 * ubsan.c: Likewise.
18910 * varasm.c: Likewise.
18911 * varpool.c: Likewise.
18912 * tree.c: Remove functions moved to attribs.[ch].
18913 * tree.h: Likewise.
18914 * config/aarch64/aarch64.c: Add attrs.h header file.
18915 * config/alpha/alpha.c: Likewise.
18916 * config/arc/arc.c: Likewise.
18917 * config/arm/arm.c: Likewise.
18918 * config/avr/avr.c: Likewise.
18919 * config/bfin/bfin.c: Likewise.
18920 * config/c6x/c6x.c: Likewise.
18921 * config/cr16/cr16.c: Likewise.
18922 * config/cris/cris.c: Likewise.
18923 * config/darwin.c: Likewise.
18924 * config/epiphany/epiphany.c: Likewise.
18925 * config/fr30/fr30.c: Likewise.
18926 * config/frv/frv.c: Likewise.
18927 * config/ft32/ft32.c: Likewise.
18928 * config/h8300/h8300.c: Likewise.
18929 * config/i386/winnt.c: Likewise.
18930 * config/ia64/ia64.c: Likewise.
18931 * config/iq2000/iq2000.c: Likewise.
18932 * config/lm32/lm32.c: Likewise.
18933 * config/m32c/m32c.c: Likewise.
18934 * config/m32r/m32r.c: Likewise.
18935 * config/m68k/m68k.c: Likewise.
18936 * config/mcore/mcore.c: Likewise.
18937 * config/microblaze/microblaze.c: Likewise.
18938 * config/mips/mips.c: Likewise.
18939 * config/mmix/mmix.c: Likewise.
18940 * config/mn10300/mn10300.c: Likewise.
18941 * config/moxie/moxie.c: Likewise.
18942 * config/msp430/msp430.c: Likewise.
18943 * config/nds32/nds32-isr.c: Likewise.
18944 * config/nds32/nds32.c: Likewise.
18945 * config/nios2/nios2.c: Likewise.
18946 * config/nvptx/nvptx.c: Likewise.
18947 * config/pa/pa.c: Likewise.
18948 * config/pdp11/pdp11.c: Likewise.
18949 * config/powerpcspe/powerpcspe.c: Likewise.
18950 * config/riscv/riscv.c: Likewise.
18951 * config/rl78/rl78.c: Likewise.
18952 * config/rx/rx.c: Likewise.
18953 * config/s390/s390.c: Likewise.
18954 * config/sh/sh.c: Likewise.
18955 * config/sol2.c: Likewise.
18956 * config/sparc/sparc.c: Likewise.
18957 * config/spu/spu.c: Likewise.
18958 * config/stormy16/stormy16.c: Likewise.
18959 * config/tilegx/tilegx.c: Likewise.
18960 * config/tilepro/tilepro.c: Likewise.
18961 * config/v850/v850.c: Likewise.
18962 * config/vax/vax.c: Likewise.
18963 * config/visium/visium.c: Likewise.
18964 * config/xtensa/xtensa.c: Likewise.
18966 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18969 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
18970 constraints since the -mupper-regs-* switches have been
18972 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
18973 into a vector from a double word element that was extracted from
18974 another vector, and eliminate extra XXPERMDI instructions.
18975 (vsx_concat_<mode>_2): Likewise.
18976 (vsx_concat_<mode>_3): Likewise.
18977 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
18978 concat to allow optimizing inserts from previous extracts.
18980 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
18982 * config/i386/i386.c (ix86_stack_protect_guard): Generate
18983 memory reference to a SSP offset in TLS address space.
18984 (ix86_print_operand) <case '@'>: Remove.
18985 (ix86_print_operand_punct_valid_p): Remove '@' code.
18986 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
18987 UNSPEC_SP_TLS_TEST.
18988 (stack_tls_protect_set_<mode>): Remove.
18989 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
18990 (stack_tls_protect_test_<mode>): Remove.
18991 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
18993 2017-08-07 Olivier Hainque <hainque@adacore.com>
18996 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
18998 2017-08-07 Douglas Rupp <rupp@adacore.com>
19000 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
19001 variable was referenced as multidir in command.
19003 2017-08-07 Jakub Jelinek <jakub@redhat.com>
19006 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
19009 2017-08-07 Martin Liska <mliska@suse.cz>
19011 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
19012 * config/rl78/rl78.c: Add include of attribs.h.
19013 * config/sh/sh.c: Likewise.
19014 * config/v850/v850.c: Likewise.
19016 2017-08-07 Tom de Vries <tom@codesourcery.com>
19018 PR middle-end/78266
19019 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
19021 2017-08-07 Martin Liska <mliska@suse.cz>
19023 * config/mips/mips.c: Include attribs.h.
19025 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
19028 * doc/invoke.texi: Document change in behvaior for -Ofast for
19031 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
19033 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
19035 (aarch64_pop_regs): Likewise.
19036 (aarch64_gen_load_pair): Likewise.
19037 (aarch64_save_callee_saves): Likewise.
19038 (aarch64_restore_callee_saves): Likewise.
19040 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
19042 * config/i386/i386.c: Revert the last change.
19044 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
19047 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
19049 (ix86_finalize_stack_frame_flags): This. Also clear
19050 frame_pointer_needed if -fno-omit-frame-pointer is used without
19052 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
19053 with ix86_finalize_stack_frame_flags.
19054 (ix86_expand_epilogue): Likewise.
19055 (ix86_expand_split_stack_prologue): Likewise.
19057 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
19060 * config/i386/i386.c (get_builtin_code_for_version): Set priority
19061 to P_AES for Westmere.
19063 2017-08-07 Jonathan Yong <10walls@gmail.com>
19065 * config/i386/mingw.opt (fset-stack-executable): Removed.
19066 * config/i386/cygming.opt (fset-stack-executable): Moved
19068 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
19070 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
19072 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
19074 2017-08-07 Marek Polacek <polacek@redhat.com>
19076 PR middle-end/81737
19077 * fold-const.c (fold_indirect_ref_1): Check type_domain.
19079 2017-08-07 Martin Liska <mliska@suse.cz>
19081 * attribs.h (canonicalize_attr_name): New function.
19082 (cmp_attribs): Move from c-format.c and adjusted.
19083 (is_attribute_p): Moved from tree.h.
19084 * tree-inline.c: Add new includes.
19085 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
19086 (private_is_attribute_p): Remove.
19087 (private_lookup_attribute): Likewise.
19088 (private_lookup_attribute_by_prefix): Simplify.
19089 (remove_attribute): Use is_attribute_p.
19090 * tree.h: Remove removed declarations.
19092 2017-08-07 Jakub Jelinek <jakub@redhat.com>
19094 PR middle-end/81698
19095 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
19096 instead of computing it in the function. Formatting fix.
19097 (expand_case): Don't rely on default_edge being the first edge,
19098 clear it if removing it, pass default_edge to
19099 emit_case_dispatch_table.
19100 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
19103 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
19105 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
19106 insn in the function, emit NOP after the insn.
19108 2017-08-06 Tom de Vries <tom@codesourcery.com>
19110 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
19113 2017-08-06 Tom de Vries <tom@codesourcery.com>
19115 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
19118 2017-08-04 Yury Gribov <tetra2005@gmail.com>
19120 PR tree-optimization/57371
19121 * match.pd: New pattern.
19123 2017-08-04 Marek Polacek <polacek@redhat.com>
19125 PR middle-end/81695
19126 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
19127 perform the computation in offset_int.
19129 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19131 PR tree-optimization/81136
19132 * tree-vectorizer.h: Include tree-hash-traits.h.
19133 (vec_base_alignments): New typedef.
19134 (vec_info): Add a base_alignments field.
19135 (vect_record_base_alignments): Declare.
19136 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
19138 (DR_IS_CONDITIONAL_IN_STMT): New macro.
19139 (create_data_ref): Add an is_conditional_in_stmt argument.
19140 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
19141 the is_conditional_in_stmt field.
19142 (data_ref_loc): Add an is_conditional_in_stmt field.
19143 (get_references_in_stmt): Set the is_conditional_in_stmt field.
19144 (find_data_references_in_stmt): Update call to create_data_ref.
19145 (graphite_find_data_references_in_stmt): Likewise.
19146 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
19147 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
19148 (vect_record_base_alignment): New function.
19149 (vect_record_base_alignments): Likewise.
19150 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
19151 for nested statements even if we fail to compute a misalignment.
19152 Use pooled base alignments for unconditional references.
19153 (vect_find_same_alignment_drs): Compare base addresses instead
19155 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
19156 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
19158 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19160 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
19161 Add an explicit name for the enum. Use auto_vec for slp_instances
19162 and grouped_stores.
19163 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
19165 (_bb_vec_info): Add a constructor and destructor.
19166 (vinfo_for_stmt): Return NULL for uids of -1 as well.
19167 (destroy_loop_vec_info): Delete.
19168 (vect_destroy_datarefs): Likewise.
19169 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
19170 (vec_info::vec_info): New function.
19171 (vec_info::~vec_info): Likewise.
19172 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
19173 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
19174 destroy_loop_vec_info.
19175 * tree-vect-loop.c (new_loop_vec_info): Replace with...
19176 (_loop_vec_info::_loop_vec_info): ...this.
19177 (destroy_loop_vec_info): Replace with...
19178 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
19179 the stmt_vec_infos. Leave handling of vec_info information to its
19180 destructor. Remove explicit vector releases.
19181 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
19182 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
19183 * tree-vect-slp.c (new_bb_vec_info): Replace with...
19184 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
19185 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
19186 (destroy_bb_vec_info): Replace with...
19187 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
19188 information to its destructor.
19189 (vect_slp_analyze_bb_1): Use new and delete instead of
19190 new_bb_vec_info and destroy_bb_vec_info.
19191 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
19194 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19196 * tree-data-ref.h (subscript): Add access_fn field.
19197 (data_dependence_relation): Add could_be_independent_p.
19198 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
19199 (same_access_functions): Move to tree-data-ref.c.
19200 * tree-data-ref.c (ref_contains_union_access_p): New function.
19201 (access_fn_component_p): Likewise.
19202 (access_fn_components_comparable_p): Likewise.
19203 (dr_analyze_indices): Add a reference to access_fn_component_p.
19204 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
19206 (constant_access_functions): Likewise.
19207 (add_other_self_distances): Likewise.
19208 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
19209 (initialize_data_dependence_relation): Use XCNEW and remove
19210 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
19211 of access functions that have the same type. Allow the
19212 subsequence to end with different bases in some circumstances.
19213 Record the chosen access functions in SUB_ACCESS_FN.
19214 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
19215 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
19216 (subscript_dependence_tester_1): Likewise dra and drb.
19217 (build_classic_dist_vector): Update calls accordingly.
19218 (subscript_dependence_tester): Likewise.
19219 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
19220 DDR_COULD_BE_INDEPENDENT_P.
19221 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
19222 comp_alias_ddrs instead of may_alias_ddrs.
19223 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
19225 (vect_analyze_data_ref_dependence): Use it if
19226 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
19227 distance vectors if that fails.
19228 (dependence_distance_ge_vf): New function.
19229 (vect_prune_runtime_alias_test_list): Use it. Don't clear
19230 LOOP_VINFO_MAY_ALIAS_DDRS.
19232 2017-08-04 Richard Biener <rguenther@suse.de>
19234 PR middle-end/81705
19235 * fold-const.c (fold_binary_loc): Properly restrict
19236 minus_var0 && minus_var1 case when associating undefined overflow
19239 2017-08-04 Tom de Vries <tom@codesourcery.com>
19241 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
19243 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19245 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19246 Don't start diagnostic messages with a capital letter.
19247 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19249 (rs6000_invalid_builtin): Likewise.
19250 (rs6000_trampoline_init): Likewise.
19252 2017-08-03 Jakub Jelinek <jakub@redhat.com>
19255 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
19256 after setting changeable df flags.
19258 2017-08-03 Richard Biener <rguenther@suse.de>
19260 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
19261 up if the use is in USE - X.
19263 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
19265 * toplev.c (dumpfile.h): New include.
19266 (internal_error_reentered): New static function. Use it...
19267 (internal_error_function): ...here to handle reentered internal_error.
19269 2017-08-03 Richard Biener <rguenther@suse.de>
19271 PR middle-end/81148
19272 * fold-const.c (split_tree): Add minus_var and minus_con
19273 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
19274 here but always use minus_*.
19275 (associate_trees): Assert we never associate with MINUS_EXPR
19276 and NULL first operand. Do not recurse for PLUS_EXPR operands
19277 when associating as MINUS_EXPR either.
19278 (fold_binary_loc): Track minus_var and minus_con.
19280 2017-08-03 Tom de Vries <tom@codesourcery.com>
19283 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
19284 ACCEL_COMPILER, apply finish_options on
19285 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
19287 2017-08-03 Tom de Vries <tom@codesourcery.com>
19290 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
19291 function_entry_patch_area_size > 0.
19293 2017-08-03 Jakub Jelinek <jakub@redhat.com>
19296 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
19297 instead of 10??LU, perform unit multiplication in wide_int,
19298 don't change alloc_object_size_limit if the limit is larger
19301 PR tree-optimization/81655
19302 PR tree-optimization/81588
19303 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
19304 the case when ranges[i].low and high are 1 for unsigned type with
19307 PR middle-end/81052
19308 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
19309 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
19311 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19313 * tree-vrp.h: Add include guard.
19315 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
19318 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
19319 (ud2): New insn pattern.
19320 * config/i386/i386.c (ix86_expand_epilogue):
19321 For naked functions, generate ud2 instead of trap insn.
19323 2017-08-02 Marek Polacek <polacek@redhat.com>
19326 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
19328 2017-08-02 Tom de Vries <tom@codesourcery.com>
19329 Cesar Philippidis <cesar@codesourcery.com>
19331 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
19332 Add missing edge probabilities.
19334 2017-08-02 Tamar Christina <tamar.christina@arm.com>
19336 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
19337 Correct endianness.
19339 2017-08-02 Jakub Jelinek <jakub@redhat.com>
19341 PR middle-end/79499
19342 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
19343 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
19344 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
19345 prologue_seq sequences - if any.
19347 2017-08-02 Richard Biener <rguenther@suse.de>
19349 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
19350 via vectors if supported, integer extracts via punning if supported
19351 or otherwise vector extracts.
19353 2017-08-02 Richard Biener <rguenther@suse.de>
19355 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
19357 (bitmap_insert_into_set): ... this.
19359 2017-08-02 Richard Biener <rguenther@suse.de>
19361 PR tree-optimization/81633
19363 2015-08-17 Alan Hayward <alan.hayward@arm.com>
19365 PR tree-optimization/71752
19366 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
19368 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
19370 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
19371 (machine_function::call_ms2sysv_pad_out): Remove field.
19372 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
19373 (ix86_compute_frame_layout): Likewise.
19375 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
19378 * config/i386/i386.c (ix86_set_func_type): Disallow naked
19379 attribute with interrupt attribute.
19381 2017-08-01 Andrew Pinski <apinski@cavium.com>
19383 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
19384 BIT_INSERT_EXPR's operand 1
19385 to see if the types precision matches.
19387 2017-08-01 Martin Liska <mliska@suse.cz>
19389 PR middle-end/70140
19390 * builtins.c (expand_builtin_memcpy_args): Remove.
19391 (expand_builtin_memcpy): Call newly added function
19392 expand_builtin_memory_copy_args.
19393 (expand_builtin_memcpy_with_bounds): Likewise.
19394 (expand_builtin_mempcpy): Remove last argument.
19395 (expand_builtin_mempcpy_with_bounds): Likewise.
19396 (expand_builtin_memory_copy_args): New function created from
19397 expand_builtin_mempcpy_args with small modifications.
19398 (expand_builtin_mempcpy_args): Remove.
19399 (expand_builtin_stpcpy): Remove unused argument.
19400 (expand_builtin): Likewise.
19401 (expand_builtin_with_bounds): Likewise.
19403 2017-08-01 Martin Liska <mliska@suse.cz>
19406 Make mempcpy more optimal (PR middle-end/70140).
19408 2017-08-01 Jakub Jelinek <jakub@redhat.com>
19411 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
19412 __builtin_vec_cmpne verify both arguments are compatible vectors
19413 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
19414 verify arg1_type is a pointer or array type. For __builtin_vec_st,
19415 move computation of aligned to after checking the argument types.
19419 * config/rs6000/vsx.md (vextract_fp_from_shorth,
19420 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
19423 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
19424 Doug Rupp <rupp@adacore.com>
19425 Olivier Hainque <hainque@adacore.com>
19427 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
19428 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
19430 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
19431 for TARGET_OS_CPP_BUILTIN.
19432 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
19433 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
19435 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
19436 passing required abi options to the assembler for EABI configurations.
19437 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
19438 of .text.hot and .text.unlikely sections for kernel modules when
19439 using ARM style exceptions.
19440 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
19441 options. Add EXTRA_CC1_SPEC.
19442 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
19444 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
19446 (ARM_TARGET2_DWARF_FORMAT): Define.
19447 * config/arm/t-vxworks: Adjust multilib control to removal of the
19448 Diab command line options.
19450 2017-08-01 Martin Liska <mliska@suse.cz>
19452 PR gcov-profile/81561
19453 * gcov.c (unblock): Make unblocking safe as we need to preserve
19454 index correspondence of blocks and block_lists.
19456 2017-08-01 Richard Biener <rguenther@suse.de>
19458 PR tree-optimization/81181
19459 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
19460 (compute_antic): ... end of iteration here.
19462 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
19464 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
19465 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
19466 (ftree-slp-vectorize): Likewise.
19467 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
19468 can no longer be set independent of flag_tree_loop_vectorize.
19469 * omp-general.c (emp_max_vf): Likewise.
19470 * opts.c (enable_fdo_optimizations): Remove references to
19471 flag_tree_vectorize, these are now implicit.
19472 (common_handle_option): Remove handling for OPT_ftree_vectorize,
19473 and leave it for the options machinery.
19475 2017-08-01 Martin Liska <mliska@suse.cz>
19477 PR middle-end/70140
19478 * builtins.c (expand_builtin_memcpy_args): Remove.
19479 (expand_builtin_memcpy): Call newly added function
19480 expand_builtin_memory_copy_args.
19481 (expand_builtin_memcpy_with_bounds): Likewise.
19482 (expand_builtin_mempcpy): Remove last argument.
19483 (expand_builtin_mempcpy_with_bounds): Likewise.
19484 (expand_builtin_memory_copy_args): New function created from
19485 expand_builtin_mempcpy_args with small modifications.
19486 (expand_builtin_mempcpy_args): Remove.
19487 (expand_builtin_stpcpy): Remove unused argument.
19488 (expand_builtin): Likewise.
19489 (expand_builtin_with_bounds): Likewise.
19491 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
19494 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
19495 print "ds:" only for immediates in generic address space.
19497 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
19500 * config/i386/i386.c (ix86_funciton_naked): New prototype.
19501 (ix86_function_ok_for_sibcall): Return false for naked functions.
19503 2017-08-01 Richard Biener <rguenther@suse.de>
19505 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
19506 (compute_antic): Seed worklist with exit block predecessors.
19507 * cfganal.c (dfs_find_deadend): For a cycle return the source
19508 of the edge closing it.
19510 2017-08-01 Tamar Christina <tamar.christina@arm.com>
19512 * config/aarch64/aarch64.c
19513 (aarch64_can_const_movi_rtx_p): Move 0 check.
19515 2017-08-01 Bin Cheng <bin.cheng@arm.com>
19517 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
19518 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
19520 * match.pd: Ditto in address comparison pattern.
19522 2017-08-01 Bin Cheng <bin.cheng@arm.com>
19524 PR tree-optimization/81627
19525 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
19526 closed ssa form for store-store chain.
19528 2017-08-01 Bin Cheng <bin.cheng@arm.com>
19530 PR tree-optimization/81620
19531 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
19532 for store-store chain.
19534 2017-08-01 Jakub Jelinek <jakub@redhat.com>
19536 PR tree-optimization/81588
19537 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
19538 ranges[i].in_p, invert comparison code ccode. For >/>=,
19539 swap rhs1 and rhs2 and comparison code unconditionally,
19540 for </<= don't do that. Don't swap rhs1/rhs2 again if
19541 ranges[i].in_p, instead invert comparison code ccode if
19542 opcode or oe->rank is BIT_IOR_EXPR.
19545 * optabs.def (vec_extract_optab, vec_init_optab): Change from
19546 a direct optab to conversion optab.
19547 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
19548 with GET_MODE_INNER as last argument instead of optab_handler.
19549 * expmed.c (extract_bit_field_1): Likewise. Use vector from
19550 vector extraction if possible and optab is available.
19551 * expr.c (store_constructor): Use convert_optab_handler instead
19552 of optab_handler. Use vector initialization from smaller
19553 vectors if possible and optab is available.
19554 * tree-vect-stmts.c (vectorizable_load): Likewise.
19555 * doc/md.texi (vec_extract, vec_init): Document that the optabs
19556 now have two modes.
19557 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
19558 of vec_init from half-sized vectors with the same element mode.
19559 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
19560 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
19561 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
19562 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
19563 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
19564 after mode in gen_vec_extract* calls.
19565 (vec_extract<mode>): Renamed to ...
19566 (vec_extract<mode><ssescalarmodelower>): ... this.
19567 (vec_extract<mode><ssehalfvecmodelower>): New expander.
19568 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
19569 element mode after mode in gen_vec_init* calls.
19570 (VEC_INIT_HALF_MODE): New mode iterator.
19571 (vec_init<mode>): Renamed to ...
19572 (vec_init<mode><ssescalarmodelower>): ... this.
19573 (vec_init<mode><ssehalfvecmodelower>): New expander.
19574 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
19575 (vec_extractv2sfsf): ... this.
19576 (vec_initv2sf): Renamed to ...
19577 (vec_initv2sfsf): ... this.
19578 (vec_extractv2si): Renamed to ...
19579 (vec_extractv2sisi): ... this.
19580 (vec_initv2si): Renamed to ...
19581 (vec_initv2sisi): ... this.
19582 (vec_extractv4hi): Renamed to ...
19583 (vec_extractv4hihi): ... this.
19584 (vec_initv4hi): Renamed to ...
19585 (vec_initv4hihi): ... this.
19586 (vec_extractv8qi): Renamed to ...
19587 (vec_extractv8qiqi): ... this.
19588 (vec_initv8qi): Renamed to ...
19589 (vec_initv8qiqi): ... this.
19590 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
19591 (vec_init<mode>): Renamed to ...
19592 (vec_init<mode><VEC_base_l>): ... this.
19593 (vec_extract<mode>): Renamed to ...
19594 (vec_extract<mode><VEC_base_l>): ... this.
19595 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
19596 (vec_initv2sfsf): ... this.
19597 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
19598 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
19599 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
19600 element mode after mode in gen_vec_init* calls.
19601 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
19602 (vec_init<mode><Vel>): ... this.
19603 (vec_extract<mode>): Renamed to ...
19604 (vec_extract<mode><Vel>): ... this.
19605 * config/aarch64/iterators.md (Vel): New mode attribute.
19606 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
19607 Add element mode after mode in gen_vec_extract* calls.
19608 * config/s390/vector.md (non_vec_l): New mode attribute.
19609 (vec_extract<mode>): Renamed to ...
19610 (vec_extract<mode><non_vec_l>): ... this.
19611 (vec_init<mode>): Renamed to ...
19612 (vec_init<mode><non_vec_l>): ... this.
19613 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
19614 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
19616 * config/arm/iterators.md (V_elem_l): New mode attribute.
19617 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
19618 (vec_extract<mode><V_elem_l>): ... this.
19619 (vec_extractv2di): Renamed to ...
19620 (vec_extractv2didi): ... this.
19621 (vec_init<mode>): Renamed to ...
19622 (vec_init<mode><V_elem_l>): ... this.
19623 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
19624 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
19625 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
19626 Add element mode after gen_vec_extract* calls.
19627 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
19628 (vec_init<mode><unitmode>): ... this.
19629 (vec_extract<mode>): Renamed to ...
19630 (vec_extract<mode><unitmode>): ... this.
19631 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
19632 (vec_init<mode><unitmode>): ... this.
19633 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
19634 (vec_initv2sfsf): ... this.
19635 (vec_extractv2sf): Renamed to ...
19636 (vec_extractv2sfsf): ... this.
19637 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
19638 Add element mode after gen_vec_extract* calls.
19639 * config/mips/mips.md (unitmode): New mode iterator.
19640 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
19641 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
19642 * config/spu/spu.md (inner_l): New mode attribute.
19643 (vec_init<mode>): Renamed to ...
19644 (vec_init<mode><inner_l>): ... this.
19645 (vec_extract<mode>): Renamed to ...
19646 (vec_extract<mode><inner_l>): ... this.
19647 * config/sparc/sparc.md (veltmode): New mode iterator.
19648 (vec_init<VMALL:mode>): Renamed to ...
19649 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
19650 * config/ia64/vect.md (vec_initv2si): Renamed to ...
19651 (vec_initv2sisi): ... this.
19652 (vec_initv2sf): Renamed to ...
19653 (vec_initv2sfsf): ... this.
19654 (vec_extractv2sf): Renamed to ...
19655 (vec_extractv2sfsf): ... this.
19656 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
19657 (vec_init<mode>): Renamed to ...
19658 (vec_init<mode><VEC_base_l>): ... this.
19659 (vec_extract<mode>): Renamed to ...
19660 (vec_extract<mode><VEC_base_l>): ... this.
19661 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
19662 (vec_initv2sfsf): ... this.
19663 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
19664 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
19665 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
19666 gen_vec_init* calls.
19668 2017-08-01 Richard Biener <rguenther@suse.de>
19670 PR tree-optimization/81297
19671 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
19672 TREE_OVERFLOW from INTEGER_CSTs.
19674 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
19676 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
19678 2017-07-31 Carl Love <cel@us.ibm.com>
19680 * config/rs6000/rs6000-c: Add support for built-in functions
19681 vector signed char vec_xl_be (signed long long, signed char *);
19682 vector unsigned char vec_xl_be (signed long long, unsigned char *);
19683 vector signed int vec_xl_be (signed long long, signed int *);
19684 vector unsigned int vec_xl_be (signed long long, unsigned int *);
19685 vector signed long long vec_xl_be (signed long long, signed long long *);
19686 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
19687 vector signed short vec_xl_be (signed long long, signed short *);
19688 vector unsigned short vec_xl_be (signed long long, unsigned short *);
19689 vector double vec_xl_be (signed long long, double *);
19690 vector float vec_xl_be (signed long long, float *);
19691 * config/rs6000/altivec.h (vec_xl_be): Add #define.
19692 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
19693 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
19695 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
19696 (altivec_expand_builtin): Add switch statement to call
19697 altivec_expand_xl_be for each builtin.
19698 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
19699 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
19700 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
19701 __builtin_vsx_le_be_v16qi.
19702 * doc/extend.texi: Update the built-in documentation file for the
19703 new built-in functions.
19705 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19708 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
19710 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
19712 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19714 * config.gcc: Add z14.
19715 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
19716 CPU model numbers for z13s and z14.
19717 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
19719 * config/s390/s390-opts.h (enum processor_type): Rename
19720 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19721 * config/s390/s390.c (processor_table): Add field for CPU name to
19722 be passed to Binutils.
19723 (s390_asm_output_machine_for_arch): Use the new field in
19724 processor_table for Binutils.
19725 (s390_expand_builtin): Replace arch12 with z14.
19726 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19727 (s390_get_sched_attrmask): Likewise.
19728 (s390_get_unit_mask): Likewise.
19729 * config/s390/s390.opt: Add z14 to processor_type enum.
19731 2017-07-31 Martin Jambor <mjambor@suse.cz>
19734 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
19735 regardless of optimization level.
19737 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
19738 Martin Liska <mliska@suse.cz>
19740 * predict.def: Remove old comment and adjust probability.
19741 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
19742 PREDICT statements.
19744 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19747 * config/i386/i386.c (ix86_function_naked): New function.
19748 (ix86_can_use_return_insn_p): Return false for naked functions.
19749 (ix86_expand_prologue): Skip prologue for naked functions.
19750 (ix86_expand_epilogue): Skip epilogue for naked functions
19751 and emit trap instruction.
19752 (ix86_warn_func_return): New function.
19753 (ix86_attribute_table): Add "naked" attribute specification.
19754 (TARGET_WARN_FUNC_RETURN): Define.
19755 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
19757 2017-07-31 Martin Liska <mliska@suse.cz>
19759 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
19760 (dump_gimple_bb_header): Always dump BB info.
19761 (pp_cfg_jump): Do not append info about BB when dumping a jump.
19763 2017-07-31 Martin Liska <mliska@suse.cz>
19766 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
19767 also with current_function_decl non-null equality.
19769 2017-07-31 Jakub Jelinek <jakub@redhat.com>
19772 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
19773 change type to the element type, instead add eltype variable and
19774 use it where we are interested in the element type.
19776 PR tree-optimization/81603
19777 * ipa-polymorphic-call.c
19778 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
19779 offset arithmetic in offset_int, bail out if the resulting bit offset
19780 doesn't fit into shwi.
19782 2017-07-31 Martin Liska <mliska@suse.cz>
19784 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
19785 (gimplify_save_expr): Fix comment.
19787 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
19790 * config/i386/i386.c (ix86_function_arg): Update arguments for
19792 (ix86_compute_frame_layout): Set the initial stack offset to
19793 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
19794 INCOMING_FRAME_SP_OFFSET.
19795 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
19796 stack before exception handler returns.
19797 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
19798 the 'ERROR_CODE' for exception handler.
19800 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
19802 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
19803 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
19804 (ASM_OUTPUT_REG_POP): Ditto.
19805 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
19806 instead of asm_fprintf to output pure string.
19808 2017-07-29 Jakub Jelinek <jakub@redhat.com>
19810 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
19811 to imported_module_or_decl hook.
19812 (debug_nothing_tree_tree_tree_bool): Remove.
19813 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
19814 * debug.c (do_nothing_debug_hooks): Use
19815 debug_nothing_tree_tree_tree_bool_bool instead of
19816 debug_nothing_tree_tree_tree_bool.
19817 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19818 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
19819 * sdbout.c (sdb_debug_hooks): Likewise.
19820 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
19821 (gen_namespace_die): Add DW_AT_export_symbols attribute if
19823 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
19824 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
19825 attribute, don't add anything.
19827 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19829 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
19830 (fold_build2_stat_loc): Likewise.
19831 (fold_build3_stat_loc): Likewise.
19832 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
19833 (fold_build1_loc): Remove macro.
19834 (fold_build2_loc): Likewise.
19835 (fold_build3_loc): Likewise.
19837 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19839 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
19840 (gimple_build_debug_bind_source_stat): Likewise.
19841 * gimple.h (gimple_build_debug_bind): Remove macro.
19842 (gimple_build_debug_bind_source): Likewise.
19844 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19846 * bitmap.c (bitmap_alloc): Adjust.
19847 (bitmap_gc_alloc): Likewise.
19848 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
19850 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19852 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
19853 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
19854 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
19855 (bitmap_gc_alloc_stat): Likewise.
19856 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
19858 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19860 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
19861 * rtl.h (shallow_copy_rtx): Remove macro.
19863 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19865 * emit-rtl.c (gen_raw_REG): Adjust.
19866 * gengenrtl.c (gendef): Likewise.
19867 * rtl.c (rtx_alloc_stat): Remove _stat from name.
19868 * rtl.h (rtx_alloc): Remove macro.
19870 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19872 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
19873 (build_tree_list_stat): Likewise.
19874 * tree.h (build_tree_list): Remove macro.
19875 (build_tree_list_vec): Likewise.
19877 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19879 * tree.c (make_vector_stat): Remove _stat from name.
19880 (build_vector_stat): Likewise.
19881 * tree.h (make_vector_stat): Remove macro.
19882 (build_vector_stat): Likewise.
19884 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19886 * tree.h (build_var_debug_value): Remove prototype.
19888 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19890 * tree.c (tree_cons_stat): Remove _stat from name.
19891 * tree.h (tree_cons): Remove macro.
19893 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19895 * tree.c (build_vl_exp_stat): Remove _stat from name.
19896 * tree.h (build_vl_exp): Remove macro.
19898 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19900 * tree.c (build_decl_stat): Remove _stat from name.
19901 * tree.h (build_decl): Remove macro.
19903 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19905 * gimple.c (gimple_build_with_ops_stat): Adjust.
19906 (gimple_alloc_stat): Remove _stat from name.
19907 * gimple.h (gimple_alloc): Remove macro.
19909 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19911 * tree.c (make_tree_vec_stat): Remove _stat from name.
19912 (grow_tree_vec_stat): Likewise.
19913 * tree.h (make_tree_vec_stat): Adjust prototype.
19914 (grow_tree_vec_stat): Likewise.
19915 (make_tree_vec): Remove macro.
19916 (grow_tree_vec): Likewise.
19918 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19920 * fold-const.c (fold_build1_stat_loc): Adjust.
19921 (fold_build2_stat_loc): Likewise.
19922 (fold_build3_stat_loc): Likewise.
19923 * tree.c (build0_stat): Remove _stat from name.
19924 (build1_stat): Likewise.
19925 (build2_stat): Likewise.
19926 (build3_stat): Likewise.
19927 (build4_stat): Likewise.
19928 (build5_stat): Likewise.
19929 * tree.h (build1_loc): Remove macro, and rename _stat function
19931 (build2_loc): Likewise.
19932 (build3_loc): Likewise.
19933 (build4_loc): Likewise.
19934 (build5_loc): Likewise.
19936 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19938 * tree.c (make_int_cst_stat): Remove _stat from name.
19939 * tree.h (make_int_cst_stat): Adjust prototype.
19940 (make_int_cst): Remove macro.
19942 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19944 * tree.c (make_tre_binfo_stat): Remove _stat from name.
19945 * tree.h (make_tree_binfo_stat): Adjust prototype.
19946 (make_tree_binfo): Remove.
19948 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19950 * tree.c (copy_node_stat): Rename to copy_node.
19951 (build_distinct_type_copy): Adjust.
19952 * tree.h (copy_node_stat): Adjust prototype.
19953 (copy_node): Remove macro.
19955 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19957 * tree.c (make_node_stat): rename to make_node.
19958 (build_tree_list_stat): Adjust.
19959 (build0_stat): Likewise.
19960 (build2_stat): Likewise.
19961 (build3_stat): Likewise.
19962 (build4_stat): Likewise.
19963 (build5_stat): Likewise.
19964 (build_decl_stat): Likewise.
19965 * tree.h (make_node_stat): Adjust prototype.
19966 (make_node): remove macro.
19968 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
19970 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
19971 (PPC_FEATURE2_SCV): Likewise.
19972 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
19974 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19976 * config/aarch64/aarch64.c
19977 (aarch64_internal_mov_immediate): Add new special pattern.
19978 * config/aarch64/aarch64.md (*movdi_aarch64):
19979 Add reg/32bit const mov case.
19981 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19982 Richard Sandiford <richard.sandiford@linaro.org>
19984 * config/aarch64/aarch64.md (mov<mode>): Generalize.
19985 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
19986 Add integer and movi cases.
19987 (movi-split-hf-df-sf split, fp16): New.
19988 (enabled): Added TARGET_FP_F16INST.
19989 * config/aarch64/iterators.md (GPF_HF): New.
19990 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
19992 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19994 * config/aarch64/aarch64.c
19995 (aarch64_simd_container_mode): Add prototype.
19996 (aarch64_expand_mov_immediate): Add HI support.
19997 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
19998 (aarch64_can_const_movi_rtx_p): New.
19999 (aarch64_preferred_reload_class):
20000 Remove restrictions of using FP registers for certain SIMD operations.
20001 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
20002 (aarch64_valid_floating_const): Add integer move validation.
20003 (aarch64_simd_imm_scalar_p): Remove.
20004 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
20005 (aarch64_legitimate_constant_p): Expand list of supported cases.
20006 * config/aarch64/aarch64-protos.h
20007 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
20008 (aarch64_reinterpret_float_as_int): New.
20009 (aarch64_simd_imm_scalar_p): Remove.
20010 * config/aarch64/constraints.md (Uvi): New.
20011 (Dd): Split into Ds and new Dd.
20012 * config/aarch64/aarch64.md (*movsi_aarch64):
20014 (*movdi_aarch64): Add SIMD mov case.
20016 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20018 * tree-predcom.c: (struct chain): Handle store-store chain in which
20019 stores for elimination only store loop invariant values.
20020 (execute_pred_commoning_chain): Ditto.
20021 (prepare_initializers_chain_store_elim): Ditto.
20022 (prepare_finalizers): Ditto.
20023 (is_inv_store_elimination_chain): New function.
20024 (initialize_root_vars_store_elim_1): New function.
20026 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20028 * tree-predcom.c: Revise general description of the pass.
20029 (enum chain_type): New enum type for store elimination.
20030 (struct chain): New field supporting store elimination.
20031 (struct component): Ditto.
20032 (dump_chain): Dump store-stores chain.
20033 (release_chain): Release resources.
20034 (split_data_refs_to_components): Compute and create component
20035 contains only stores for elimination.
20036 (get_chain_last_ref_at): New function.
20037 (make_invariant_chain): Initialization.
20038 (make_rooted_chain): Specify chain type in parameter and record it.
20039 (add_looparound_copies): Skip for store-stores chain.
20040 (determine_roots_comp): Compute type of chain and pass it to
20042 (initialize_root_vars_store_elim_2): New function.
20043 (finalize_eliminated_stores): New function.
20044 (remove_stmt): Handle store for elimination.
20045 (execute_pred_commoning_chain): Execute predictive commoning on
20046 store-store chains.
20047 (determine_unroll_factor): Skip unroll for store-stores chain.
20048 (prepare_initializers_chain_store_elim): New function.
20049 (prepare_initializers_chain): Hanlde store-store chain.
20050 (prepare_finalizers_chain, prepare_finalizers): New function.
20051 (tree_predictive_commoning_loop): Return integer value indicating
20052 if loop is unrolled or lcssa form is corrupted.
20053 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
20055 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20057 * tree-predcom.c (initialize_root): Delete.
20058 (execute_pred_commoning_chain): Initialize root vars and replace
20059 reference of non-combined chain directly, rather than call above
20062 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20064 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
20065 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
20067 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20069 * tree-predcom.c (struct chain): New field init_seq.
20070 (release_chain): Release init_seq.
20071 (prepare_initializers_chain): Record intialization stmts in above
20073 (insert_init_seqs): New function.
20074 (tree_predictive_commoning_loop): Call insert_init_seqs.
20076 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20078 * tree-predcom.c (determine_roots_comp): Skip trivial components.
20080 2017-07-28 Richard Biener <rguenther@suse.de>
20082 * match.pd: Remove superfluous :c.
20083 * genmatch.c (simplify::id): Add member.
20084 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
20086 (current_id): New global.
20087 (dt_node::parent): Move from ...
20088 (dt_operand::parent): ... here. Add for_id member.
20089 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
20090 (decision_tree::find_node): Relax order requirement when
20091 merging DT_TRUE nodes to ones inbetween the current simplify
20092 and the one we try to merge with. Add diagnostic whenever
20093 we need to enforce pattern order by not merging.
20094 (decision_tree::insert): Set current_id.
20095 (decision_tree::print_node): Dump parent node and for_id.
20096 (parser::last_id): Add member.
20097 (parser::push_simplify): Assign unique id.
20098 (parser::parser): Initialize last_id.
20100 2017-07-28 Martin Liska <mliska@suse.cz>
20103 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
20104 gimple_build_debug_bind.
20106 2017-07-28 Richard Biener <rguenther@suse.de>
20108 PR tree-optimization/81502
20109 * match.pd: Add pattern combining BIT_INSERT_EXPR with
20111 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
20113 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
20114 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
20115 for BIT_FIELD_REF args.
20116 * fold-const.c (make_bit_field_ref): Likewise.
20117 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
20119 2017-07-28 Jakub Jelinek <jakub@redhat.com>
20122 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
20123 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
20124 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
20125 Or it into SANITIZER_UNDEFINED.
20126 * ubsan.c: Include gimple-fold.h and varasm.h.
20127 (ubsan_expand_ptr_ifn): New function.
20128 (instrument_pointer_overflow): New function.
20129 (maybe_instrument_pointer_overflow): New function.
20130 (instrument_object_size): Formatting fix.
20131 (pass_ubsan::execute): Call instrument_pointer_overflow
20132 and maybe_instrument_pointer_overflow.
20133 * internal-fn.c (expand_UBSAN_PTR): New function.
20134 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
20135 * sanitizer.def (__ubsan_handle_pointer_overflow,
20136 __ubsan_handle_pointer_overflow_abort): New builtins.
20137 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
20138 * internal-fn.def (UBSAN_PTR): New internal function.
20139 * opts.c (sanitizer_opts): Add pointer-overflow.
20140 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
20141 * fold-const.c (build_range_check): Compute pointer range check in
20142 integral type if pointer arithmetics would be needed. Formatting
20145 2017-07-28 Martin Liska <mliska@suse.cz>
20148 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
20149 parameters that are of a variable-length.
20151 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
20153 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
20155 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
20156 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
20157 (CRT_CALL_STATIC_FUNCTION): Likewise.
20158 (ASM_DEFAULT_SPEC): New define.
20159 (ASM_SPEC32): Likewise.
20160 (ASM_SPEC64): Likewise.
20161 (ASM_SPEC_COMMON): Likewise.
20162 (ASM_SPEC): Likewise.
20163 (INVALID_64BIT): Likewise.
20164 (LINK_OS_DEFAULT_SPEC): Likewise.
20165 (LINK_OS_SPEC32): Likewise.
20166 (LINK_OS_SPEC64): Likewise.
20167 (POWERPC_LINUX): Likewise.
20168 (PTRDIFF_TYPE): Likewise.
20169 (RESTORE_FP_PREFIX): Likewise.
20170 (RESTORE_FP_SUFFIX): Likewise.
20171 (SAVE_FP_PREFIX): Likewise.
20172 (SAVE_FP_SUFFIX): Likewise.
20173 (SIZE_TYPE): Likewise.
20174 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
20175 (TARGET_64BIT): Likewise.
20176 (TARGET_64BIT): Likewise.
20177 (TARGET_AIX): Likewise.
20178 (WCHAR_TYPE_SIZE): Likewise.
20179 (WCHAR_TYPE): Undefine.
20180 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
20181 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
20182 (CPP_OS_RTEMS_SPEC): Delete.
20183 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
20184 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
20186 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
20188 2017-07-28 Jakub Jelinek <jakub@redhat.com>
20190 PR tree-optimization/81578
20191 * tree-parloops.c (build_new_reduction): Bail out if
20192 reduction_code isn't one of the standard OpenMP reductions.
20193 Move the details printing after that decision.
20195 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20197 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
20198 related to reload_in_progress.
20199 (splat_input_operand): Likewise.
20200 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
20202 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
20204 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
20205 (TARGET_INSTANTIATE_DECLS): Likewise.
20206 (legitimate_indexed_address_p): Delete reload_in_progress code.
20207 (rs6000_debug_legitimate_address_p): Likewise.
20208 (rs6000_eliminate_indexed_memrefs): Likewise.
20209 (rs6000_emit_le_vsx_store): Likewise.
20210 (rs6000_emit_move_si_sf_subreg): Likewise.
20211 (rs6000_emit_move): Likewise.
20212 (register_to_reg_type): Likewise.
20213 (rs6000_pre_atomic_barrier): Likewise.
20214 (rs6000_machopic_legitimize_pic_address): Likewise.
20215 (rs6000_allocate_stack_temp): Likewise.
20216 (rs6000_address_for_fpconvert): Likewise.
20217 (rs6000_address_for_altivec): Likewise.
20218 (rs6000_secondary_memory_needed_rtx): Delete function.
20219 (rs6000_check_sdmode): Likewise.
20220 (rs6000_alloc_sdmode_stack_slot): Likewise.
20221 (rs6000_instantiate_decls): Likewise.
20222 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
20223 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
20224 Delete reload_in_progress.
20225 (*vec_reload_and_plus_<mptrsize>): Likewise.
20226 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
20227 (vsx_div_v2di): Likewise.
20228 (vsx_udiv_v2di): Likewise.
20230 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20232 * config/rs6000/rs6000.opt (mlra): Replace with stub.
20233 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
20234 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
20235 (rs6000_debug_reg_global): Delete print of LRA status.
20236 (rs6000_option_override_internal): Delete dead LRA related code.
20237 (rs6000_lra_p): Delete function.
20238 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
20240 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
20242 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
20243 * config/riscv/rtems.h: New file.
20245 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20246 Sudakshina Das <sudi.das@arm.com>
20248 * config/aarch64/aarch64.md
20249 (define_split for and<mode>3nr_compare): Move
20250 non aarch64_logical_operand to a register.
20251 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
20252 register immediate operand to a register.
20253 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
20255 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20257 PR middle-end/81564
20258 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
20260 2017-07-27 Richard Biener <rguenther@suse.de>
20262 PR tree-optimization/81573
20263 PR tree-optimization/81494
20264 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
20265 multi defuse cycle case.
20267 2017-07-27 Richard Biener <rguenther@suse.de>
20269 PR tree-optimization/81571
20270 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
20273 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
20275 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
20276 earlier and only if MASK_FPU is set. Adjust formatting.
20278 2017-07-27 Martin Liska <mliska@suse.cz>
20280 * opt-functions.awk: Add validation of value of Init.
20281 * optc-gen.awk: Pass new argument.
20283 2017-07-27 Martin Liska <mliska@suse.cz>
20285 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
20286 Fix wrong condition.
20288 2017-07-27 Martin Liska <mliska@suse.cz>
20290 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
20291 BBs and edges seen by autoFDO.
20293 2017-07-27 Richard Biener <rguenther@suse.de>
20295 PR tree-optimization/81502
20296 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
20297 with incompatible but same sized type.
20298 (execute_update_addresses_taken): Likewise.
20300 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
20302 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
20303 flag_tree_loop_vectorize rather than flag_tree_vectorize.
20305 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20308 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
20309 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
20310 Change s_operand to memory_operand.
20312 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
20314 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
20315 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
20316 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
20317 Emit instructions rather than returning an expression. Handle TFmode
20318 and KFmode by casting to TImode.
20319 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
20320 (rs6000_emit_le_vsx_store): Likewise.
20321 * config/rs6000/vsx.md (VSX_TI): New iterator.
20322 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
20323 (*vsx_le_undo_permute_<mode>): Likewise.
20324 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
20325 emit the split sequence.
20326 (*vsx_le_perm_store_<mode>): Likewise.
20328 2017-07-27 Jakub Jelinek <jakub@redhat.com>
20330 PR tree-optimization/81555
20331 PR tree-optimization/81556
20332 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
20333 if true, force CHANGED for the recursive invocation.
20334 (reassociate_bb): Remember original length of ops array, pass
20335 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
20337 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
20338 attributes for noipa attribute. For naked attribute use
20339 lookup_attribute first before lookup_attribute_spec.
20340 * final.c (rest_of_handle_final): Disable IPA RA for functions with
20342 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
20343 for functions with noipa attribute.
20344 (cgraph_externally_visible_p): Return true for functions with noipa
20346 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
20347 for functions with noipa attribute.
20348 * doc/extend.texi: Document noipa function attribute.
20349 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
20350 also for functions with noipa attribute.
20351 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
20353 2017-07-26 Andrew Pinski <apinski@cavium.com>
20355 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
20356 vec_unalign_load_cost and vec_unalign_store_cost.
20358 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
20360 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
20361 -mvsx-small-integer option.
20362 (ISA_3_0_MASKS_IEEE): Likewise.
20363 (OTHER_VSX_VECTOR_MASKS): Likewise.
20364 (POWERPC_MASKS): Likewise.
20365 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
20366 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
20367 code, only testing for DImode being allowed in non-VSX floating
20369 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
20370 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
20372 (rs6000_option_override_internal): Delete -mvsx-small-integer.
20373 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
20374 TARGET_P8_VECTOR test.
20375 (rs6000_secondary_reload_simple_move): Likewise.
20376 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
20377 since TARGET_P9_VECTOR was already tested.
20378 (rs6000_opt_masks): Remove -mvsx-small-integer.
20379 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
20380 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
20382 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
20383 test for TARGET_VEXTRACTUB was used, and that uses
20385 (p9 extract splitter): Likewise.
20386 (vsx_extract_<mode>_di_p9): Likewise.
20387 (vsx_extract_<mode>_store_p9): Likewise.
20388 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
20389 for TARGET_P9_VECTOR was used. Delete code that is now dead with
20390 the elimination of TARGET_VSX_SMALL_INTEGER.
20391 (vsx_extract_<mode>_p8): Likewise.
20392 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
20393 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
20394 (vsx_set_<mode>_p9): Likewise.
20395 (vsx_set_v4sf_p9): Likewise.
20396 (vsx_set_v4sf_p9_zero): Likewise.
20397 (vsx_insert_extract_v4sf_p9): Likewise.
20398 (vsx_insert_extract_v4sf_p9_2): Likewise.
20399 * config/rs6000/rs6000.md (sign extend splitter): Change
20400 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
20401 (floatsi<mode>2_lfiwax_mem): Likewise.
20402 (floatunssi<mode>2_lfiwzx_mem): Likewise.
20403 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
20404 since a test for TARGET_P9_VECTOR was used.
20405 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
20406 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
20407 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
20408 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
20409 TARGET_P8_VECTOR test.
20410 (fix_trunc<mode>si2_stfiwx): Likewise.
20411 (fix_trunc<mode>si2_internal): Likewise.
20412 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
20413 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
20415 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
20416 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
20417 TARGET_P8_VECTOR test.
20418 (fixuns_trunc<mode>si2_stfiwx): Likewise.
20419 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
20420 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
20422 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
20423 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
20424 since a test for TARGET_P9_VECTOR was used.
20425 (splitter for loading small constants): Likewise.
20427 2017-07-26 Andrew Pinski <apinski@cavium.com>
20429 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
20432 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
20435 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
20436 (fp_valid_at): Likewise.
20438 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
20440 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
20441 (qdf24xx_addrcost_table): Likewise.
20442 (cortexa57_tunings): Update to use generic_branch_cost.
20443 (cortexa72_tunings): Likewise.
20444 (cortexa73_tunings): Likewise.
20445 (qdf24xx_tunings): Likewise.
20447 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
20449 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
20450 (thunderx2t99_branch_cost): Likewise.
20451 (cortexa35_tunings): Update to use generic_branch_cost.
20452 (cortexa53_tunings): Likewise.
20453 (cortexa57_tunings): Likewise.
20454 (cortexa72_tunings): Likewise.
20455 (cortexa73_tunings): Likewise.
20456 (thunderx2t99_tunings): Likewise.
20458 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
20460 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
20461 (sparc_option_override): Honour MASK_FSMULD.
20462 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
20463 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
20464 * config/sparc/sparc.opt (mfsmuld): New option.
20465 * doc/invoke.texi (mfsmuld): Document option.
20467 2017-07-26 Marek Polacek <polacek@redhat.com>
20469 PR middle-end/70992
20470 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
20472 2017-07-26 Richard Biener <rguenther@suse.de>
20474 * gimple-match-head.c (do_valueize): Return OP if valueize
20476 (get_def): New helper to get at the def stmt of a SSA name
20477 if valueize allows.
20478 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
20479 do_valueize to get at the def stmt.
20480 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
20482 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
20484 PR middle-end/46932
20485 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
20487 2017-07-26 Martin Liska <mliska@suse.cz>
20490 * function.c (expand_function_start): Make expansion of
20491 nonlocal_goto_save_area after parm_birth_insn.
20493 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
20495 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
20496 from all CPU target flags enable members.
20498 2017-07-26 Richard Biener <rguenther@suse.de>
20500 * genmatch.c (dt_simplify::gen): Make iterator vars const.
20501 (decision_tree::gen): Make 'type' const.
20502 (write_predicate): Likewise.
20504 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
20506 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
20507 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
20508 (rs6000_option_override_internal): Likewise.
20509 (rs6000_expand_vector_set): Likewise.
20510 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
20511 (TARGET_UPPER_REGS_SF): Likewise.
20512 (TARGET_UPPER_REGS_DI): Likewise.
20513 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
20514 (TARGET_DIRECT_MOVE_64BIT): Likewise.
20515 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
20516 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
20517 (Splitters for DI constants in Altivec registers): Likewise.
20518 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
20519 (vsx_set_v4sf_p9): Likewise.
20520 (vsx_set_v4sf_p9_zero): Likewise.
20521 (vsx_insert_extract_v4sf_p9): Likewise.
20522 (vsx_insert_extract_v4sf_p9_2): Likewise.
20524 2017-07-25 Carl Love <cel@us.ibm.com>
20526 * doc/extend.texi: Update the built-in documentation file for the
20527 existing built-in functions
20528 vector signed char vec_cnttz (vector signed char);
20529 vector unsigned char vec_cnttz (vector unsigned char);
20530 vector signed short vec_cnttz (vector signed short);
20531 vector unsigned short vec_cnttz (vector unsigned short);
20532 vector signed int vec_cnttz (vector signed int);
20533 vector unsigned int vec_cnttz (vector unsigned int);
20534 vector signed long long vec_cnttz (vector signed long long);
20535 vector unsigned long long vec_cnttz (vector unsigned long long);
20537 2017-07-25 Andrew Pinski <apinski@cavium.com>
20539 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
20540 accesses where the use is for the first operand of a BIT_INSERT.
20542 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
20545 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
20546 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
20548 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
20550 * config/i386/gstabs.h: Delete.
20551 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
20553 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
20555 * config/i386/i386.c (ix86_decompose_address): Do not check for
20556 register RTX when looking at index_reg or base_reg.
20557 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
20559 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
20561 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
20562 to update EH info here.
20564 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20566 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
20568 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20570 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
20572 2017-07-25 Torsten Duwe <duwe@suse.de>
20574 * common.opt: Introduce -fpatchable-function-entry
20575 command line option, and its variables function_entry_patch_area_size
20576 and function_entry_patch_area_start.
20577 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
20578 including a two-value parser.
20579 * target.def (print_patchable_function_entry): New target hook.
20580 * targhooks.h (default_print_patchable_function_entry): New function.
20581 * targhooks.c (default_print_patchable_function_entry): Likewise.
20582 * toplev.c (process_options): Switch off IPA-RA if
20583 patchable function entries are being generated.
20584 * varasm.c (assemble_start_function): Look at the
20585 patchable-function-entry command line switch and current
20586 function attributes and maybe generate NOP instructions by
20587 calling the print_patchable_function_entry hook.
20588 * doc/extend.texi: Document patchable_function_entry attribute.
20589 * doc/invoke.texi: Document -fpatchable_function_entry
20590 command line option.
20591 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
20593 * doc/tm.texi: Re-generate.
20595 2017-07-25 Jakub Jelinek <jakub@redhat.com>
20598 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
20599 TARGET_AVX512DQ rather than TARGET_AVX512BW.
20601 2017-07-25 Tamar Christina <tamar.christina@arm.com>
20603 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
20605 2017-07-25 Richard Biener <rguenther@suse.de>
20607 PR tree-optimization/81455
20608 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
20609 not walk in cycles when looking for guards.
20611 2017-07-25 Richard Biener <rguenther@suse.de>
20613 PR tree-optimization/81529
20614 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
20615 when optimizing backedge uses.
20617 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
20619 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
20621 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
20622 to dl_section_ref. On AIX, append an expression to subtract
20623 the size of the section length to dl_section_ref.
20625 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
20627 * configure.ac: If any of the config.* scripts fail, exit 1.
20628 * configure: Regenerate.
20630 2017-07-25 Richard Biener <rguenther@suse.de>
20632 PR middle-end/81546
20633 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
20634 of immediate uses, be more verbose on errors.
20636 2017-07-25 Richard Biener <rguenther@suse.de>
20638 PR tree-optimization/81510
20639 * tree-vect-loop.c (vect_is_simple_reduction): When the
20640 reduction stmt is not inside the loop bail out.
20642 2017-07-25 Richard Biener <rguenther@suse.de>
20644 PR tree-optimization/81303
20645 * tree-vect-loop-manip.c (vect_loop_versioning): Build
20646 profitability check against LOOP_VINFO_NITERSM1.
20648 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20650 * domwalk.c (cmp_bb_postorder): Simplify.
20651 (sort_bbs_postorder): New function. Use it...
20652 (dom_walker::walk): ...here to optimize common cases.
20654 2017-07-25 Martin Liska <mliska@suse.cz>
20657 * ipa-visibility.c (function_and_variable_visibility): Make the
20658 redirection just on target that supports aliasing.
20659 Fix GNU coding style.
20661 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
20664 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
20666 * config/arm/rtems.h: Likewise.
20667 * config/bfin/rtems.h: Likewise.
20668 * config/i386/rtemself.h: Likewise.
20669 * config/lm32/rtems.h: Likewise.
20670 * config/m32c/rtems.h: Likewise.
20671 * config/m68k/rtemself.h: Likewise.
20672 * config/microblaze/rtems.h: Likewise.
20673 * config/mips/rtems.h: Likewise.
20674 * config/moxie/rtems.h: Likewise.
20675 * config/nios2/rtems.h: Likewise.
20676 * config/powerpcspe/rtems.h: Likewise.
20677 * config/rs6000/rtems.h: Likewise.
20678 * config/rtems.h: Likewise.
20679 * config/sh/rtems.h: Likewise.
20680 * config/sh/rtemself.h: Likewise.
20681 * config/sparc/rtemself.h: Likewise.
20683 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20686 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
20687 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
20688 * tree-ssa-structalias.c (alias_get_name): Same.
20690 2017-07-25 Bin Cheng <bin.cheng@arm.com>
20693 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
20694 instructions if no du chain is found.
20696 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20698 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
20700 2017-07-25 Richard Biener <rguenther@suse.de>
20702 PR middle-end/81505
20703 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
20706 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
20708 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
20709 upper-regs options.
20710 (ISA_2_7_MASKS_SERVER): Likewise.
20711 (ISA_3_0_MASKS_IEEE): Likewise.
20712 (OTHER_P8_VECTOR_MASKS): Likewise.
20713 (OTHER_VSX_VECTOR_MASKS): Likewise.
20714 (POWERPC_MASKS): Likewise.
20715 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
20716 duplicate list of options.
20717 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
20718 explicit -mupper-regs options.
20719 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
20720 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
20721 alias for -mupper-regs-df.
20722 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20723 (rs6000_init_hard_regno_mode_ok): Likewise.
20724 (rs6000_option_override_internal): Likewise.
20725 (rs6000_opt_masks): Likewise.
20726 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
20727 options in terms of whether -mvsx or -mpower8-vector was used.
20728 (TARGET_UPPER_REGS_DI): Likewise.
20729 (TARGET_UPPER_REGS_SF): Likewise.
20730 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
20731 -mupper-regs-* options.
20733 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20735 * passes.c (emergency_dump_function): Print some empty lines and a
20736 header before the RTL dump.
20738 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20740 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
20742 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
20745 * config/aarch64/aarch64.c (aarch64_classify_symbol):
20746 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
20748 2017-07-24 Carl Love <cel@us.ibm.com>
20750 * config/rs6000/rs6000-c.c: Add support for built-in functions
20751 vector float vec_extract_fp32_from_shorth (vector unsigned short);
20752 vector float vec_extract_fp32_from_shortl (vector unsigned short);
20753 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
20754 vec_extract_fp_from_shortl): Add defines for the two builtins.
20755 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
20756 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
20758 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
20759 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
20760 * doc/extend.texi: Update the built-in documentation file for the
20761 new built-in function.
20763 2017-07-24 Jakub Jelinek <jakub@redhat.com>
20766 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
20768 * doc/generic.texi: Likewise.
20769 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
20770 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
20772 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
20774 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
20775 (aarch64_mls_elt_merge<mode>): Likewise.
20777 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
20779 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
20780 having __cxa_atexit.
20782 2017-07-23 Michael Collison <michael.collison@arm.com>
20784 * config/arm/arm.c (arm_option_override): Deprecate
20785 use of -mstructure-size-boundary.
20786 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
20787 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
20789 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20792 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
20793 Reduce cost estimate for direct moves.
20795 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
20798 * config/i386/i386.c (ix86_option_override_internal): Disable
20799 BMI, BMI2 and TBM instructions for -m16.
20801 2017-07-21 Carl Love <cel@us.ibm.com>
20803 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20804 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20805 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20806 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20807 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20808 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20809 VMULOSW): New enum "unspec" values.
20810 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20811 altivec_vmulosw): New patterns.
20812 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20813 VMULOSW): Add definitions.
20815 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
20817 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
20818 (qdf24xx): Likewise.
20819 * config/aarch64/aarch64-options-extensions.def (rdma); New.
20820 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
20821 (AARCH64_FL_V8_1): Renumber.
20822 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
20823 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
20824 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
20825 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
20826 rdma to feature modifiers list.
20828 2017-07-21 Yury Gribov <tetra2005@gmail.com>
20830 PR middle-end/56727
20831 * ipa-visibility (function_and_variable_visibility): Convert
20832 recursive PLT call to direct call if appropriate.
20834 2017-07-21 Andrew Pinski <apinski@cavium.com>
20836 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
20837 operand 1 to see if the types precision matches.
20838 * fold-const.c (operand_equal_p): Likewise.
20840 2017-07-21 Richard Biener <rguenther@suse.de>
20842 PR tree-optimization/81303
20843 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
20844 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
20845 (vect_peeling_hash_get_lowest_cost): Adjust.
20846 (vect_enhance_data_refs_alignment): Likewise. Use
20847 vect_get_peeling_costs_all_drs to compute the penalty for no
20848 peeling to match up costs.
20850 2017-07-21 Richard Biener <rguenther@suse.de>
20852 PR tree-optimization/81500
20853 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
20854 we didn't identify a reduction path.
20856 2017-07-21 Tom de Vries <tom@codesourcery.com>
20857 Cesar Philippidis <cesar@codesourcery.com>
20859 PR gcov-profile/81442
20860 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
20863 2017-07-21 Tom de Vries <tom@codesourcery.com>
20866 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
20868 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
20869 nvptx_override_options_after_change.
20871 2017-07-21 Ulrich Drepper <drepper@redhat.com>
20873 * dwarf2out.c (output_file_names): Avoid double testing for
20874 dwarf_version >= 5.
20876 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
20878 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
20880 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20882 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
20884 (try_crossjump_to_edge): Do not punt on partitioned functions.
20886 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20888 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
20889 Put all BBs reachable only via paths crossing cold region to cold
20891 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
20893 2016-07-21 Richard Biener <rguenther@suse.de>
20895 PR tree-optimization/81303
20896 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
20897 into account prologue and epilogue iterations when raising
20898 min_profitable_iters to sth at least covering one vector iteration.
20900 2017-07-21 Tamar Christina <tamar.christina@arm.com>
20902 * config/arm/arm.c (arm_test_cpu_arch_dat):
20905 2017-07-20 Nathan Sidwell <nathan@acm.org>
20907 Remove TYPE_METHODS.
20908 * tree.h (TYPE_METHODS): Delete.
20909 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
20910 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
20911 (dbxout_type_methods): Scan TYPE_FIELDS.
20912 (dbxout_type): Don't check TYPE_METHODS here.
20913 * function.c (use_register_for_decl): Always ignore register for
20914 class types when not optimizing.
20915 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
20916 * tree.c (free_lang_data_in_type): Stitch out member functions and
20917 templates from TYPE_FIELDS.
20918 (build_distinct_type_copy, verify_type_variant,
20919 verify_type): Member fns are on TYPE_FIELDS.
20920 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
20921 * tree-pretty-print.c (dump_generic_node): Likewise.
20923 2017-07-20 Jakub Jelinek <jakub@redhat.com>
20926 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
20927 V2TImode and V4TImode.
20928 (ix86_expand_vector_extract): Likewise.
20929 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
20930 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
20931 (ssescalarmode): Handle V4TImode and V2TImode.
20932 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
20933 (*vec_extractv2ti, *vec_extractv4ti): New insns.
20934 (VEXTRACTI128_MODE): New mode iterator.
20935 (splitter for *vec_extractv?ti first element): New.
20936 (VEC_INIT_MODE): New mode iterator.
20937 (vec_init<mode>): Consolidate 3 expanders into one using
20938 VEC_INIT_MODE mode iterator.
20940 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20942 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
20943 non_spilled_static_chain_regno_p.
20945 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20947 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
20949 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
20951 * bb-reorder.c (connect_traces): Allow copying of blocks within
20954 2017-07-20 Richard Biener <rguenther@suse.de>
20956 * gimple.h (gimple_phi_result): Add gphi * overload.
20957 (gimple_phi_result_ptr): Likewise.
20958 (gimple_phi_arg): Likewise. Adjust index assert to only
20959 allow actual argument accesses rather than all slots available
20961 (gimple_phi_arg_def): Add gphi * overload.
20962 * tree-phinodes.c (make_phi_node): Initialize only actual
20964 (resize_phi_node): Clear memory not covered by old node,
20965 do not initialize excess argument slots.
20966 (reserve_phi_args_for_new_edge): Initialize new argument slot
20969 2017-07-20 Bin Cheng <bin.cheng@arm.com>
20971 PR tree-optimization/81388
20973 2016-07-21 Bin Cheng <bin.cheng@arm.com>
20975 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
20976 by removing computation of may_be_zero.
20978 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20979 Tom de Vries <tom@codesourcery.com>
20981 PR middle-end/81030
20982 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
20983 when gimple level profile disagrees with what RTL expander did.
20985 2017-07-20 Richard Biener <rguenther@suse.de>
20987 PR tree-optimization/61171
20988 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
20989 (vect_analyze_stmt): Add slp instance parameter.
20990 (vectorizable_reduction): Likewise.
20991 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
20992 (vect_is_simple_reduction): Deal with chains not detected
20993 as SLP reduction chain, specifically not properly associated
20994 chains containing a mix of plus/minus.
20995 (get_reduction_op): Remove.
20996 (get_initial_defs_for_reduction): Simplify, pass in whether
20997 this is a reduction chain, pass in the SLP node for the PHIs.
20998 (vect_create_epilog_for_reduction): Get the SLP instance as
21000 (vectorizable_reduction): Get the SLP instance as arg.
21001 During analysis remember the SLP node with the PHIs in the
21002 instance. Simplify getting at the vectorized reduction PHIs.
21003 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
21004 through SLP instance.
21005 (vect_slp_analyze_operations): Likewise.
21006 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
21007 (vect_transform_stmt): Likewise.
21009 2017-07-20 Tom de Vries <tom@codesourcery.com>
21011 PR tree-optimization/81489
21012 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
21013 read of phi arg location to before loop that modifies phi.
21015 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21017 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
21020 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
21022 PR middle-end/81331
21023 * except.c (execute): Fix ordering issue.
21025 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
21027 PR rtl-optimization/81423
21028 * combine.c (make_compound_operation_int): Don't try to optimize
21029 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
21031 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
21033 PR rtl-optimization/81423
21034 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
21035 with a constant that is -1 in the truncated to mode.
21037 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
21039 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
21040 (determine_unlikely_bbs): ... here.
21041 * predict.h (propagate_unlikely_bbs_forward): Declare.
21042 * cfgexpand.c (pass_expand::execute): Use it.
21043 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
21045 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
21046 propagate_unlikely_bbs_forward.
21048 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
21050 PR middle-end/81331
21051 * except.c (maybe_add_nop_after_section_switch): New function.
21054 2017-07-19 Tom de Vries <tom@codesourcery.com>
21056 * gimple.h (gimple_phi_set_arg): Make assert more strict.
21058 2017-07-19 Tom de Vries <tom@codesourcery.com>
21060 * gimple.h (gimple_phi_arg): Make assert more strict.
21062 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
21064 * config.gcc (powerpc*-*-*): Add mmintrin.h.
21065 * config/rs6000/mmintrin.h: New file.
21066 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
21068 2017-07-19 Jakub Jelinek <jakub@redhat.com>
21070 PR tree-optimization/81346
21071 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
21073 2017-07-19 Tom de Vries <tom@codesourcery.com>
21075 * config/nvptx/nvptx.md (VECIM): Add V2DI.
21077 2017-07-19 Tom de Vries <tom@codesourcery.com>
21079 * config/nvptx/nvptx-modes.def: Add V2DImode.
21080 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
21081 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
21082 (nvptx_output_mov_insn): Handle lack of mov.b128.
21083 (nvptx_print_operand): Handle 'H' and 'L' codes.
21084 (nvptx_vector_mode_supported): Allow V2DImode.
21085 (nvptx_preferred_simd_mode): New function.
21086 (nvptx_data_alignment): New function.
21087 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
21088 nvptx_preferred_simd_mode.
21089 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
21091 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
21093 2017-07-19 Tom de Vries <tom@codesourcery.com>
21095 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
21096 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
21097 (nvptx_vector_mode_supported): New function. Allow V2SImode.
21098 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
21099 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
21100 (mov<VECIM>_insn): New define_insn.
21101 (define_expand "mov<VECIM>): New define_expand.
21103 2017-07-19 Tom de Vries <tom@codesourcery.com>
21105 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
21107 2017-07-19 Jakub Jelinek <jakub@redhat.com>
21109 PR tree-optimization/81346
21110 * fold-const.h (fold_div_compare, range_check_type): Declare.
21111 * fold-const.c (range_check_type): New function.
21112 (build_range_check): Use range_check_type.
21113 (fold_div_compare): No longer static, rewritten into
21114 a match.pd helper function.
21115 (fold_comparison): Don't call fold_div_compare here.
21116 * match.pd (X / C1 op C2): New optimization using fold_div_compare
21117 as helper function.
21119 2017-07-19 Nathan Sidwell <nathan@acm.org>
21121 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
21122 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
21123 * tree.c (find_decls_types_r, verify_type): Use
21124 TYPE_{MIN,MAX}_VALUE_RAW.
21125 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
21126 (hash_tree): Likewise.
21127 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
21129 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
21132 2017-07-18 Tom de Vries <tom@codesourcery.com>
21134 PR middle-end/81464
21135 * omp-expand.c (expand_omp_for_static_chunk): Handle
21136 equal-argument loop exit phi.
21138 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
21141 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
21142 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
21143 operand 2 predicate.
21144 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
21145 operand 2 predicate.
21146 (ror,rol -> rorx splitters): Use const_int_operand as
21147 operand 2 predicate.
21149 2017-06-18 Richard Biener <rguenther@suse.de>
21151 PR tree-optimization/81410
21152 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
21153 the gap in the ! slp_perm SLP case after each group.
21155 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
21157 PR middle-end/81463
21158 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
21161 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
21163 PR middle-end/81462
21164 * predict.c (set_even_probabilities): Cleanup; do not affect
21165 probabilities that are already known.
21166 (combine_predictions_for_bb): Call even when count is set.
21168 2017-07-18 Nathan Sidwell <nathan@acm.org>
21170 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
21173 2017-07-18 Bin Cheng <bin.cheng@arm.com>
21176 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
21177 optimization for loop niter analysis.
21179 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
21182 * config/avr/avr.c (avr_optimize_casesi): Don't use
21183 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
21185 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
21187 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
21188 body_cost_vec from _vect_peel_extended_info.
21189 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
21190 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
21193 2017-07-18 Bin Cheng <bin.cheng@arm.com>
21195 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
21197 2017-07-18 Richard Biener <rguenther@suse.de>
21199 PR tree-optimization/80620
21200 PR tree-optimization/81403
21201 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
21202 info when re-using a VN table entry.
21204 2017-07-18 Richard Biener <rguenther@suse.de>
21206 PR tree-optimization/81418
21207 * tree-vect-loop.c (vectorizable_reduction): Properly compute
21208 vectype_in. Verify that with lane-reducing reduction operations
21209 we have a single def-use cycle.
21211 2017-07-17 Carl Love <cel@us.ibm.com>
21213 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
21215 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21216 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21217 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21218 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21219 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21220 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21221 VMULOSW): New enum "unspec" values.
21222 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
21223 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
21224 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21225 altivec_vmulosw): New patterns.
21226 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21227 VMULOSW): Add definitions.
21229 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
21231 * config/alpha/alpha.c: Include predict.h.
21233 2017-07-17 Yury Gribov <tetra2005@gmail.com>
21235 * tree-vrp.c (compare_assert_loc): Fix comparison function
21236 to return predictable results.
21238 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21240 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
21242 (subdi3): Likewise.
21243 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
21244 * doc/invoke.texi (mexpand-adddi): Update text.
21246 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21248 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
21249 that also clobbers the CC register. The old expand code is moved
21251 (*arc_clzsi2): ... here.
21252 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
21253 the CC register. The old expand code is moved to ...
21254 (arc_ctzsi2): ... here.
21256 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21258 * config/arc/arc.opt (mindexed-loads): Use initial value
21259 TARGET_INDEXED_LOADS_DEFAULT.
21260 (mauto-modify-reg): Use initial value
21261 TARGET_AUTO_MODIFY_REG_DEFAULT.
21262 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
21263 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
21264 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
21265 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
21267 2017-07-17 Martin Liska <mliska@suse.cz>
21270 * opts.c (finish_options): Do not allow -fgnu-tm
21271 w/ -fsanitize={kernel-,}address. Say sorry.
21273 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21276 * tree-loop-distribution.c (classify_partition): Only assert on
21277 numer of iterations.
21278 (merge_dep_scc_partitions): Delete prameter. Update function call.
21279 (distribute_loop): Remove code handling loop with unknown niters.
21280 (pass_loop_distribution::execute): Skip loop with unknown niters.
21282 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21285 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
21286 function sort_partitions_by_post_order.
21288 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21290 PR tree-optimization/81374
21291 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
21292 the max index of basic blocks, rather than number of basic blocks.
21294 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21296 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
21298 (arc_legitimate_pic_operand_p): Likewise.
21299 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
21301 (arc_needs_pcl_p): Likewise.
21302 (arc_legitimate_pc_offset_p): Likewise.
21303 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
21304 function is also used in constrains.md.
21305 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
21306 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
21307 PLUS. Only return true/false in known cases, otherwise assert.
21308 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
21309 is already called in arc_legitimate_constant_p.
21310 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
21312 (LEGITIMATE_PIC_OPERAND_P): Use
21313 arc_raw_symbolic_reference_mentioned_p function.
21314 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
21319 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21320 Andrew Burgess <andrew.burgess@embecosm.com>
21322 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
21323 (arc_return_address_register): New function.
21324 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
21325 (arc_handle_fndecl_attribute): Add naked attribute.
21326 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21327 (TARGET_WARN_FUNC_RETURN): Likewise.
21328 (arc_allocate_stack_slots_for_args): New function.
21329 (arc_warn_func_return): Likewise.
21330 (machine_function): Change type fn_type.
21331 (arc_compute_function_type): Consider new naked function type,
21332 change function return type.
21333 (arc_must_save_register): Adapt to handle new
21334 arc_compute_function_type's return type.
21335 (arc_expand_prologue): Likewise.
21336 (arc_expand_epilogue): Likewise.
21337 (arc_return_address_regs): Delete.
21338 (arc_return_address_register): New function.
21339 (arc_epilogue_uses): Use above function.
21340 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
21341 (arc_function_type): Change encoding, add naked type.
21342 (ARC_INTERRUPT_P): Change to handle the new encoding.
21343 (ARC_FAST_INTERRUPT_P): Likewise.
21344 (ARC_NORMAL_P): Define.
21345 (ARC_NAKED_P): Likewise.
21346 (arc_compute_function_type): Delete prototype.
21347 * config/arc/arc.md (in_ret_delay_slot): Use
21348 arc_return_address_register function.
21349 (simple_return): Likewise.
21350 (p_return_i): Likewise.
21352 2017-07-17 Jakub Jelinek <jakub@redhat.com>
21354 PR tree-optimization/81428
21355 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
21356 can't be built for those types.
21358 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
21360 Remove stuff dead since r239246.
21362 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
21363 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
21364 (avr_inform_devices): Remove dead stuff.
21366 2017-07-17 Tamar Christina <tamar.christina@arm.com>
21368 * config/arm/arm_neon.h: Fix softp typo.
21370 2017-07-17 Jakub Jelinek <jakub@redhat.com>
21372 PR tree-optimization/81365
21373 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
21374 aggregate moves onto bb predecessor edges, make sure there are no
21375 loads that could alias the lhs in between the start of bb and the
21378 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
21381 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
21382 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
21383 [LSHIFTRT, outer_code = TRUNCATE]: Same.
21385 2017-07-17 Jakub Jelinek <jakub@redhat.com>
21387 PR tree-optimization/81396
21388 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
21389 (init_symbolic_number): Initialize it to 1.
21390 (perform_symbolic_merge): Add n_ops from both operands into the new
21392 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
21393 without base_addr as useless if they need more than one operation.
21394 (bswap_replace): Handle !bswap case for NULL base_addr.
21396 2017-07-17 Tom de Vries <tom@codesourcery.com>
21399 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
21402 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
21404 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
21405 conditional builtin define __FIX_LEON3FT_B2BST.
21407 2017-07-17 Daniel Cederman <cederman@gaisler.com>
21409 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
21410 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
21413 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
21415 PR rtl-optimization/81424
21416 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
21417 to remove potential trapping from operands if -fnon-call-exceptions.
21419 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21421 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
21422 profile_proability for scalling.
21423 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
21425 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21427 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
21429 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21431 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
21432 fixpoint arithmetics.
21434 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21436 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
21437 fixpoint arithmetics.
21439 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21441 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
21442 fixpoint arithmetics.
21444 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21446 * profile-count.h (profile_probability::from_reg_br_prob_note,
21447 profile_probability::to_reg_br_prob_note): New functions.
21448 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
21449 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
21450 * predict.c (probability_reliable_p): Update.
21451 (edge_probability_reliable_p): Update.
21452 (br_prob_note_reliable_p): Update.
21453 (invert_br_probabilities): Update.
21454 (add_reg_br_prob_note): New function.
21455 (combine_predictions_for_insn): Update.
21456 * asan.c (asan_clear_shadow): Update.
21457 * cfgbuild.c (compute_outgoing_frequencies): Update.
21458 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21459 (update_br_prob_note): Update.
21460 (rtl_verify_edges): Update.
21461 (purge_dead_edges): Update.
21462 (fixup_reorder_chain): Update.
21463 * emit-rtl.c (try_split): Update.
21464 * ifcvt.c (cond_exec_process_insns): Update.
21465 (cond_exec_process_if_block): Update.
21466 (dead_or_predicable): Update.
21467 * internal-fn.c (expand_addsub_overflow): Update.
21468 (expand_neg_overflow): Update.
21469 (expand_mul_overflow): Update.
21470 * loop-doloop.c (doloop_modify): Update.
21471 * loop-unroll.c (compare_and_jump_seq): Update.
21472 * optabs.c (emit_cmp_and_jump_insn_1): Update.
21473 * predict.h: Update.
21474 * reorg.c (mostly_true_jump): Update.
21476 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
21477 * config/alpha/alpha.c (emit_unlikely_jump): Update.
21478 * config/arc/arc.c: (emit_unlikely_jump): Update.
21479 * config/arm/arm.c: (emit_unlikely_jump): Update.
21480 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
21481 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
21482 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
21483 (ix86_print_operand): Update.
21484 (ix86_split_fp_branch): Update.
21485 (predict_jump): Update.
21486 * config/ia64/ia64.c (ia64_print_operand): Update.
21487 * config/mmix/mmix.c (mmix_print_operand): Update.
21488 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
21489 (rs6000_expand_split_stack_prologue): Update.
21490 * config/rs6000/rs6000.c: Update.
21491 * config/s390/s390.c (s390_expand_vec_strlen): Update.
21492 (s390_expand_vec_movstr): Update.
21493 (s390_expand_cs_tdsi): Update.
21494 (s390_expand_split_stack_prologue): Update.
21495 * config/sh/sh.c (sh_print_operand): Update.
21496 (expand_cbranchsi4): Update.
21497 (expand_cbranchdi4): Update.
21498 * config/sparc/sparc.c (output_v9branch): Update.
21499 * config/spu/spu.c (get_branch_target): Update.
21500 (ea_load_store_inline): Update.
21501 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
21502 * config/tilepro/tilepro.c: Update.
21504 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
21506 * gimplify.c (mostly_copy_tree_r): Revert latest change.
21507 (gimplify_save_expr): Likewise.
21509 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21511 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
21513 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21515 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
21517 * timevar.def (TV_IPA_FNSUMMARY): Define.
21519 2017-07-16 Daniel Cederman <cederman@gaisler.com>
21521 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
21522 to back store errata sensitive sequence from being generated.
21523 (sqrtdf2_fix): Likewise.
21525 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21527 * tree-ssa-threadupdate.c (compute_path_counts,
21528 update_joiner_offpath_counts): Use profile_probability.
21530 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
21533 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21535 * config/arm/arm-c.c (arm_cpu_builtins): Define
21536 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
21538 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
21540 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21541 array entries to represent __ieee128 versions of the
21542 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
21543 scalar_extract_sig, and scalar_insert_exp built-in functions.
21544 (altivec_resolve_overloaded_builtin): Add special case handling
21545 for the __builtin_scalar_insert_exp function, as represented by
21546 the P9V_BUILTIN_VEC_VSIEDP constant.
21547 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
21548 exponent support for __ieee128 argument.
21549 (VSESQP): Add scalar extract signature support for __ieee128
21551 (VSTDCNQP): Add scalar test negative support for __ieee128
21553 (VSIEQP): Add scalar insert exponent support for __int128 argument
21554 with __ieee128 result.
21555 (VSIEQPF): Add scalar insert exponent support for __ieee128
21556 argument with __ieee128 result.
21557 (VSTDCQP): Add scalar test data class support for __ieee128
21559 (VSTDCNQP): Add overload support for scalar test negative with
21560 __ieee128 argument.
21561 (VSTDCQP): Add overload support for scalar test data class
21562 __ieee128 argument.
21563 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
21564 UNSPEC_VSX_SXSIGDP.
21565 (UNSPEC_VSX_SIEXPQP): New constant.
21566 (xsxexpqp): New insn for VSX scalar extract exponent quad
21568 (xsxsigqp): New insn for VSX scalar extract significand quad
21570 (xsiexpqpf): New insn for VSX scalar insert exponent quad
21571 precision with floating point argument.
21572 (xststdcqp): New expand for VSX scalar test data class quad
21574 (xststdcnegqp): New expand for VSX scalar test negative quad
21576 (xststdcqp): New insn to match expansions for VSX scalar test data
21577 class quad precision and VSX scalar test negative quad precision.
21578 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
21579 special case operand checking to enforce that second operand of
21580 VSX scalar test data class with quad precision argument is a 7-bit
21582 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21583 prototypes and descriptions of __ieee128 versions of
21584 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
21585 scalar_test_data_class, and scalar_test_neg built-in functions.
21587 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21589 PR tree-optimization/81162
21590 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
21591 replace a negate with an add.
21593 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
21595 * doc/invoke.texi (arm/-mcpu): Document +crypto.
21597 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21599 * config/arm/arm-c.c (arm_cpu_builtins): Define
21600 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
21602 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21604 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
21605 (armv8-r): Set ARM Cortex-R52 as default CPU.
21606 * config/arm/arm-tables.opt: Regenerate.
21607 * config/arm/arm-tune.md: Regenerate.
21608 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
21610 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
21611 extension for -mcpu=cortex-r52.
21613 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21615 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
21616 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
21617 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
21618 (fp-armv8): Define it as FP_ARMv8 only.
21619 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
21620 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
21622 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
21623 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
21624 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
21625 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
21626 than TARGET_FPU_ARMV8.
21627 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
21628 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
21629 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
21631 * config/arm/neon.md (neon_vrint): Likewise.
21632 (neon_vcvt): Likewise.
21633 (neon_<fmaxmin_op><mode>): Likewise.
21634 (<fmaxmin><mode>3): Likewise.
21635 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
21636 * config/arm/predicates.md (arm_cond_move_operator): Check against
21637 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
21639 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
21641 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
21642 to top of function.
21644 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21646 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
21647 loop in comment with memset.
21649 2017-07-14 Martin Liska <mliska@suse.cz>
21651 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
21652 * dwarf2out.c (is_java): Remove the function.
21653 (output_pubname): Remove usage of the function.
21654 (lower_bound_default): Remove usage of DW_LANG_Java.
21655 (gen_compile_unit_die): Likewise.
21656 * gcc.c: Remove compiler defaults for .java and .zip files.
21657 * gimple-expr.c (remove_suffix): Change as there's no longer
21658 extension than 4-letter one.
21659 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
21660 (gimplify_save_expr): Likewise.
21661 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
21662 as it's possible even for other languages than Java.
21663 * langhooks.h (struct lang_hooks): Remove Java from a comment.
21664 * lto-opts.c (lto_write_options): Remove reference to Java.
21665 * opts.c (strip_off_ending): Update file extension handling.
21666 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
21667 * tree-eh.c (lower_resx): Likewise.
21668 * tree.c (free_lang_data_in_type): Remove dead code.
21669 (find_decls_types_r): Likewise.
21670 (build_common_builtin_nodes): Remove Java from a comment.
21671 (verify_type): Remove dead code.
21672 * varasm.c (assemble_external): Remove Java from a comment.
21674 2017-07-14 Martin Liska <mliska@suse.cz>
21676 * opts.c (finish_options): Add quotes.
21677 (common_handle_option): Likewise.
21679 2017-07-14 Martin Liska <mliska@suse.cz>
21681 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
21682 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
21683 Remove N_SO_PASCAL.
21684 * dwarf2out.c (lower_bound_default): Do not handle
21686 (gen_compile_unit_die): Likewise.
21687 * gcc.c: Remove default extension binding for GNU Pascal.
21688 * stmt.c: Remove Pascal language from a comment.
21689 * xcoffout.c: Likewise.
21691 2017-07-13 David Malcolm <dmalcolm@redhat.com>
21694 * diagnostic-show-locus.c (fixit_cmp): New function.
21695 (layout::layout): Sort m_fixit_hints.
21696 (column_range::column_range): Assert that the values are valid.
21697 (struct char_span): New struct.
21698 (correction::overwrite): New method.
21699 (struct source_line): New struct.
21700 (line_corrections::add_hint): Add assertions. Reimplement memcpy
21701 calls in terms of classes source_line and char_span, and
21702 correction::overwrite.
21703 (selftest::test_overlapped_fixit_printing_2): New function.
21704 (selftest::diagnostic_show_locus_c_tests): Call it.
21706 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
21708 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
21709 early if there is no lhs.
21711 2017-07-13 Martin Liska <mliska@suse.cz>
21713 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
21714 (gen_reference_type_die): Likewise.
21715 * stor-layout.c: Remove Pascal-related comment.
21717 2017-07-13 Martin Liska <mliska@suse.cz>
21719 * opts.c (finish_options): Add quotes to error messages.
21720 (parse_sanitizer_options): Likewise.
21722 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
21724 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
21726 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
21728 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
21730 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
21732 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
21734 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
21736 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
21739 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
21740 provides the hardware capability bits, define the macro
21741 __BUILTIN_CPU_SUPPORTS__.
21742 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
21743 if GLIBC does not provide the hardware capability bits. Add a
21744 gcc_unreachable call if the built-in cpu function is neither
21745 __builtin_cpu_is nor __builtin_cpu_supports.
21746 (rs6000_get_function_versions_dispatcher): Change the warning
21747 that an old GLIBC is used which does not export the capability
21748 bits to be an error.
21749 * doc/extend.texi (target_clones attribute): Document the
21750 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
21751 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
21752 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
21753 the macros defined by GCC if the newer GLIBC is available.
21755 2017-07-12 Jeff Law <law@redhat.com>
21757 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
21758 remaining includes slightly.
21759 * config/riscv/riscv-builtins.c: Include profile-count.h.
21761 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21764 * config/avr/avr.c (avr_set_current_function): In diagnostic
21765 messages: Quote keywords and (parts of) identifiers.
21766 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
21769 2017-07-12 Carl Love <cel@us.ibm.com>
21771 * config/rs6000/rs6000-c.c: Add support for built-in functions
21772 vector bool char vec_revb (vector bool char);
21773 vector bool short vec_revb (vector short char);
21774 vector bool int vec_revb (vector bool int);
21775 vector bool long long vec_revb (vector bool long long);
21776 * doc/extend.texi: Update the built-in documentation file for the
21777 new built-in functions.
21779 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21781 * config/s390/s390.md: Remove movcc splitter.
21783 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21785 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
21786 load/store on condition.
21788 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21791 * config/avr/avr.c (avr_encode_section_info)
21792 [progmem && !TREE_READONLY]: Error if progmem object needs
21795 2017-07-11 Michael Collison <michael.collison@arm.com>
21797 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
21800 2017-07-11 Carl Love <cel@us.ibm.com>
21802 * config/rs6000/rs6000-c.c: Add support for builtins
21803 vector unsigned int vec_parity_lsbb (vector signed int);
21804 vector unsigned int vec_parity_lsbb (vector unsigned int);
21805 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
21806 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
21807 vector unsigned long long vec_parity_lsbb (vector signed long long);
21808 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
21809 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
21810 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
21811 * doc/extend.texi: Update the built-in documentation file for the
21812 new built-in functions.
21814 2017-07-11 David Malcolm <dmalcolm@redhat.com>
21816 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
21817 (layout::m_primary_loc): New field.
21818 (layout::layout): Initialize new field. Move location filtering
21819 logic from here to...
21820 (layout::maybe_add_location_range): ...this new method. Add
21821 support for filtering to just the lines already specified by other
21823 (layout::will_show_line_p): New method.
21824 (gcc_rich_location::add_location_if_nearby): New method.
21825 (selftest::test_add_location_if_nearby): New test function.
21826 (selftest::diagnostic_show_locus_c_tests): Call it.
21827 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
21830 2017-07-11 Tom de Vries <tom@codesourcery.com>
21832 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
21833 (bb_first_real_insn): New function.
21834 (nvptx_single): Add extra initialization of broadcasted condition
21837 2017-07-11 Nathan Sidwell <nathan@acm.org>
21839 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
21841 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
21843 * doc/extend.texi (AVR Function Attributes): Remove weblink to
21844 Binutils doc as TEXI will mess them up.
21845 * doc/invoke.texi (AVR Options): Same here.
21847 2017-07-11 Daniel Cederman <cederman@gaisler.com>
21849 * config/sparc/sparc.opt (mfix-ut700): New option.
21850 (mfix-gr712rc): Likewise.
21851 (sparc_fix_b2bst): New variable.
21852 * doc/invoke.texi (SPARC options): Document them.
21853 (ARM options): Fix warnings.
21854 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
21855 instructions to prevent sequences that can trigger the store-store
21856 errata for certain LEON3FT processors.
21857 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
21858 (sparc_option_override): Set sparc_fix_b2bst appropriately.
21859 * config/sparc/sparc.md (fix_b2bst): New attribute.
21860 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
21862 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
21865 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
21867 (*rsqrtsf2_sse): Ditto.
21869 (div<mode>3): Macroize insn from divdf3 and divsf3
21870 using MODEF mode iterator.
21872 2017-07-10 Martin Sebor <msebor@redhat.com>
21874 PR tree-optimization/80397
21875 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
21876 instead of testing for equality to INTEGER_TYPE.
21878 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
21880 * config.gcc: Remove uclibc from arc target spec.
21882 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
21884 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
21886 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21889 * lto-wrapper.c (remove_option): New function.
21890 (merge_and_complain): Merge PIC/PIE options more realistically.
21892 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21894 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
21898 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
21899 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
21900 * config.in: Regenerate.
21901 * configure: Regenerate.
21902 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
21903 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
21904 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
21905 (TARGET_GASISR_PROLOGUES): ...target mask.
21906 * common/config/avr/avr-common.c
21907 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
21908 Set -mgas-isr-prologues.
21909 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
21910 INSERT_PASS_BEFORE for it.
21911 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
21912 * config/avr/avr.c (avr_option_override)
21913 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
21914 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
21915 (avr_attribute_table) <no_gccisr>: Add new function attribute.
21916 (avr_set_current_function) <is_no_gccisr>: Init machine field.
21917 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
21919 (make_avr_pass_pre_proep): New function.
21920 (emit_push_sfr) <treg>: Add argument to function and use it
21921 instead of TMP_REG.
21922 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
21923 and set machine->gasisr.yes.
21924 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
21925 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
21926 __gcc_isr.n_pushed to .L__stack_usage.
21927 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
21928 (avr_asm_final_postscan_insn): ...this new static function.
21929 * config/avr/avr.h (machine_function)
21930 <is_no_gccisr, use_L__stack_usage>: New fields.
21931 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
21932 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
21933 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
21934 (gasisr, *gasisr): New expander and insn.
21935 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21936 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
21937 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
21939 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
21941 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
21944 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21946 Move jump-tables out of .text again.
21949 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
21950 (ASM_OUTPUT_ADDR_VEC): New function.
21951 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
21952 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
21953 INSN_ADDRESSes as asm comment.
21954 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
21955 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
21956 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
21957 * config/avr/avr.md (*tablejump): Adjust comment.
21958 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
21959 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
21961 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
21962 (avr_output_addr_vec): New proto.
21963 (avr_log_t) <insn_addresses>: New field.
21965 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
21968 * config/i386/i386.c (ix86_function_arg_advance): Set
21969 outgoing_args_on_stack to true if there are outgoing arguments
21971 (ix86_function_arg): Likewise.
21972 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
21973 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
21974 * config/i386/i386.h (machine_function): Add
21975 outgoing_args_on_stack.
21977 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
21979 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
21980 supporting pthreds.
21981 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
21983 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
21985 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
21986 (REAL_H): Remove $(MACHMODE_H).
21987 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
21989 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
21990 $(MACHMODE_H) and double-int.h.
21991 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
21993 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
21996 2017-07-07 Andrew Pinski <apinski@cavium.com>
21998 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
21999 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
22001 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
22003 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
22004 Add warning if GCC was not configured to link against a GLIBC that
22005 exports the hardware capability bits.
22006 (make_resolver_func): Make resolver function private and not a
22007 COMDAT function. Create the name with clone_function_name instead
22008 of make_unique_name.
22011 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
22012 correct operand in doing the split.
22014 2017-07-07 Carl Love <cel@us.ibm.com>
22016 * config/rs6000/rs6000-c: Add support for built-in function
22017 vector unsigned short vec_pack_to_short_fp32 (vector float,
22019 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
22020 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
22021 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
22022 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
22023 (convert_4f32_8i16): Add define_expand.
22024 * doc/extend.texi: Update the built-in documentation file for the
22025 new built-in function.
22027 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22029 * config/sparc/m8.md: New file.
22030 * config/sparc/sparc.md: Include m8.md.
22032 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22034 * config/sparc/sparc.opt: New option -mvis4b.
22035 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
22036 (sparc_option_override): Handle VIS4B.
22037 (enum sparc_builtins): Define
22038 SPARC_BUILTIN_DICTUNPACK{8,16,32},
22039 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
22040 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
22041 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
22042 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
22043 (check_constant_argument): New function.
22044 (sparc_vis_init_builtins): Define builtins
22045 __builtin_vis_dictunpack{8,16,32},
22046 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
22047 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
22048 __builtin_vis_fpcmpde{8,16,32}shl and
22049 __builtin_vis_fpcmpur{8,16,32}shl.
22050 (sparc_expand_builtin): Check that the constant operands to
22051 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
22052 constant and in range.
22053 * config/sparc/sparc-c.c (sparc_target_macros): Handle
22055 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
22056 (SPARC_IMM5_P): Likewise.
22057 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
22058 (enabled): Handle vis4b.
22059 (UNSPEC_DICTUNPACK): New unspec.
22060 (UNSPEC_FPCMPSHL): Likewise.
22061 (UNSPEC_FPUCMPSHL): Likewise.
22062 (UNSPEC_FPCMPDESHL): Likewise.
22063 (UNSPEC_FPCMPURSHL): Likewise.
22064 (cpu_feature): New CPU feature `vis4b'.
22065 (dictunpack{8,16,32}): New insns.
22066 (FPCSMODE): New mode iterator.
22067 (fpcscond): New code iterator.
22068 (fpcsucond): Likewise.
22069 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
22070 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
22071 (fpcmpde{8,16,32}{si,di}shl): Likewise.
22072 (fpcmpur{8,16,32}{si,di}shl): Likewise.
22073 * config/sparc/constraints.md: Define constraints `q' for unsigned
22074 2-bit integer constants and `t' for unsigned 5-bit integer
22076 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
22078 (imm5_operand_dictunpack16): Likewise.
22079 (imm5_operand_dictunpack32): Likewise.
22080 (imm2_operand): Likewise.
22081 * doc/invoke.texi (SPARC Options): Document -mvis4b.
22082 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
22083 ditunpack* and fpcmp*shl builtins.
22085 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22087 * config.gcc: Handle m8 in --with-{cpu,tune} options.
22088 * config.in: Add HAVE_AS_SPARC6 define.
22089 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
22091 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
22093 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
22094 (CPP_CPU_SPEC): Handle m8.
22095 (ASM_CPU_SPEC): Likewise.
22096 * config/sparc/sparc-opts.h (enum processor_type): Add
22098 * config/sparc/sparc.c (m8_costs): New struct.
22099 (sparc_option_override): Handle TARGET_CPU_m8.
22100 (sparc32_initialize_trampoline): Likewise.
22101 (sparc64_initialize_trampoline): Likewise.
22102 (sparc_issue_rate): Likewise.
22103 (sparc_register_move_cost): Likewise.
22104 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
22105 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
22106 (ASM_CPU64_DEFAULT_SPEC): Likewise.
22107 (CPP_CPU_SPEC): Handle M8.
22108 (ASM_CPU_SPEC): Likewise.
22109 (AS_M8_FLAG): Define.
22110 * config/sparc/sparc.md: Add m8 to the cpu attribute.
22111 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
22112 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
22114 * configure: Regenerate.
22115 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
22118 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22120 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
22122 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
22123 ("*movdi_insn_sp32"): Do not set v3pipe.
22124 ("*movsi_insn"): Likewise.
22125 ("*movdi_insn_sp64"): Likewise.
22126 ("*movsf_insn"): Likewise.
22127 ("*movdf_insn_sp32"): Likewise.
22128 ("*movdf_insn_sp64"): Likewise.
22129 ("*zero_extendsidi2_insn_sp64"): Likewise.
22130 ("*sign_extendsidi2_insn"): Likewise.
22131 ("*mov<VM32:mode>_insn"): Likewise.
22132 ("*mov<VM64:mode>_insn_sp64"): Likewise.
22133 ("*mov<VM64:mode>_insn_sp32"): Likewise.
22134 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
22135 ("<vlop:code><VL:mode>3"): Likewise.
22136 ("*not_<vlop:code><VL:mode>3"): Likewise.
22137 ("*nand<VL:mode>_vis"): Likewise.
22138 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
22139 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
22140 ("one_cmpl<VL:mode>2"): Likewise.
22141 ("faligndata<VM64:mode>_vis"): Likewise.
22142 ("alignaddrsi_vis"): Likewise.
22143 ("alignaddrdi_vis"): Likweise.
22144 ("alignaddrlsi_vis"): Likewise.
22145 ("alignaddrldi_vis"): Likewise.
22146 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
22147 ("bmaskdi_vis"): Likewise.
22148 ("bmasksi_vis"): Likewise.
22149 ("bshuffle<VM64:mode>_vis"): Likewise.
22150 ("cmask8<P:mode>_vis"): Likewise.
22151 ("cmask16<P:mode>_vis"): Likewise.
22152 ("cmask32<P:mode>_vis"): Likewise.
22153 ("pdistn<P:mode>_vis"): Likewise.
22154 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
22156 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22158 * config/sparc/sparc.md ("subtype"): New insn attribute.
22159 ("*wrgsr_sp64"): Set insn subtype.
22160 ("*rdgsr_sp64"): Likewise.
22161 ("alignaddrsi_vis"): Likewise.
22162 ("alignaddrdi_vis"): Likewise.
22163 ("alignaddrlsi_vis"): Likewise.
22164 ("alignaddrldi_vis"): Likewise.
22165 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
22166 ("fexpand_vis"): Likewise.
22167 ("fpmerge_vis"): Likewise.
22168 ("faligndata<VM64:mode>_vis"): Likewise.
22169 ("bshuffle<VM64:mode>_vis"): Likewise.
22170 ("cmask8<P:mode>_vis"): Likewise.
22171 ("cmask16<P:mode>_vis"): Likewise.
22172 ("cmask32<P:mode>_vis"): Likewise.
22173 ("fchksm16_vis"): Likewise.
22174 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
22175 ("fmean16_vis"): Likewise.
22176 ("fp<plusminus_insn>64_vis"): Likewise.
22177 ("<plusminus_insn>v8qi3"): Likewise.
22178 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
22179 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
22180 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
22181 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
22182 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
22183 ("*movqi_insn"): Likewise.
22184 ("*movhi_insn"): Likewise.
22185 ("*movsi_insn"): Likewise.
22186 ("movsi_pic_gotdata_op"): Likewise.
22187 ("*movdi_insn_sp32"): Likewise.
22188 ("*movdi_insn_sp64"): Likewise.
22189 ("movdi_pic_gotdata_op"): Likewise.
22190 ("*movsf_insn"): Likewise.
22191 ("*movdf_insn_sp32"): Likewise.
22192 ("*movdf_insn_sp64"): Likewise.
22193 ("*zero_extendhisi2_insn"): Likewise.
22194 ("*zero_extendqihi2_insn"): Likewise.
22195 ("*zero_extendqisi2_insn"): Likewise.
22196 ("*zero_extendqidi2_insn"): Likewise.
22197 ("*zero_extendhidi2_insn"): Likewise.
22198 ("*zero_extendsidi2_insn_sp64"): Likewise.
22199 ("ldfsr"): Likewise.
22200 ("prefetch_64"): Likewise.
22201 ("prefetch_32"): Likewise.
22202 ("tie_ld32"): Likewise.
22203 ("tie_ld64"): Likewise.
22204 ("*tldo_ldub_sp32"): Likewise.
22205 ("*tldo_ldub1_sp32"): Likewise.
22206 ("*tldo_ldub2_sp32"): Likewise.
22207 ("*tldo_ldub_sp64"): Likewise.
22208 ("*tldo_ldub1_sp64"): Likewise.
22209 ("*tldo_ldub2_sp64"): Likewise.
22210 ("*tldo_ldub3_sp64"): Likewise.
22211 ("*tldo_lduh_sp32"): Likewise.
22212 ("*tldo_lduh1_sp32"): Likewise.
22213 ("*tldo_lduh_sp64"): Likewise.
22214 ("*tldo_lduh1_sp64"): Likewise.
22215 ("*tldo_lduh2_sp64"): Likewise.
22216 ("*tldo_lduw_sp32"): Likewise.
22217 ("*tldo_lduw_sp64"): Likewise.
22218 ("*tldo_lduw1_sp64"): Likewise.
22219 ("*tldo_ldx_sp64"): Likewise.
22220 ("*mov<VM32:mode>_insn"): Likewise.
22221 ("*mov<VM64:mode>_insn_sp64"): Likewise.
22222 ("*mov<VM64:mode>_insn_sp32"): Likewise.
22224 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22226 * config/sparc/sparc.md ("type"): New insn type viscmp.
22227 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
22229 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
22230 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
22231 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
22232 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
22234 ("n7_vis_logical_11cycle"): Likewise.
22235 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
22236 * config/sparc/niagara2.md ("niag3_vis": Likewise.
22237 * config/sparc/niagara.md ("niag_vis"): Likewise.
22238 * config/sparc/ultra3.md ("us3_fga"): Likewise.
22239 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
22241 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22243 * config/sparc/sparc.md: New instruction type `bmask'.
22244 (bmaskdi_vis): Use the `bmask' type.
22245 (bmasksi_vis): Likewise.
22246 * config/sparc/ultra3.md (us3_array): Likewise.
22247 * config/sparc/niagara7.md (n7_array): Likewise.
22248 * config/sparc/niagara4.md (n4_array): Likewise.
22249 * config/sparc/niagara2.md (niag2_vis): Likewise.
22250 (niag3_vis): Likewise.
22251 * config/sparc/niagara.md (niag_vis): Likewise.
22253 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22255 * ipa-comdats.c: Remove optimize check from gate.
22256 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
22257 for functions not optimized.
22258 (ipa_fn_summary_read): Skip optimize check.
22259 (ipa_fn_summary_write): Likewise.
22260 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
22262 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
22264 (can_inline_edge_p): Check flag_pcc_struct_return for match.
22265 (check_callers): Give up on caller which is not optimized.
22266 (inline_small_functions): Likewise.
22267 (ipa_inline): Do not give up when not optimizing.
22268 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
22269 away unoptimizes cdtors.
22270 (whole_program_function_and_variable_visibility): Do
22271 ipa_discover_readonly_nonaddressable_vars in LTO mode.
22272 * ipa.c (process_references): Do not check optimize.
22273 (symbol_table::remove_unreachable_nodes): Update optimize check.
22274 (set_writeonly_bit): Update optimize check.
22275 (pass_ipa_cdtor_merge::gate): Do not check optimize.
22276 (pass_ipa_single_use::gate): Remove.
22278 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22280 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
22281 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
22282 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
22283 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
22284 permute_load, permute_store, adjust_extract, adjust_splat,
22285 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
22286 replace_swap_with_copy, dump_swap_insn_table,
22287 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
22288 recombine_lvx_pattern, recombine_stvx_pattern,
22289 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
22290 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
22291 to file rs6000-p8swap.c.
22292 * config/rs6000/rs6000-p8swap.c: New file.
22293 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
22294 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
22295 and rs6000*-*-* targets.
22297 2017-07-06 David Malcolm <dmalcolm@redhat.com>
22299 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
22301 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22303 * lto-wrapper.c (merge_and_complain): Do not merge
22304 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
22305 fsigned_zeros, ftrapping_math, fwrapv.
22306 (append_compiler_options): Do not track these options.
22307 (append_linker_options): Likewie
22309 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22311 * cgraphunit.c (cgraph_node::finalize_function): When
22312 !flag_toplevel_reorde set no_reorder flag.
22313 (varpool_node::finalize_decl): Likewise.
22314 (symbol_table::compile): Drop no toplevel reorder path.
22316 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22318 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
22319 edges; zero probability is not better than uninitialized.
22321 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
22323 * asan.h (asan_sanitize_allocas_p): Declare.
22324 * asan.c (asan_sanitize_allocas_p): New function.
22325 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
22326 (handle_builtin_alloca): Likewise.
22327 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
22328 if !asan_sanitize_allocas_p.
22329 * params.def (asan-instrument-allocas): Add new option.
22330 * params.h (ASAN_PROTECT_ALLOCAS): Define.
22331 * opts.c (common_handle_option): Disable allocas sanitization for
22334 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
22336 * asan.c: Include gimple-fold.h.
22337 (get_last_alloca_addr): New function.
22338 (handle_builtin_stackrestore): Likewise.
22339 (handle_builtin_alloca): Likewise.
22340 (asan_emit_allocas_unpoison): Likewise.
22341 (get_mem_refs_of_builtin_call): Add new parameter, remove const
22342 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
22343 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
22344 (instrument_builtin_call): Pass gimple iterator to
22345 get_mem_refs_of_builtin_call.
22346 (last_alloca_addr): New global.
22347 * asan.h (asan_emit_allocas_unpoison): Declare.
22348 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
22349 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
22350 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
22351 if function calls alloca.
22352 * gimple-fold.c (replace_call_with_value): Remove static keyword.
22353 * gimple-fold.h (replace_call_with_value): Declare.
22354 * internal-fn.c: Include asan.h.
22355 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
22356 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
22358 2017-07-06 David Malcolm <dmalcolm@redhat.com>
22360 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
22361 (C_SELFTEST_FLAGS): New.
22362 (CPP_SELFTEST_FLAGS): New.
22363 (SELFTEST_DEPS): New, from deps of s-selftest.
22364 (C_SELFTEST_DEPS): New, from deps of s-selftest.
22365 (CPP_SELFTEST_DEPS): New.
22366 (selftest): Add dependency on s-selftest-c++.
22367 (s-selftest): Rename to...
22368 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
22369 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
22370 than SELFTEST_FLAGS.
22371 (selftest-gdb): Rename to...
22372 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
22374 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
22375 (selftest-valgrind): Rename to...
22376 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
22378 (selftest-valgrind): Reintroduce as an alias for
22379 selftest-c-valgrind.
22380 (s-selftest-c++): New.
22381 (selftest-c++-gdb): New.
22382 (selftest-c++-valgrind): New.
22384 2017-07-06 Olivier Hainque <hainque@adacore.com>
22386 * gcc.c (process_command): When deciding if undefined variables
22387 should be ignored when processing specs, accept "gcc -v" as well.
22389 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22391 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
22392 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
22394 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22396 * config/arm/arm-cpus.in (armv8-r): Add new entry.
22397 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
22398 * config/arm/arm-tables.opt: Regenerate.
22399 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
22401 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
22403 2017-07-06 Carl Love <cel@us.ibm.com>
22405 * ChangeLog: Clean up from mid air collision
22407 2017-07-06 Carl Love <cel@us.ibm.com>
22409 * config/rs6000/rs6000-c.c: Add support for built-in functions
22410 vector signed int vec_subc (vector signed int, vector signed int);
22411 vector signed __int128 vec_subc (vector signed __int128,
22412 vector signed __int128);
22413 vector unsigned __int128 vec_subc (vector unsigned __int128,
22414 vector unsigned __int128);
22415 vector signed int vec_sube (vector signed int, vector signed int,
22416 vector signed int);
22417 vector unsigned int vec_sube (vector unsigned int,
22418 vector unsigned int,
22419 vector unsigned int);
22420 vector signed __int128 vec_sube (vector signed __int128,
22421 vector signed __int128,
22422 vector signed__int128);
22423 vector unsigned __int128 vec_sube (vector unsigned __int128,
22424 vector unsigned __int128,
22425 vector unsigned __int128);
22426 vector signed int vec_subec (vector signed int, vector signed int,
22427 vector signed int);
22428 vector unsigned int vec_subec (vector unsigned int,
22429 vector unsigned int,
22430 vector unsigned int);
22431 vector signed __int128 vec_subec (vector signed __int128,
22432 vector signed __int128,
22433 vector signed__int128);
22434 vector unsigned __int128 vec_subec (vector unsigned __int128,
22435 vector unsigned __int128,
22436 vector unsigned __int128);
22437 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
22438 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
22439 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
22440 BU_ALTIVEC_OVERLOAD_X definitions.
22441 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
22442 * doc/extend.texi: Update the built-in documentation file for the new
22443 built-in functions.
22445 2017-07-06 David Malcolm <dmalcolm@redhat.com>
22448 * diagnostic-show-locus.c (layout::layout): Use start and finish
22449 spelling location for the start and finish of each range.
22450 * genmatch.c (linemap_client_expand_location_to_spelling_point):
22451 Add unused aspect param.
22452 * input.c (expand_location_1): Add "aspect" param, and use it
22453 to access the correct part of the location.
22454 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
22456 (expand_location_to_spelling_point): Likewise.
22457 (linemap_client_expand_location_to_spelling_point): Add "aspect"
22458 param, and pass it to expand_location_1.
22460 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
22462 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
22463 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
22464 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
22465 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
22466 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
22467 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
22468 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
22469 _mm_maskz_getmant_ss): New intrinsics.
22470 (__builtin_ia32_getexpss128_mask): Changed to ...
22471 __builtin_ia32_getexpss128_round ... this.
22472 (__builtin_ia32_getexpsd128_mask): Changed to ...
22473 __builtin_ia32_getexpsd128_round ... this.
22474 * config/i386/i386-builtin-types.def
22475 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
22476 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
22477 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
22478 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
22479 __builtin_ia32_getmantss_mask_round): New builtins.
22480 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
22481 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
22482 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
22483 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
22484 * config/i386/sse.md
22485 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
22486 avx512f_sgetexp<mode><mask_scalar_name>
22487 <round_saeonly_scalar_name> ... this.
22488 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
22489 %0, %1, %2<round_saeonly_op3>}): Changed to ...
22490 vgetexp<ssescalarmodesuffix>
22491 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22492 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
22493 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
22494 avx512f_vgetmant<mode><mask_scalar_name>
22495 <round_saeonly_scalar_name> ... this.
22496 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
22497 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
22498 vgetmant<ssescalarmodesuffix>
22499 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
22500 %0<mask_scalar_operand4>, %1, %2
22501 <round_saeonly_scalar_mask_op4>, %3} ... this.
22502 * config/i386/subst.md (mask_scalar_operand4,
22503 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
22504 round_saeonly_scalar_nimm_predicate): New subst attributes.
22506 2017-07-06 Julia Koval <julia.koval@intel.com>
22508 * config/i386/i386.c (ix86_erase_embedded_rounding):
22509 Remove code for old rounding pattern.
22511 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
22513 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
22515 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
22517 * doc/sourcebuild.texi (Test Directives, Variants of
22518 dg-require-support): Add documentation for dg-require-stack-check.
22520 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
22522 * config/i386/subst.md (mask_scalar, round_scalar,
22523 round_saeonly_scalar): New meta-templates.
22524 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
22525 round_scalar_mask_operand3, round_scalar_mask_op3,
22526 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
22527 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
22528 round_saeonly_scalar_constraint,
22529 round_saeonly_scalar_prefix): New subst attribute.
22530 * config/i386/sse.md
22531 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
22532 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
22533 <round_scalar_name> ... this.
22534 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
22535 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
22536 <round_scalar_name> ... this.
22537 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
22538 <sse>_vm<code><mode>3<mask_scalar_name>
22539 <round_saeonly_scalar_name> ... this.
22540 (v<plusminus_mnemonic><ssescalarmodesuffix>
22541 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
22542 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
22543 v<plusminus_mnemonic><ssescalarmodesuffix>
22544 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22545 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
22546 (v<multdiv_mnemonic><ssescalarmodesuffix>
22547 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
22548 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
22549 v<multdiv_mnemonic><ssescalarmodesuffix>
22550 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22551 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
22552 (v<maxmin_float><ssescalarmodesuffix>
22553 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
22554 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
22555 v<maxmin_float><ssescalarmodesuffix>
22556 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22557 %0<mask_scalar_operand3>, %1, %<iptr>2
22558 <round_saeonly_scalar_mask_op3>} ... this.
22560 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
22562 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
22563 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
22565 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
22566 Alan Hayward <alan.hayward@arm.com>
22567 David Sherwood <david.sherwood@arm.com>
22569 * combine.c (simplify_if_then_else): Remove "enum" before
22571 * compare-elim.c (can_eliminate_compare): Likewise.
22572 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
22574 (aarch64_lookup_simd_builtin_type): Likewise.
22575 (aarch64_simd_builtin_type): Likewise.
22576 (aarch64_init_simd_builtin_types): Likewise.
22577 (aarch64_simd_expand_args): Likewise.
22578 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
22580 (aarch64_reverse_mask): Likewise.
22581 (aarch64_simd_emit_reg_reg_move): Likewise.
22582 (aarch64_gen_adjusted_ldpstp): Likewise.
22583 (aarch64_ccmp_mode_to_code): Likewise.
22584 (aarch64_operands_ok_for_ldpstp): Likewise.
22585 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
22586 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
22588 (aarch64_min_divisions_for_recip_mul): Likewise.
22589 (aarch64_reassociation_width): Likewise.
22590 (aarch64_get_condition_code_1): Likewise.
22591 (aarch64_simd_emit_reg_reg_move): Likewise.
22592 (aarch64_simd_attr_length_rglist): Likewise.
22593 (aarch64_reverse_mask): Likewise.
22594 (aarch64_operands_ok_for_ldpstp): Likewise.
22595 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
22596 (aarch64_gen_adjusted_ldpstp): Likewise.
22597 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
22599 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
22600 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
22601 (arm_lookup_simd_builtin_type): Likewise.
22602 (arm_simd_builtin_type): Likewise.
22603 (arm_init_simd_builtin_types): Likewise.
22604 (arm_expand_builtin_args): Likewise.
22605 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
22606 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
22607 (ft32_setup_incoming_varargs): Likewise.
22608 (ft32_function_arg): Likewise.
22609 (ft32_function_arg_advance): Likewise.
22610 (ft32_pass_by_reference): Likewise.
22611 (ft32_arg_partial_bytes): Likewise.
22612 (ft32_valid_pointer_mode): Likewise.
22613 (ft32_addr_space_pointer_mode): Likewise.
22614 (ft32_addr_space_legitimate_address_p): Likewise.
22615 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
22617 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
22618 (ix86_emit_outlined_ms2sysv_restore): Likewise.
22619 (iamcu_alignment): Likewise.
22620 (canonicalize_vector_int_perm): Likewise.
22621 (ix86_noce_conversion_profitable_p): Likewise.
22622 (ix86_mpx_bound_mode): Likewise.
22623 (ix86_operands_ok_for_move_multiple): Likewise.
22624 * config/microblaze/microblaze-protos.h
22625 (microblaze_expand_conditional_branch_reg): Likewise.
22626 * config/microblaze/microblaze.c
22627 (microblaze_expand_conditional_branch_reg): Likewise.
22628 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
22630 (rs6000_reassociation_width): Likewise.
22631 (rs6000_invalid_binary_op): Likewise.
22632 (fusion_p9_p): Likewise.
22633 (emit_fusion_p9_load): Likewise.
22634 (emit_fusion_p9_store): Likewise.
22635 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
22637 (riscv_hard_regno_mode_ok_p): Likewise.
22638 (riscv_address_insns): Likewise.
22639 (riscv_split_symbol): Likewise.
22640 (riscv_legitimize_move): Likewise.
22641 (riscv_function_value): Likewise.
22642 (riscv_hard_regno_nregs): Likewise.
22643 (riscv_expand_builtin): Likewise.
22644 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
22645 (riscv_build_integer): Likewise.
22646 (riscv_split_integer): Likewise.
22647 (riscv_legitimate_constant_p): Likewise.
22648 (riscv_cannot_force_const_mem): Likewise.
22649 (riscv_regno_mode_ok_for_base_p): Likewise.
22650 (riscv_valid_base_register_p): Likewise.
22651 (riscv_valid_offset_p): Likewise.
22652 (riscv_valid_lo_sum_p): Likewise.
22653 (riscv_classify_address): Likewise.
22654 (riscv_legitimate_address_p): Likewise.
22655 (riscv_address_insns): Likewise.
22656 (riscv_load_store_insns): Likewise.
22657 (riscv_force_binary): Likewise.
22658 (riscv_split_symbol): Likewise.
22659 (riscv_force_address): Likewise.
22660 (riscv_legitimize_address): Likewise.
22661 (riscv_move_integer): Likewise.
22662 (riscv_legitimize_const_move): Likewise.
22663 (riscv_legitimize_move): Likewise.
22664 (riscv_address_cost): Likewise.
22665 (riscv_subword): Likewise.
22666 (riscv_output_move): Likewise.
22667 (riscv_canonicalize_int_order_test): Likewise.
22668 (riscv_emit_int_order_test): Likewise.
22669 (riscv_function_arg_boundary): Likewise.
22670 (riscv_pass_mode_in_fpr_p): Likewise.
22671 (riscv_pass_fpr_single): Likewise.
22672 (riscv_pass_fpr_pair): Likewise.
22673 (riscv_get_arg_info): Likewise.
22674 (riscv_function_arg): Likewise.
22675 (riscv_function_arg_advance): Likewise.
22676 (riscv_arg_partial_bytes): Likewise.
22677 (riscv_function_value): Likewise.
22678 (riscv_pass_by_reference): Likewise.
22679 (riscv_setup_incoming_varargs): Likewise.
22680 (riscv_print_operand): Likewise.
22681 (riscv_elf_select_rtx_section): Likewise.
22682 (riscv_save_restore_reg): Likewise.
22683 (riscv_for_each_saved_reg): Likewise.
22684 (riscv_register_move_cost): Likewise.
22685 (riscv_hard_regno_mode_ok_p): Likewise.
22686 (riscv_hard_regno_nregs): Likewise.
22687 (riscv_class_max_nregs): Likewise.
22688 (riscv_memory_move_cost): Likewise.
22689 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
22690 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
22691 (rl78_addr_space_address_mode): Likewise.
22692 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22694 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
22695 (rs6000_reassociation_width): Likewise.
22696 (rs6000_invalid_binary_op): Likewise.
22697 (fusion_p9_p): Likewise.
22698 (emit_fusion_p9_load): Likewise.
22699 (emit_fusion_p9_store): Likewise.
22700 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
22701 (ok_for_simple_move_operands): Likewise.
22702 (ok_for_simple_move_strict_operands): Likewise.
22703 (ok_for_simple_arith_logic_operands): Likewise.
22704 (visium_legitimize_reload_address): Likewise.
22705 (visium_select_cc_mode): Likewise.
22706 (output_cbranch): Likewise.
22707 (visium_split_double_move): Likewise.
22708 (visium_expand_copysign): Likewise.
22709 (visium_expand_int_cstore): Likewise.
22710 (visium_expand_fp_cstore): Likewise.
22711 * config/visium/visium.c (visium_pass_by_reference): Likewise.
22712 (visium_function_arg): Likewise.
22713 (visium_function_arg_advance): Likewise.
22714 (visium_libcall_value): Likewise.
22715 (visium_setup_incoming_varargs): Likewise.
22716 (visium_legitimate_constant_p): Likewise.
22717 (visium_legitimate_address_p): Likewise.
22718 (visium_legitimize_address): Likewise.
22719 (visium_secondary_reload): Likewise.
22720 (visium_register_move_cost): Likewise.
22721 (visium_memory_move_cost): Likewise.
22722 (prepare_move_operands): Likewise.
22723 (ok_for_simple_move_operands): Likewise.
22724 (ok_for_simple_move_strict_operands): Likewise.
22725 (ok_for_simple_arith_logic_operands): Likewise.
22726 (visium_function_value_1): Likewise.
22727 (rtx_ok_for_offset_p): Likewise.
22728 (visium_legitimize_reload_address): Likewise.
22729 (visium_split_double_move): Likewise.
22730 (visium_expand_copysign): Likewise.
22731 (visium_expand_int_cstore): Likewise.
22732 (visium_expand_fp_cstore): Likewise.
22733 (visium_split_cstore): Likewise.
22734 (visium_select_cc_mode): Likewise.
22735 (visium_split_cbranch): Likewise.
22736 (output_cbranch): Likewise.
22737 (visium_print_operand_address): Likewise.
22738 * expmed.c (flip_storage_order): Likewise.
22739 * expmed.h (emit_cstore): Likewise.
22740 (flip_storage_order): Likewise.
22741 * genrecog.c (validate_pattern): Likewise.
22742 * hsa-gen.c (gen_hsa_addr): Likewise.
22743 * internal-fn.c (expand_arith_overflow): Likewise.
22744 * ira-color.c (allocno_copy_cost_saving): Likewise.
22745 * lra-assigns.c (find_hard_regno_for_1): Likewise.
22746 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
22747 (process_invariant_for_inheritance): Likewise.
22748 * lra-eliminations.c (move_plus_up): Likewise.
22749 * omp-low.c (lower_oacc_reductions): Likewise.
22750 * simplify-rtx.c (simplify_subreg): Likewise.
22751 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
22752 (TARGET_CHKP_BOUND_MODE): Likewise..
22753 * targhooks.c (default_chkp_bound_mode): Likewise.
22754 (default_setup_incoming_vararg_bounds): Likewise.
22755 * targhooks.h (default_chkp_bound_mode): Likewise.
22756 (default_setup_incoming_vararg_bounds): Likewise.
22757 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
22758 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
22759 (have_whole_vector_shift): Likewise.
22760 * tree-vect-stmts.c (vectorizable_load): Likewise.
22761 * doc/tm.texi: Regenerate.
22763 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22765 Graceful degrade if Binutils PR21472 is not available.
22768 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
22769 .rodata in flash test fails.
22770 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
22771 * confgure: Regenerate.
22772 * config.in: Regenerate.
22773 * config/avr/avr.c (avr_asm_named_section)
22774 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
22775 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
22776 (avr_asm_init_sections): Same.
22778 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22780 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
22781 (fma<VH:mode>4_intrinsic): Likewise.
22782 (*fmsub<VCVTF:mode>4): Likewise.
22783 (*fmsub<VH:mode>4_intrinsic): Likewise.
22785 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22788 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
22789 Don't depend on "optimize > 0".
22790 (out_movhi_r_mr, out_movqi_mr_r): Same.
22791 (out_movhi_mr_r, out_movqi_r_mr): Same.
22792 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
22793 io_address_operand on "optimize > 0".
22795 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22797 * tree-loop-distribution.c: Add general explanantion on the pass.
22798 (generate_loops_for_partition): Mark distributed loop.
22799 (pg_add_dependence_edges): New parameter. Handle alias data
22800 dependence specially and record it in the parameter if asked.
22801 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
22802 (init_partition_graph_vertices, add_partition_graph_edge): New.
22803 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
22804 (free_partition_graph_vdata, build_partition_graph): New.
22805 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
22806 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
22807 (data_ref_segment_size, latch_dominated_by_data_ref): New.
22808 (compute_alias_check_pairs, version_loop_by_alias_check): New.
22809 (version_for_distribution_p, finalize_partitions): New.
22810 (distribute_loop): Handle alias data dependence specially. Factor
22811 out loop fusion code as functions and call these functions.
22813 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22815 * tree-loop-distribution.c (classify_partition): New parameter and
22816 better handle reduction statement.
22817 (rdg_build_partitions): Revise comment.
22818 (distribute_loop): Compute statements in all partitions and pass it
22819 to classify_partition.
22821 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22823 * tree-loop-distribution.c (enum partition_type): New.
22824 (struct partition): New field type.
22825 (partition_merge_into): Add parameter. Update partition type.
22826 (data_dep_in_cycle_p, update_type_for_merge): New functions.
22827 (build_rdg_partition_for_vertex): Compute partition type.
22828 (rdg_build_partitions): Dump partition type.
22829 (distribute_loop): Update calls to partition_merge_into.
22831 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22833 * tree-loop-distribution.c (struct ddr_hasher): New.
22834 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
22836 (classify_partition): Call get_data_dependence.
22837 (pg_add_dependence_edges): Ditto.
22838 (distribute_loop): Release data dependence hash table.
22840 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22842 * tree-loop-distribution.c (ref_base_address): Delete.
22843 (similar_memory_accesses): Rename ...
22844 (share_memory_accesses): ... to this. Check if partitions access
22845 the same memory reference.
22846 (distribute_loop): Call share_memory_accesses.
22848 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22850 * tree-loop-distribution.c (struct partition): New field recording
22851 its data reference.
22852 (partition_alloc, partition_free): Init and release data refs.
22853 (partition_merge_into): Merge data refs.
22854 (build_rdg_partition_for_vertex): Collect data refs for partition.
22855 (pg_add_dependence_edges): Change parameters from vector to bitmap.
22857 (distribute_loop): Remve data refs from vertice data of partition
22860 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22862 * tree-loop-distribution.c (params.h): Include header file.
22863 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
22864 (datarefs_vec): New global var.
22865 (create_rdg_vertices): Use datarefs_vec directly.
22866 (free_rdg): Don't free data references.
22867 (build_rdg): Update use. Don't free data references.
22868 (distribute_loop): Compute global variable for data references.
22869 Bail out if there are too many data references.
22871 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22873 * tree-loop-distribution.c (loop_nest): New global var.
22874 (build_rdg): Use loop directly, rather than loop nest.
22875 (pg_add_dependence_edges): Remove loop nest parameter. Use global
22877 (distribute_loop): Compute global variable loop nest. Update use.
22879 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22881 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
22882 (partition_merge_into): New parameter. Dump reason for fusion.
22883 (distribute_loop): Update use of partition_merge_into.
22885 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22887 * tree-loop-distribution.c (bb_top_order_index): New.
22888 (bb_top_order_index_size, bb_top_order_cmp): New.
22889 (stmts_from_loop): Use topological order.
22890 (pass_loop_distribution::execute): Compute and release topological
22891 order for basic blocks.
22893 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22895 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
22898 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22900 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
22901 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
22902 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
22903 * internal-fn.def (LOOP_DIST_ALIAS): New.
22904 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
22905 (fold_loop_internal_call): ... this.
22906 (vect_loop_dist_alias_call): New function.
22907 (set_uid_loop_bbs): Call fold_loop_internal_call.
22908 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
22911 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22914 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
22915 Require dead FLAGS_REG at the beginning of a peephole.
22917 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22920 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
22921 arguments in the call to __builtin_ia32_sbb_u32.
22922 (_subborrow_u64): Swap _X and _Y arguments in the call to
22923 __builtin_ia32_sbb_u64.
22925 2017-07-04 Jakub Jelinek <jakub@redhat.com>
22928 * tree-vrp.c (compare_assert_loc): Turn into a function template
22929 with stable template parameter. Only test if a->e is NULL,
22930 !a->e == !b->e has been verified already. Use e == NULL or
22931 e != NULL instead of e or ! e tests. If stable is true, don't use
22932 iterative_hash_expr, on the other side allow a or b or both NULL
22933 and sort the NULLs last.
22934 (process_assert_insertions): Sort using compare_assert_loc<false>
22935 instead of compare_assert_loc, later sort using
22936 compare_assert_loc<true> before calling process_assert_insertions_for
22937 in a loop. Use break instead of continue once seen NULL pointer.
22939 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22941 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
22942 Cortex-R7 and Cortex-R8 processors.
22944 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22946 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
22947 uninitialized while src is not.
22949 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
22951 * common/config/arm/arm-common.c: Adjust include path for
22953 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
22954 (arm-cpu.h): Create in build directory. Adjust dependency rules.
22955 (arm-cpu-data.h): Likewise.
22956 (arm-cpu-cdata.h): Likewise.
22957 * config/arm/arm-cpu.h: Delete.
22958 * config/arm/arm-cpu-cdata.h: Delete.
22959 * config/arm/arm-cpu-data.h: Delete.
22961 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
22963 * config/arm/arm-cpus.in (cortex-a55): New.
22964 (cortex-a75): Likewise.
22965 (cortex-a75.cortex-a55): Likewise.
22966 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
22968 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
22969 * config/arm/arm-cpu-cdata.h: Regenerate.
22970 * config/arm/arm-cpu-data.h: Regenerate.
22971 * config/arm/arm-cpu.h: Regenerate.
22972 * config/arm/arm-tables.opt: Regenerate.
22973 * config/arm/arm-tune.md: Regenerate.
22975 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22977 * haifa-sched.c (sched_create_recovery_edges): Update profile.
22979 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22981 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
22984 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
22986 PR tree-optimization/81292
22987 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
22988 full_string_p, also call adjust_related_strinfos if the adjustment
22989 is simple, otherwise invalidate related strinfos.
22991 2017-07-04 Martin Liska <mliska@suse.cz>
22994 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
22995 newly created variable as DECL_IGNORED_P.
22997 2017-07-04 Martin Liska <mliska@suse.cz>
23000 * ipa-inline.c (inline_small_functions):
23001 Use xstrdup_for_dump.
23003 2017-07-04 Tom de Vries <tom@codesourcery.com>
23005 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
23007 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
23010 * config/darwin.c (darwin_function_switched_text_sections):
23013 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
23015 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
23017 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
23019 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
23021 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23023 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
23024 min_profitable_iters, and th as inclusive lower bounds.
23025 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
23026 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
23027 for min_profitable_iters and min_profitable_estimate.
23028 (vect_transform_loop): Treat th as an inclusive lower bound.
23029 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
23031 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
23034 * config/darwin.c (darwin_function_switched_text_sections):
23035 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
23036 in two pieces, and suppress the use of buf.
23038 2017-07-03 Nathan Sidwell <nathan@acm.org>
23040 * hash-table.h (hash_table_mod1): Fix indentation.
23042 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23044 PR middle-end/81290
23045 * predict.c (force_edge_cold): Be more careful about propagation
23047 * profile-count.h (profile_probability::guessed,
23048 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
23050 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
23052 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
23054 * doc/invoke.texi (rcpc architecture extension): Document it.
23056 2017-07-03 Richard Biener <rguenther@suse.de>
23058 PR tree-optimization/60510
23059 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
23060 the scalar reduction PHI and use it.
23061 (vectorizable_reduction): Properly guard the single_defuse_cycle
23062 path for non-SLP reduction chains where we cannot use it.
23063 Rework reduc_def/index and vector type deduction. Rework
23064 vector operand gathering during reduction op code-gen.
23065 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
23066 chains dissolve the chain and leave it to non-SLP reduction
23069 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23071 * tree-data-ref.h (dr_alignment): Declare.
23072 * tree-data-ref.c (dr_alignment): New function.
23073 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
23074 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
23076 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
23078 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23080 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
23081 and base_misalignment fields.
23082 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
23083 * tree-data-ref.c: Include builtins.h.
23084 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
23085 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
23086 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
23087 * tree-vect-data-refs.c: Include tree-cfg.h.
23088 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
23089 fields instead of calculating an alignment here.
23090 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
23091 innermost_loop_behavior fields.
23093 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23095 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
23097 (DR_STEP_ALIGNMENT): New macro.
23098 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
23099 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
23100 (create_data_ref): Print it.
23101 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
23102 to tell whether the step preserves vector (mis)alignment.
23103 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
23104 Move the check for an integer step and generalise to all INTEGER_CST.
23105 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
23106 Print the outer step alignment.
23108 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23110 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
23111 with offset_alignment.
23112 (DR_ALIGNED_TO): Delete.
23113 (DR_OFFSET_ALIGNMENT): New macro.
23114 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
23115 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
23116 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
23117 (create_data_ref): Likewise.
23118 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
23119 (vect_analyze_data_refs): Likewise.
23120 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
23121 creating dummy innermost behavior.
23123 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23125 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
23126 with a "innermost_loop_behavior *" and refeence tree.
23127 * tree-data-ref.c (dr_analyze_innermost): Likewise.
23128 (create_data_ref): Update call accordingly.
23129 * tree-predcom.c (find_looparound_phi): Likewise.
23131 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23133 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
23134 fields with dr_wrt_vec_loop.
23135 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
23136 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
23137 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
23138 (vect_dr_behavior): New function.
23139 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
23140 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
23141 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
23142 track whether the step preserves the misalignment.
23143 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
23144 Use vect_dr_behavior.
23145 (vect_setup_realignment): Update call accordingly.
23146 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
23147 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
23148 call to vect_create_addr_base_for_vector_ref.
23149 (vect_create_cond_for_align_checks): Likewise.
23150 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
23151 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
23152 (vect_recog_mask_conversion_pattern): Likewise.
23153 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
23154 (new_stmt_vec_info): Remove redundant zeroing.
23156 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
23158 * common/config/arm/arm-common.c (arm_be8_option): New function.
23159 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
23160 (ISA_ARMv6): Add isa_bit_be8.
23161 * config/arm/arm.h (arm_be8_option): Add prototype.
23162 (BE8_SPEC_FUNCTION): New define.
23163 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
23164 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
23165 (mlittle-endian): Similarly.
23166 (mbe8, mbe32): New options.
23167 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
23168 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
23170 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23172 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
23174 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23176 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
23177 (cleanup_tree_cfg_bb): Use it.
23178 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
23180 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
23182 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23185 * loop-doloop.c (add_test): Update profile.
23187 2017-07-03 Martin Liska <mliska@suse.cz>
23190 * sanopt.c (rewrite_usage_of_param): New function.
23191 (sanitize_rewrite_addressable_params): Likewise.
23192 (pass_sanopt::execute): Call rewrite_usage_of_param.
23194 2017-07-03 Richard Biener <rguenther@suse.de>
23196 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
23197 back to using VIEW_CONVERT_EXPR.
23199 2017-07-03 Martin Liska <mliska@suse.cz>
23202 * doc/extend.texi: Document when a resolver function is
23203 generated for target_clones.
23205 2017-07-03 Martin Liska <mliska@suse.cz>
23207 * asan.c (asan_emit_stack_protection): Unpoison just red zones
23208 and shadow memory of auto variables which are subject of
23209 use-after-scope sanitization.
23210 (asan_expand_mark_ifn): Add do set only when is_poison.
23212 2016-07-03 Richard Biener <rguenther@suse.de>
23214 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
23216 (vect_force_simple_reduction): Record reduction def -> phi mapping.
23217 (vectorizable_reduction): Perform reduction PHI creation when
23218 visiting a reduction PHI and adjust and simplify code generation
23219 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
23220 (vect_transform_loop): Visit reduction PHIs.
23221 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
23222 defs into the SLP tree.
23223 (vect_build_slp_tree): Reduction defs terminate the recursion.
23224 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
23226 (vect_get_vec_defs_for_stmt_copy): Export.
23227 (vect_get_vec_defs): Likewise.
23228 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
23230 (vect_get_vec_defs_for_stmt_copy): Declare.
23231 (vect_get_vec_defs): Likewise.
23233 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23235 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
23236 parameter with a "loop" parameter and use it instead of the
23237 loop containing DR_STMT. Don't check simple_iv when doing
23238 BB analysis. Describe the two analysis modes in the comment.
23240 2017-07-03 Tom de Vries <tom@codesourcery.com>
23242 PR tree-optimization/69468
23243 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
23244 (find_same_succ_bb): Handle ignore_edge_flags.
23246 2017-07-03 Tom de Vries <tom@codesourcery.com>
23248 PR tree-optimization/81192
23249 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
23251 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
23253 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
23255 2017-07-03 Tom de Vries <tom@codesourcery.com>
23257 PR tree-optimization/81192
23258 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
23259 BB_SAME_SUCC (bb) == NULL.
23261 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23263 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
23266 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23268 * dumpfile.c: Include profile-count.h
23269 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
23271 (insert_cond_bb): Update profile.
23272 * tree-cfg.h (insert_cond_bb): Update prototype.
23273 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
23274 * tree-dump.c: Do not include tree-cfg.
23276 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23278 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
23280 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23282 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
23285 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23287 * tree-complex.c (expand_complex_div_wide): update profile.
23289 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23290 Alan Hayward <alan.hayward@arm.com>
23291 David Sherwood <david.sherwood@arm.com>
23293 * Makefile.in (MACHMODE_H): Remove insn-modes.h
23294 (CORETYPES_H): New define.
23295 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
23296 (insn-modes-inline.h, s-modes-inline-h): New rules.
23297 (generated_files): Add insn-modes-inline.h.
23298 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
23299 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
23300 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
23301 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
23302 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
23303 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
23304 (build/gencodes.o, build/genconditions.o): Likewise.
23305 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
23306 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
23307 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
23308 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
23309 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
23310 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
23311 * coretypes.h: Include everything up to real.h for generators.
23312 Include insn-modes.h first. Include wide-int-print.h after
23313 wide-int.h. Include insn-modes-inline.h and then machmode.h.
23314 * machmode.h: Don't include insn-modes.h here.
23315 * function-tests.c: Remove includes of signop.h, machmode.h,
23316 double-int.h and wide-int.h.
23318 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
23320 * optc-save-gen.awk: Likewise.
23321 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
23322 * godump.c: Remove include of wide-int-print.h.
23323 * pretty-print.h: Likewise.
23324 * wide-int-print.cc: Likewise.
23325 * wide-int.cc: Likewise.
23326 * hash-map-tests.c: Remove include of signop.h.
23327 * hash-set-tests.c: Likewise.
23328 * rtl-tests.c: Likewise.
23329 * mkconfig.sh: Remove include of machmode.h.
23330 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
23332 (emit_insn_modes_inline_h): ...this new function. Emit the code
23333 into an insn-modes-inline.h header file, adding appropriate
23334 include guards and end comments.
23335 (emit_insn_modes_c_header): Remove include of machmode.h.
23336 (emit_min_insn_modes_c_header): Include coretypes.h rather than
23338 (main): Handle -i flag and call emit_insn_modes_inline_h when
23341 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23343 * tree-ssa-strlen.c (strinfo): Rename the length field to
23344 nonzero_chars. Add a full_string_p field.
23345 (compare_nonzero_chars, zero_length_string_p): New functions.
23346 (get_addr_stridx): Add an offset_out parameter.
23347 Use compare_nonzero_chars.
23348 (get_stridx): Update accordingly. Use compare_nonzero_chars.
23349 (new_strinfo): Update after above changes to strinfo.
23350 (set_endptr_and_length): Set full_string_p.
23351 (get_string_length): Update after above changes to strinfo.
23352 (unshare_strinfo): Update call to new_strinfo.
23353 (maybe_invalidate): Likewise.
23354 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
23355 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
23356 as a uhwi instead of an shwi. Update after above changes to
23357 strinfo and new_strinfo.
23358 (zero_length_string): Assert that chainsi contains full strings.
23359 Use zero_length_string_p. Update call to new_strinfo.
23360 (adjust_related_strinfos): Update after above changes to strinfo.
23361 Copy full_string_p from origsi.
23362 (adjust_last_stmt): Use zero_length_string_p.
23363 (handle_builtin_strlen): Update after above changes to strinfo and
23364 new_strinfo. Install the lhs as the string length if the previous
23365 entry didn't describe a full string.
23366 (handle_builtin_strchr): Update after above changes to strinfo
23368 (handle_builtin_strcpy): Likewise.
23369 (handle_builtin_strcat): Likewise.
23370 (handle_builtin_malloc): Likewise.
23371 (handle_pointer_plus): Likewise.
23372 (handle_builtin_memcpy): Likewise. Track nonzero characters
23373 that aren't necessarily followed by a nul terminator.
23374 (handle_char_store): Likewise.
23376 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23378 PR tree-optimization/80769
23379 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
23380 for malloc and calloc. Document the new invariant that all related
23381 strinfos have delayed lengths or none do.
23382 (verify_related_strinfos): Move earlier in file.
23383 (set_endptr_and_length): New function, split out from...
23384 (get_string_length): ...here. Also set the lengths of related
23386 (zero_length_string): Assert that chainsi has known (rather than
23388 (adjust_related_strinfos): Likewise.
23390 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23392 PR tree-optimization/81136
23393 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
23394 assert that two references with the same misalignment have the same
23395 compile-time misalignment if those compile-time misalignments
23398 2017-07-01 Andi Kleen <ak@linux.intel.com>
23400 * print-tree.c (print_node): Print all attributes.
23402 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23404 * cfg.c (scale_bbs_frequencies): New function.
23405 * cfg.h (scale_bbs_frequencies): Declare it.
23406 * cfgloopanal.c (single_likely_exit): Cleanup.
23407 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
23409 (scale_loop_profile): Likewise.
23410 (loop_version): Likewise.
23411 (create_empty_loop_on_edge): Update.
23412 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
23413 scale_loop_frequencies, scale_loop_profile, loopify,
23414 loop_version): Update prototypes.
23415 * modulo-sched.c (sms_schedule): Update.
23416 * predict.c (unlikely_executed_edge_p): Also check probability.
23417 (probably_never_executed_edge_p): Fix typo.
23418 * tree-if-conv.c (version_loop_for_if_conversion): Update.
23419 * tree-parloops.c (gen_parallel_loop): Update.
23420 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
23421 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23422 * tree-ssa-loop-split.c (split_loop): Update.
23423 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
23424 * tree-vect-loop-manip.c (vect_do_peeling): Update.
23425 (vect_loop_versioning): Update.
23426 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23428 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23430 * trans-mem.c (split_bb_make_tm_edge): Update profile.
23432 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23434 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
23435 to keep profile consistent.
23437 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23439 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
23440 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
23441 * profile-count.h (max_safe_multiplier): Make unsigned.
23442 (profile_count::guessed_zero): New.
23444 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23446 * bb-reorder.c (fix_up_crossing_landing_pad,
23447 fix_crossing_conditional_branches): Use make_single_succ_edge
23448 to keep profile consistent.
23450 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23452 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
23455 2017-07-01 Jakub Jelinek <jakub@redhat.com>
23458 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
23459 the right scopes, make sure cond_jump isn't preserved between multiple
23460 iterations. Search for fallthru edge whenever there are 3+ edges and
23461 use find_fallthru_edge for it.
23463 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23465 Patch by Alexander Monakov <amonakov@ispras.ru>
23466 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
23467 probabilities consistently.
23469 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23471 * pa.c (pa_expand_compare_and_swap_loop): Update call of
23472 emit_cmp_and_jump_insns.
23474 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23477 * tree-inline.c (expand_call_inline): Combine profile statuses.
23479 2017-06-30 Andrew Pinski <apinski@cavium.com>
23481 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
23482 fold_stmt returned true.
23484 2017-06-30 Nathan Sidwell <nathan@acm.org>
23486 * ggc.h (empty_string): Delete.
23487 * cfgexpand.c (expand_asm_stmt): Use plain "".
23488 * optabs.c (expand_asm_memory_barrier): Likewise.
23489 * stringpool.c (empty_string): Delete.
23490 (digit_vector, digit_string): Delete.
23491 (ggc_alloc_string): Use plain "", don't optimize single digit
23492 strings. Use ggc_alloc_atomic.
23494 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
23496 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
23497 comparison set and one other set, use the cost of the non-comparison
23500 2017-06-30 Nathan Sidwell <nathan@acm.org>
23502 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
23505 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
23507 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
23508 loops. Remove now unneeded calls to gimple_switch_set_label() that
23509 just set removed labels to NULL_TREE.
23511 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
23513 * tree-ssanames.c (set_range_info_raw): Abstract from ...
23514 (set_range_info): ...here. Only call set_range_info_raw if domain
23516 (set_nonzero_bits): Call set_range_info_raw.
23517 * tree-ssanames.h (set_range_info_raw): New.
23519 2017-06-30 Jakub Jelinek <jakub@redhat.com>
23522 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
23523 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
23524 of nonimmediate_operand and <store_mask_constraint> instead of m
23525 for the input operand. For V8FI iterator, always split if input
23526 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
23527 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
23528 <store_mask_predicate> instead of register_operand and
23529 <store_mask_constraint> instead of v for the input operand. Make
23530 sure both operands aren't MEMs for if not <mask_applied>.
23532 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
23534 * lto-wrapper.c (copy_file) Close both file descriptors before
23537 2017-06-30 Martin Liska <mliska@suse.cz>
23540 * multiple_target.c (create_dispatcher_calls): Make ifunc
23541 also for function that don't have calls or are not referenced.
23543 2017-06-30 Richard Biener <rguenther@suse.de>
23545 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
23546 analyze the first scalar stmt. Move vector type computation
23547 for the BB case here from ...
23548 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
23549 live operation processing in the SLP case properly.
23551 2017-06-30 Richard Biener <rguenther@suse.de>
23553 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
23555 2017-06-30 Martin Liska <mliska@suse.cz>
23558 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
23559 before BUILT_IN_UNWIND_RESUME when ASAN is used.
23561 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
23563 * doc/invoke.texi (AArch64): Add missing options and remove redundant
23566 2017-06-30 Richard Biener <rguenther@suse.de>
23568 PR tree-optimization/81249
23569 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
23570 condition reduction result to original scalar type.
23572 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23574 * profile-count.h (enum profile_quality): Fix typos and whitespace
23577 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23579 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
23580 type for branch probabilities.
23582 2017-06-29 Julian Brown <julian@codesourcery.com>
23583 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23585 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
23586 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
23587 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
23588 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
23590 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23592 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
23593 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
23594 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
23595 CC usage from generic code to here.
23596 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
23597 CC usage into the target macros.
23599 2017-06-29 Maya Rashish <coypu@sdf.org>
23601 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
23604 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23606 * arm/arm-builtins.c: Include profile-count.h
23607 * except.c (sjlj_emit_function_enter): Use
23608 profile_probability::unlikely.
23610 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23612 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
23613 and tocrel_offset be pointer args rather than implicitly using
23615 (legitimate_constant_pool_address_p, rs6000_emit_move,
23616 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
23617 tocrel_offset and use in toc_relative_expr_p call.
23618 (print_operand, print_operand_address): Use static tocrel_base_oac
23619 and tocrel_offset_oac.
23620 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
23623 2017-06-29 Maya Rashish <coypu@sdf.org>
23625 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
23627 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
23629 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
23630 objects, take into account only the alignment of 'op0' and 'mode1' if
23633 2017-06-29 Steve Ellcey <sellcey@cavium.com>
23635 * ccmp.c (ccmp_tree_comparison_p): New function.
23636 (ccmp_candidate_p): Update to use above function.
23637 (get_compare_parts): New function.
23638 (expand_ccmp_next): Update to use new functions.
23639 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
23641 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
23642 take mode as argument.
23643 * ccmp.h (expand_ccmp_expr): Add mode as argument.
23644 * expr.c (expand_expr_real_1): Pass mode as argument.
23646 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
23648 * combine.c (combine_instructions): Print insns to dump_file, together
23651 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23653 * asan.c (asan_emit_stack_protection): Update.
23654 (create_cond_insert_point): Update.
23655 * auto-profile.c (afdo_propagate_circuit): Update.
23656 * basic-block.h (struct edge_def): Turn probability to
23657 profile_probability.
23658 (EDGE_FREQUENCY): Update.
23659 * bb-reorder.c (find_traces_1_round): Update.
23660 (better_edge_p): Update.
23661 (sanitize_hot_paths): Update.
23662 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
23663 (make_single_succ_edge): Update.
23664 (check_bb_profile): Update.
23665 (dump_edge_info): Update.
23666 (update_bb_profile_for_threading): Update.
23667 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
23668 probabilitycount to 0.
23669 * cfgbuild.c (compute_outgoing_frequencies): Update.
23670 * cfgcleanup.c (try_forward_edges): Update.
23671 (outgoing_edges_match): Update.
23672 (try_crossjump_to_edge): Update.
23673 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
23674 (expand_gimple_tailcall): Update.
23675 (construct_init_block): Use make_single_succ_edge.
23676 (construct_exit_block): Use make_single_succ_edge.
23677 * cfghooks.c (verify_flow_info): Update.
23678 (redirect_edge_succ_nodup): Update.
23679 (split_edge): Update.
23680 (account_profile_record): Update.
23681 * cfgloopanal.c (single_likely_exit): Update.
23682 * cfgloopmanip.c (scale_loop_profile): Update.
23683 (set_zero_probability): Remove.
23684 (duplicate_loop_to_header_edge): Update.
23685 * cfgloopmanip.h (loop_version): Update prototype.
23686 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
23687 (force_nonfallthru_and_redirect): Update.
23688 (update_br_prob_note): Update.
23689 (rtl_verify_edges): Update.
23690 (purge_dead_edges): Update.
23691 (rtl_lv_add_condition_to_bb): Update.
23692 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
23693 * cgraphunit.c (init_lowered_empty_function): Update.
23694 (cgraph_node::expand_thunk): Update.
23695 * cilk-common.c: Include profile-count.h
23696 * dojump.c (inv): Remove.
23697 (jumpifnot): Update.
23698 (jumpifnot_1): Update.
23699 (do_jump_1): Update.
23701 (do_jump_by_parts_greater_rtx): Update.
23702 (do_compare_rtx_and_jump): Update.
23703 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
23704 do_jump_1. do_compare_rtx_and_jump): Update prototype.
23705 * dwarf2cfi.c: Include profile-count.h
23706 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
23707 (sjlj_emit_dispatch_table): Likewise.
23708 * explow.c: Include profile-count.h
23709 * expmed.c (emit_store_flag_force): Update.
23710 (do_cmp_and_jump): Update.
23711 * expr.c (compare_by_pieces_d::generate): Update.
23712 (compare_by_pieces_d::finish_mode): Update.
23713 (emit_block_move_via_loop): Update.
23714 (store_expr_with_bounds): Update.
23715 (store_constructor): Update.
23716 (expand_expr_real_2): Update.
23717 (expand_expr_real_1): Update.
23718 * expr.h (try_casesi, try_tablejump): Update prototypes.
23719 * gimple-pretty-print.c (dump_probability): Update.
23720 (dump_profile): New.
23721 (dump_gimple_label): Update.
23722 (dump_gimple_bb_header): Update.
23723 * graph.c (draw_cfg_node_succ_edges): Update.
23724 * hsa-gen.c (convert_switch_statements): Update.
23725 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
23726 (find_if_case_1): Update.
23727 (find_if_case_2): Update.
23728 * internal-fn.c (expand_arith_overflow_result_store): Update.
23729 (expand_addsub_overflow): Update.
23730 (expand_neg_overflow): Update.
23731 (expand_mul_overflow): Update.
23732 (expand_vector_ubsan_overflow): Update.
23733 * ipa-cp.c (good_cloning_opportunity_p): Update.
23734 * ipa-split.c (split_function): Use make_single_succ_edge.
23735 * ipa-utils.c (ipa_merge_profiles): Update.
23736 * loop-doloop.c (add_test): Update.
23737 (doloop_modify): Update.
23738 * loop-unroll.c (compare_and_jump_seq): Update.
23739 (unroll_loop_runtime_iterations): Update.
23740 * lra-constraints.c (lra_inheritance): Update.
23741 * lto-streamer-in.c (input_cfg): Update.
23742 * lto-streamer-out.c (output_cfg): Update.
23743 * mcf.c (adjust_cfg_counts): Update.
23744 * modulo-sched.c (sms_schedule): Update.
23745 * omp-expand.c (expand_omp_for_init_counts): Update.
23746 (extract_omp_for_update_vars): Update.
23747 (expand_omp_ordered_sink): Update.
23748 (expand_omp_for_ordered_loops): Update.
23749 (expand_omp_for_generic): Update.
23750 (expand_omp_for_static_nochunk): Update.
23751 (expand_omp_for_static_chunk): Update.
23752 (expand_cilk_for): Update.
23753 (expand_omp_simd): Update.
23754 (expand_omp_taskloop_for_outer): Update.
23755 (expand_omp_taskloop_for_inner): Update.
23756 * omp-simd-clone.c (simd_clone_adjust): Update.
23757 * optabs.c (expand_doubleword_shift): Update.
23758 (expand_abs): Update.
23759 (emit_cmp_and_jump_insn_1): Update.
23760 (expand_compare_and_swap_loop): Update.
23761 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
23762 * predict.c (predictable_edge_p): Update.
23763 (edge_probability_reliable_p): Update.
23764 (set_even_probabilities): Update.
23765 (combine_predictions_for_insn): Update.
23766 (combine_predictions_for_bb): Update.
23767 (propagate_freq): Update.
23768 (estimate_bb_frequencies): Update.
23769 (force_edge_cold): Update.
23770 * profile-count.c (profile_count::dump): Add missing space into dump.
23771 (profile_count::debug): Add newline.
23772 (profile_count::differs_from_p): Explicitly convert to unsigned.
23773 (profile_count::stream_in): Update.
23774 (profile_probability::dump): New member function.
23775 (profile_probability::debug): New member function.
23776 (profile_probability::differs_from_p): New member function.
23777 (profile_probability::differs_lot_from_p): New member function.
23778 (profile_probability::stream_in): New member function.
23779 (profile_probability::stream_out): New member function.
23780 * profile-count.h (profile_count_quality): Rename to ...
23781 (profile_quality): ... this one.
23782 (profile_probability): New.
23783 (profile_count): Update.
23784 * profile.c (compute_branch_probabilities): Update.
23785 * recog.c (peep2_attempt): Update.
23786 * sched-ebb.c (schedule_ebbs): Update.
23787 * sched-rgn.c (find_single_block_region): Update.
23788 (compute_dom_prob_ps): Update.
23789 (schedule_region): Update.
23790 * sel-sched-ir.c (compute_succs_info): Update.
23791 * stmt.c (struct case_node): Update.
23792 (do_jump_if_equal): Update.
23793 (get_outgoing_edge_probs): Update.
23794 (conditional_probability): Update.
23795 (emit_case_dispatch_table): Update.
23796 (expand_case): Update.
23797 (expand_sjlj_dispatch_table): Update.
23798 (emit_case_nodes): Update.
23799 * targhooks.c: Update.
23800 * tracer.c (better_p): Update.
23801 (find_best_successor): Update.
23802 * trans-mem.c (expand_transaction): Update.
23803 * tree-call-cdce.c: Update.
23804 * tree-cfg.c (gimple_split_edge): Upate.
23805 (move_sese_region_to_fn): Upate.
23806 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
23807 * tree-eh.c (lower_resx): Upate.
23808 (cleanup_empty_eh_move_lp): Upate.
23809 * tree-if-conv.c (version_loop_for_if_conversion): Update.
23810 * tree-inline.c (copy_edges_for_bb): Update.
23811 (copy_cfg_body): Update.
23812 * tree-parloops.c (gen_parallel_loop): Update.
23813 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
23814 (gimple_gen_time_profiler): Update.
23815 * tree-ssa-dce.c (remove_dead_stmt): Update.
23816 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23817 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
23818 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
23819 (unloop_loops): Update.
23820 (try_peel_loop): Update.
23821 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23822 * tree-ssa-loop-split.c (connect_loops): Update.
23823 (split_loop): Update.
23824 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
23825 (hoist_guard): Update.
23826 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
23827 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
23828 (value_replacement): Update.
23829 * tree-ssa-reassoc.c (branch_fixup): Update.
23830 * tree-ssa-tail-merge.c (replace_block_by): Update.
23831 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
23832 (create_edge_and_update_destination_phis): Update.
23833 (compute_path_counts): Update.
23834 (recompute_probabilities): Update.
23835 (update_joiner_offpath_counts): Update.
23836 (freqs_to_counts_path): Update.
23837 (duplicate_thread_path): Update.
23838 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
23839 (struct switch_conv_info): Update.
23840 (gen_inbound_check): Update.
23841 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23842 (vect_do_peeling): Update.
23843 (vect_loop_versioning): Update.
23844 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23845 (optimize_mask_stores): Update.
23846 * ubsan.c (ubsan_expand_null_ifn): Update.
23847 * value-prof.c (gimple_divmod_fixed_value): Update.
23848 (gimple_divmod_fixed_value_transform): Update.
23849 (gimple_mod_pow2): Update.
23850 (gimple_mod_pow2_value_transform): Update.
23851 (gimple_mod_subtract): Update.
23852 (gimple_mod_subtract_transform): Update.
23853 (gimple_ic): Update.
23854 (gimple_stringop_fixed_value): Update.
23855 (gimple_stringops_transform): Update.
23856 * value-prof.h: Update.
23858 2017-06-29 Carl Love <cel@us.ibm.com>
23860 * config/rs6000/rs6000-c.c: Add support for built-in functions
23861 vector signed int vec_signed (vector float);
23862 vector signed long long vec_signed (vector double);
23863 vector signed int vec_signed2 (vector double, vector double);
23864 vector signed int vec_signede (vector double);
23865 vector signed int vec_signedo (vector double);
23866 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
23867 instruction generator.
23868 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23869 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
23870 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
23872 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
23873 vunsignede_v2df): Add define_expands.
23874 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
23875 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
23876 VEC_UNSIGNEDO): Add definitions.
23877 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23878 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
23879 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
23880 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
23881 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
23882 * config/rs6000/altivec.h (vec_signed, vec_signed2,
23883 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
23884 vec_unsignede, vec_unsignedo): Add builtin defines.
23885 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
23887 * doc/extend.texi: Update the built-in documentation file for the
23888 new built-in functions.
23890 2017-06-29 Richard Biener <rguenther@suse.de>
23892 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
23893 reduction chains to LOOP_VINFO_REDUCTIONS.
23894 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
23895 SLP reductions after processing reduction chains.
23897 2017-06-29 Nathan Sidwell <nathan@acm.org>
23899 * builtins.c (fold_builtin_FUNCTION): Use
23900 lang_hooks.decl_printable_name.
23902 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
23904 PR middle-end/81194
23905 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
23906 with only one label.
23907 * stmt.c (expand_case): Assert NCASES is greater than one.
23909 2017-06-29 Richard Biener <rguenther@suse.de>
23911 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
23913 (group_case_labels): Likewise.
23914 (find_taken_edge): Push sanity checking on val to workers...
23915 (find_taken_edge_cond_expr): ... here
23916 (find_taken_edge_switch_expr): ... and here, handle cases
23917 with just a default label.
23918 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
23919 (group_case_labels): Likewise.
23920 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
23921 group_case_labels does anything cleanup the CFG again.
23923 2017-06-29 Bin Cheng <bin.cheng@arm.com>
23925 PR tree-optimization/81196
23926 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
23927 exit condition comparing two IVs.
23929 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
23931 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
23932 profile to the dummy entry at the end of the list of architectures.
23933 * config/arm/arm-cpu-cdata.h: Regenerated.
23935 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23936 Michael Collison <michael.collison@arm.com>
23939 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
23941 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
23942 (*aarch64_reg_<mode>3_minus_mask): New pattern.
23943 (*aarch64_<optab>_reg_di3_mask2): New pattern.
23944 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
23945 of shift when the shift amount is masked with constant equal to
23946 the size of the mode.
23947 * config/aarch64/predicates.md (subreg_lowpart_operator): New
23950 2017-06-29 Martin Liska <mliska@suse.cz>
23952 * config/i386/i386.opt: Change range from [1,5] to [0,5].
23954 2017-06-29 Yury Gribov <tetra2005@gmail.com>
23957 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
23960 (edge_growth_cache_entry::edge_growth_cache_entry): New
23962 (reset_edge_growth_cache): Update to use constructor.
23964 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23966 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
23967 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23968 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23970 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
23972 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
23973 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
23975 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
23977 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
23978 (*-linux-uclibc*): Add t-uclibc tmake_file.
23979 * config/t-musl: New.
23980 * config/t-uclibc: New.
23982 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
23984 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
23986 (gen_comm_data): Emit architectural setting of arch_prof.
23987 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
23989 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
23990 (armv8-m.base, armv8-m.main): Likewise.
23991 * arm-protos.h (arm_build_target): Add profile field.
23992 (arch_option): Likewise.
23993 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
23995 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
23996 arm_active_target.profile.
23998 2017-06-28 Richard Biener <rguenther@suse.de>
24000 PR middle-end/81227
24001 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
24002 TYPE_OVERFLOW_WRAPS.
24003 * match.pd (negate_expr_p): Likewise.
24004 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
24005 fold_build2, not fold_binary.
24007 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
24009 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
24010 Convert memory address to Pmode.
24011 (aarch64_print_operand): Assert MEM operands are always Pmode.
24013 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
24016 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
24017 Remove redundant if.
24018 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
24019 * config/arm/aarch-common-protos.h
24020 (aarch_forward_to_shift_is_not_shifted_re): Remove.
24021 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
24023 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
24026 * multiple_target.c (create_dispatcher_calls): Set the default
24027 clone to be static, not public.
24029 2017-06-28 Richard Biener <rguenther@suse.de>
24031 * tree-vect-loop.c (vectorizable_reduction): Move special
24032 cond reduction IV var creation ...
24033 (vect_create_epilog_for_reduction): ... here. Remove induction_index
24034 parameter. Use STMT_VINFO_VECTYPE.
24035 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
24038 2017-06-28 Martin Liska <mliska@suse.cz>
24041 * ipa-visibility.c (non_local_p): Handle visibility.
24043 2017-06-28 Martin Liska <mliska@suse.cz>
24046 * common.opt: Add IntegerRange to various options.
24047 * opt-functions.awk (integer_range_info): New function.
24048 * optc-gen.awk: Add integer_range_info to cl_options struct.
24049 * opts-common.c (decode_cmdline_option): Handle
24050 CL_ERR_INT_RANGE_ARG.
24051 (cmdline_handle_error): Likewise.
24052 * opts.c (print_filtered_help): Show valid interval in
24053 when --help is provided.
24054 * opts.h (struct cl_option): Add range_min and range_max fields.
24055 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
24057 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
24059 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
24060 (x * C EQ/NE y * C): New transformation.
24062 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
24064 * genmultilib (combination_space): Accept '+' in option names.
24066 2017-06-28 Martin Liska <mliska@suse.cz>
24069 * asan.c (instrument_derefs): Bail out inner references
24070 that are hard register variables.
24072 2017-06-28 Jakub Jelinek <jakub@redhat.com>
24075 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
24076 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
24078 2017-06-28 Richard Biener <rguenther@suse.de>
24080 * tree-vectorizer.h (vect_get_vec_defs): Remove.
24081 (vect_get_slp_defs): Adjust.
24082 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
24084 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
24086 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
24087 get_initial_defs_for_reduction instead of vect_get_vec_defs.
24088 (vectorizable_reduction): Adjust.
24089 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
24091 (vect_get_slp_defs): Likewise.
24092 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
24093 (vectorizable_bswap): Adjust.
24094 (vectorizable_call): Likewise.
24095 (vectorizable_conversion): Likewise.
24096 (vectorizable_assignment): Likewise.
24097 (vectorizable_shift): Likewise.
24098 (vectorizable_operation): Likewise.
24099 (vectorizable_store): Likewise.
24100 (vectorizable_condition): Likewise.
24101 (vectorizable_comparison): Likewise.
24103 2017-06-28 Michael Collison <michael.collison@arm.com>
24106 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
24108 (arm_gen_movmemqi): Removed unused variable 'i'.
24109 Convert 'for' loop into 'while' loop.
24110 (arm_expand_prologue): Remove last unnecessary set of insn.
24111 (thumb_pop): Remove unused variable 'pushed_words'.
24112 (thumb_exit): Remove last unnecessary set of regs_to_pop.
24114 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24116 * config/s390/predicates.md: Use s390_rel_address_ok_p.
24117 * config/s390/s390-protos.h: Add prototype of
24118 s390_rel_address_ok_p.
24119 * config/s390/s390.c (s390_got_symbol): New function.
24120 (s390_rel_address_ok_p): New function.
24121 (legitimize_pic_address): Use s390_rel_address_ok_p.
24122 (s390_load_got): Use s390_got_symbol.
24123 (s390_option_override): Issue error if
24124 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
24125 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
24127 * config/s390/s390.opt: New option mpic-data-is-text-relative.
24129 2017-06-27 Andrew Pinski <apinski@cavium.com>
24131 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
24132 (X * copysign (1.0, X)): New pattern.
24133 (X * copysign (1.0, -X)): New pattern.
24134 (copysign (-1.0, CST)): New pattern.
24136 2017-06-27 Joseph Myers <joseph@codesourcery.com>
24138 * genmultilib (combination_space): Remove variable.
24139 Validate reuse rules against regular expression for any sequence
24140 of multilib options in any order.
24142 2017-06-27 Michael Collison <michael.collison@arm.com>
24144 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
24145 call aarch64_split_simd_combine.
24146 * (aarch64_combine_internal<mode>): Delete pattern.
24147 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
24148 Allow register and subreg operands.
24150 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24152 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
24153 specific need, just fallback on defaults.
24154 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
24156 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24157 Olivier Hainque <hainque@adacore.com>
24159 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
24161 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
24162 targets. Pick a default if no particular attempt applied.
24163 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
24166 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24168 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
24169 (x86_64-wrs-vxworks7): Likewise.
24171 2017-06-27 Marek Polacek <polacek@redhat.com>
24174 * ubsan.c (instrument_null): Check get_base_address's result for null.
24176 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
24178 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
24180 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
24182 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
24183 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
24184 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
24185 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
24186 New function types.
24187 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
24188 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
24189 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
24190 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
24191 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
24192 BUILT_IN_FEUPDATEENV): New builtins.
24193 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
24194 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
24195 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
24196 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
24198 (builtin_structptr_types): Adjust size.
24199 * tree.c (builtin_structptr_types): Add four entries.
24201 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24202 Olivier Hainque <hainque@adacore.com>
24204 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
24205 (TLS_SYM): New local macro, forcing reference to __tls__ on
24206 link command lines for VxWorks 7 RTPs, triggering initialization
24208 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
24209 OS features TLS support, true for RTPs on VxWorks 7.
24210 * config/vxworks.c (vxworks_override_options): Setup emutls
24213 2017-06-27 Jakub Jelinek <jakub@redhat.com>
24215 * predict.c (test_prediction_value_range): Use -1U instead of -1
24216 to avoid narrowing conversion warning.
24217 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
24218 to avoid narrowing conversion warning.
24219 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
24221 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
24223 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24225 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
24226 64bit configurations.
24227 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
24228 (SIZE_TYPE): Likewise.
24229 * config/vxworks.c (vxworks_emutls_var_fields): Use
24230 long_unsigned_type_node instead of unsigned_type_node as the offset
24231 field type, which is "pointer" mode in emutls.c.
24233 2017-06-27 Jakub Jelinek <jakub@redhat.com>
24236 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
24238 PR middle-end/81207
24239 * gimple-fold.c (replace_call_with_call_and_fold): Handle
24240 gimple_vuse copying separately from gimple_vdef copying.
24242 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24244 * value-prof.c (free_hist): Remove call to memset and the enclosing if
24247 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
24248 Olivier Hainque <hainque@adacore.com>
24250 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
24251 for all vxworks7 targets.
24252 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
24253 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
24254 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
24255 variations for VX6/VX7 and 32/64bits later on in ...
24256 (VXWORKS_LIB_SPEC): Leverage new macros.
24257 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
24258 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
24260 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
24262 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
24263 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
24265 2017-06-26 Carl Love <cel@us.ibm.com>
24267 * config/rs6000/rs6000-c.c: Add support for built-in functions
24268 vector bool char vec_reve (vector bool char);
24269 vector signed char vec_reve (vector signed char);
24270 vector unsigned char vec_reve (vector unsigned char);
24271 vector bool int vec_reve (vector bool int);
24272 vector signed int vec_reve (vector signed int);
24273 vector unsigned int vec_reve (vector unsigned int);
24274 vector bool long long vec_reve (vector bool long long);
24275 vector signed long long vec_reve (vector signed long long);
24276 vector unsigned long long vec_reve (vector unsigned long long);
24277 vector bool short vec_reve (vector bool short);
24278 vector signed short vec_reve (vector signed short);
24279 vector double vec_reve (vector double);
24280 vector float vec_reve (vector float);
24281 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
24282 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
24283 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
24284 (altivec_vreve): New pattern.
24285 * config/rs6000/altivec.h (vec_reve): New define.
24286 * doc/extend.texi (vec_rev): Update the built-in documentation file
24287 for the new built-in functions.
24289 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24291 PR tree-optimization/71815
24292 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
24294 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
24296 (slsr_process_phi): Likewise.
24297 (replace_uncond_cands_and_profitable_phis): Don't replace a
24298 multiply candidate with a stride of 1 (copy or cast).
24299 (phi_incr_cost): Call uses_consumed_by_stmt rather than
24301 (lowest_cost_path): Likewise.
24302 (total_savings): Likewise.
24304 2017-06-26 Richard Biener <rguenther@suse.de>
24307 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
24308 Use def_builtin_pure for all gather builtins.
24310 2017-06-26 Richard Biener <rguenther@suse.de>
24312 PR tree-optimization/81203
24313 * tree-tailcall.c (find_tail_calls): Do not move stmts into
24314 non-dominating BBs.
24316 2017-06-26 Marek Polacek <polacek@redhat.com>
24319 * doc/invoke.texi: Document -Wmultistatement-macros.
24321 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
24323 * doc/sourcebuild.texi (ARM-specific attributes): Document new
24324 arm_neon_ok_no_float_abi effective target.
24326 2017-06-26 Richard Biener <rguenther@suse.de>
24328 PR tree-optimization/80928
24329 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
24330 (copy_bbs): Set BB_DUPLICATED flag early.
24331 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
24333 (execute_on_shrinking_pred): Likewise.
24334 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
24335 BB_DUPLICATED blocks.
24336 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
24337 iterate over all PHIs considering removal of *gsi.
24339 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
24341 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
24344 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24346 * config/rs6000/rs6000-string.c: (expand_block_clear,
24347 do_load_for_compare, select_block_compare_mode,
24348 compute_current_alignment, expand_block_compare,
24349 expand_strncmp_align_check, expand_strn_compare,
24350 expand_block_move, rs6000_output_load_multiple)
24351 Move functions related to string/block move/compare
24352 to a separate file.
24353 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
24354 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
24355 for this function which is now used in two files.
24356 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
24357 * config.gcc: Add rs6000-string.o to extra_objs for
24358 targets powerpc*-*-* and rs6000*-*-*.
24360 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
24363 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
24364 32-bit, since indexed is not valid for DImode.
24365 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
24366 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
24367 (define_peephole2 for Altivec d-form load): Add 32-bit support.
24368 (define_peephole2 for Altivec d-form store): Likewise.
24371 * multiple_target.c (create_dispatcher_calls): Only create the
24372 dispatcher call if the function is the default clone of a
24373 versioned function.
24375 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
24377 PR middle-end/80902
24378 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
24379 a call, force the call to not be a tail call.
24381 2017-06-23 Jeff Law <law@redhat.com>
24383 * doc/contrib.texi: Add entry for Steven Pemberton's work on
24386 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
24388 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
24389 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
24390 handling for early expansion of vector shifts (sl,sr,sra,rl).
24391 (builtin_function_type): Add vector shift right instructions
24392 to the unsigned argument list.
24394 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
24396 rtl-optimizatoin/79286
24397 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
24398 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
24399 trap. PIC register plus a const unspec without offset can never trap.
24401 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
24403 * tree.h (builtin_structptr_type): New type.
24404 (builtin_structptr_types): Declare new array.
24405 * tree.c (builtin_structptr_types): New array.
24406 (free_lang_data, build_common_tree_nodes): Use it.
24408 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
24411 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
24414 2017-06-22 Matt Turner <mattst88@gmail.com>
24416 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
24417 Lake models to skylake case. Assume skylake for unknown
24418 models with clflushopt.
24420 2017-06-22 Jeff Law <law@redhat.com>
24422 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
24423 frame sizes that do not satisfy aarch64_uimm12_shift.
24425 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
24427 * profile-count.h (apply_probability,
24428 apply_scale, probability_in): Fix checks for zero.
24430 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24432 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
24433 * doc/cppdiropts.texi (-I @var{dir}): Document it.
24435 2016-06-22 Richard Biener <rguenther@suse.de>
24437 * tree-vect-loop.c (vect_model_reduction_cost): Handle
24438 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
24439 REDUC_MAX_EXPR support.
24440 (vectorizable_reduction): Likewise.
24441 (vect_create_epilog_for_reduction): Likewise.
24443 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
24445 * match.pd (A / (1 << B) -> A >> B): New.
24446 * generic-match-head.c: Include optabs-tree.h.
24447 * gimple-match-head.c: Likewise.
24448 * optabs-tree.h (target_supports_op_p): New.
24449 * optabs-tree.c (target_supports_op_p): New.
24451 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24453 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
24454 $gcc_cv_ld --help output.
24455 (gcc_cv_ld_demangle): Likewise.
24456 (gcc_cv_ld_eh_frame_hdr): Likewise.
24457 (gcc_cv_ld_pie): Likewise.
24458 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
24459 (gcc_cv_ld_buildid): Likewise.
24460 (gcc_cv_ld_sysroot): Likewise.
24461 (ld_bndplt_support): Likewise.
24462 (ld_pushpopstate_support): Likewise.
24463 * configure: Regenerate.
24464 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
24466 2017-06-21 Jakub Jelinek <jakub@redhat.com>
24469 * config/i386/sse.md (round<mode>2): Renumber match_dup and
24470 operands indexes to avoid gap between operands and match_dups.
24472 2017-06-21 Andrew Pinski <apinski@cavium.com>
24474 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
24475 Increment Arith_shift and Arith_shift_reg by 1.
24476 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
24478 * config/aarch64/aarch64.c (thunderx_tunings): Enable
24479 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
24480 (aarch64_strip_extend): Add new argument and test for it.
24481 (aarch64_cheap_mult_shift_p): New function.
24482 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
24483 add a cost if it is true.
24484 Update calls to aarch64_strip_extend.
24485 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
24487 2017-06-21 Andrew Pinski <apinski@cavium.com>
24489 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
24491 (thunderxt88): Likewise.
24492 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
24493 (thunderx_prefetch_tune): New variable.
24494 (thunderx2t99_prefetch_tune): Update for the correct values.
24495 (thunderxt88_tunings): New variable.
24496 (thunderx_tunings): Use thunderx_prefetch_tune instead of
24497 generic_prefetch_tune.
24498 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
24500 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24502 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
24503 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
24504 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
24505 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
24506 (aarch64_atomic_cas<mode>, GPI): Likewise.
24508 2017-06-21 Martin Liska <mliska@suse.cz>
24510 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
24511 statements on cold and hot labels.
24512 * predict.c (tree_estimate_probability_bb): Remove the
24513 prediction from this place.
24515 2017-06-21 Martin Liska <mliska@suse.cz>
24517 PR tree-optimization/79489
24518 * gimplify.c (maybe_add_early_return_predict_stmt): New
24520 (gimplify_return_expr): Call the function.
24521 * predict.c (tree_estimate_probability_bb): Remove handling
24523 * predict.def: Update comment about early return predictor.
24524 * gimple-predict.h (is_gimple_predict): New function.
24525 * predict.def: Change default value of early return to 66.
24526 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
24528 * passes.def: Put pass_strip_predict_hints to the beginning of
24531 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
24533 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
24534 FUNCTION_DECL declarations.
24535 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
24537 (dwaf2out_decl): Likewise.
24538 * godump.c (go_early_global_decl): Skip call to the real debug hook
24539 for FUNCTION_DECL declarations.
24540 * passes.c (rest_of_decl_compilation): Skip call to the
24541 early_global_decl debug hook for FUNCTION_DECL declarations, unless
24542 -fdump-go-spec is passed.
24544 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
24546 * config/i386/i386.c (struct builtin_isa): New field pure_p.
24547 Reorder for compactness.
24548 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
24549 (def_builtin_pure, def_builtin_pure2): New functions.
24550 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
24552 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
24554 * match.pd (nop_convert): New predicate.
24555 ((A +- CST1) +- CST2): Allow some NOP conversions.
24557 2017-06-21 Jakub Jelinek <jakub@redhat.com>
24560 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
24561 with ctors/dtors if GOVD_SHARED is set.
24563 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24565 * config/aarch64/aarch64.md (movti_aarch64):
24566 Emit mov rather than orr.
24567 (movtf_aarch64): Likewise.
24568 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
24569 Emit mov rather than orr.
24571 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24573 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
24574 Swap alternatives, make integer dup more expensive.
24576 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24578 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
24579 Return true for non-tls symbols.
24581 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
24583 * config/aarch64/aarch64-cores.def (cortex-a55): New.
24584 (cortex-a75): Likewise.
24585 (cortex-a75.cortex-a55): Likewise.
24586 * config/aarch64/aarch64-tune.md: Regenerate.
24587 * doc/invoke.texi (-mtune): Document new values for -mtune.
24589 2017-06-21 Tom de Vries <tom@codesourcery.com>
24591 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
24592 stack_size feature.
24593 (Effective-Target Keywords, Other attributes): Suggest using
24594 dg-add-options stack_size feature to get stack limit in stack_size
24595 effective target documentation.
24597 2017-06-21 Julian Brown <julian@codesourcery.com>
24598 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24600 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
24601 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
24602 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
24604 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
24605 attribute type list for neon_multiply.
24606 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
24607 attribute type list for neon_multiply.
24608 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
24609 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
24610 attribute type list for neon_multiply.
24611 * config/arm/types.md (crypto_pmull): Add.
24612 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
24613 attribute type list.
24615 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
24617 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
24620 2017-06-20 Jakub Jelinek <jakub@redhat.com>
24622 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
24623 to make sure not to dereference a NULL cost_classes_ptr pointer.
24625 2017-06-20 Carl Love <cel@us.ibm.com>
24627 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24628 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
24629 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
24630 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
24631 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
24632 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
24633 VMULOSW): New enum "unspec" values.
24634 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
24635 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
24636 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
24637 altivec_vmulosw): New patterns.
24638 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
24639 VMULOSW): Add definitions.
24641 2017-06-20 Julia Koval <julia.koval@intel.com>
24643 * config/i386/i386.c: Fix rounding expand for new pattern.
24644 * config/i386/subst.md: Fix pattern (parallel -> unspec).
24646 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24648 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
24649 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
24651 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24653 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
24656 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24658 * config/aarch64/aarch64-cores.def: Rearrange to sort by
24659 architecture, then by implementer ID.
24660 * config/aarch64/aarch64-tune.md: Regenerate.
24662 2017-06-20 Richard Biener <rguenther@suse.de>
24664 PR middle-end/81097
24665 * fold-const.c (split_tree): Fold to type before negating.
24667 2017-06-20 David Malcolm <dmalcolm@redhat.com>
24669 * diagnostic-show-locus.c
24670 (selftest::test_fixit_deletion_affecting_newline): New function.
24671 (selftest::diagnostic_show_locus_c_tests): Call it.
24673 2017-06-20 Andreas Schwab <schwab@suse.de>
24676 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
24679 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24681 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
24682 __ARM_FEATURE_COPROC according to support.
24684 2017-06-20 Jakub Jelinek <jakub@redhat.com>
24686 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
24687 Rewritten to avoid overflow for > 32-bit pointers.
24690 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
24691 by removing enum keyword.
24692 (ubsan_type_descriptor): Likewise. Formatting fix.
24695 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
24696 splitter): Require TARGET_SSE2 in the condition.
24698 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
24701 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
24702 for doing vector set of SFmode on ISA 3.0.
24703 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
24704 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
24706 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
24707 SFmode value into a V4SF variable that was extracted from another
24708 V4SF variable without converting the element to double precision
24709 and back to single precision vector format.
24710 (vsx_insert_extract_v4sf_p9_2): Likewise.
24712 2017-06-19 Jakub Jelinek <jakub@redhat.com>
24714 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
24715 in UWHI to avoid undefined overflow.
24718 * ubsan.h (enum ubsan_encode_value_phase): New.
24719 (ubsan_encode_value): Change second argument to
24720 enum ubsan_encode_value_phase with default value of
24721 UBSAN_ENCODE_VALUE_GENERIC.
24722 * ubsan.c (ubsan_encode_value): Change second argument to
24723 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
24724 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
24725 create_tmp_var_raw instead of create_tmp_var and use a
24727 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
24728 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
24729 ubsan_encode_value callers.
24732 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
24733 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
24734 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
24736 2017-06-19 Richard Biener <rguenther@suse.de>
24738 PR middle-end/81118
24739 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
24740 estimates if we changed anything.
24742 2017-06-19 Richard Biener <rguenther@suse.de>
24744 PR tree-optimization/80887
24745 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
24746 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
24747 simplified lookups, then reset mprts_hook.
24748 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
24750 (try_to_simplify): Likewise.
24752 2017-06-19 Martin Liska <mliska@suse.cz>
24755 * gimplify.c (gimplify_switch_expr):
24756 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
24758 2017-06-19 Martin Liska <mliska@suse.cz>
24760 * doc/install.texi: Document that PGO runs in 4 stages.
24762 2017-06-19 Martin Liska <mliska@suse.cz>
24765 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
24766 to dispatcher function name.
24767 * multiple_target.c (replace_function_decl): New function.
24768 (create_dispatcher_calls): Redirect both edges and references.
24770 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
24772 * profile-count.c (profile_count::dump): Dump quality.
24773 (profile_count::differs_from_p): Update for unsigned val.
24774 * profile-count.h (profile_count_quality): New enum.
24775 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
24777 2017-06-19 Richard Biener <rguenther@suse.de>
24779 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
24780 struct function as arg.
24781 (estimate_numbers_of_iterations): Export overload with loop arg.
24782 (free_numbers_of_iterations_estimates_loop): Use an overload of
24783 free_numbers_of_iterations_estimates instead.
24784 * tree-cfg.c (remove_bb): Adjust.
24785 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
24786 * tree-parloops.c (gen_parallel_loop): Likewise.
24787 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
24789 (tree_unroll_loops_completely): Likewise.
24790 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
24791 Use an overload instead and export.
24792 (estimated_loop_iterations): Adjust.
24793 (max_loop_iterations): Likewise.
24794 (likely_max_loop_iterations): Likewise.
24795 (estimate_numbers_of_iterations): Take struct function as arg
24797 (loop_exits_before_overflow): Adjust.
24798 (free_numbers_of_iterations_estimates_loop): Use an overload.
24799 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
24800 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
24802 2017-06-19 Richard Biener <rguenther@suse.de>
24805 * ipa-prop.c (find_constructor_constant_at_offset): Handle
24806 RANGE_EXPR conservatively.
24808 2017-06-16 Carl Love <cel@us.ibm.com>
24810 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24811 definitions for vec_float, vec_float2, vec_floato,
24812 vec_floate built-ins.
24813 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
24814 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
24816 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
24817 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
24818 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
24819 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
24820 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
24821 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
24822 vec_floato): Add builtin defines.
24823 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
24824 Update the built-in documentation file for the new built-in
24827 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24829 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
24830 (mthumb): Mark as the negative of -marm.
24832 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24834 * doc/invoke.texi (ARM Options, -mcpu): Document supported
24836 (ARM Options, -mtune): Document that this accepts the same
24837 extension options as -mcpu.
24838 (ARM Options, -mfpu): Document addition of -mfpu=auto.
24840 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24842 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
24843 permitted extensions.
24845 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24847 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
24848 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
24849 (armv8-m.main): Add option +nodsp.
24850 * config/arm/arm-cpu-cdata.h: Regenerated.
24852 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24854 * config/arm/t-fuchsia: New file.
24855 * config.gcc (arm*-*-fuchsia*): Use it.
24857 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24859 * config/arm/t-symbian: Rewrite for new option infrastructure.
24861 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24863 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
24864 (MULTILIB_REQUIRED): Likewise.
24866 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24868 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
24869 (MULTILIB_RESUE): Likewise.
24870 (MULTILIB_MATCHES): Likewise.
24871 (MULTLIB_REQUIRED): Likewise.
24873 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24875 * config/arm/t-rtems: Rewrite for new option framework.
24877 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24879 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
24880 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
24881 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
24882 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
24883 * config/arm/t-multilib: ... here.
24884 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
24885 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
24886 armv7-a and armv8*-a when A-profile libraries have not been built.
24887 * config/arm/t-rmprofile: Rewrite.
24889 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24891 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
24892 with a backslash. Remove the backslash after substituting unescaped
24894 * doc/fragments.texi (MULTILIB_REUSE): Document it.
24896 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24898 * config.gcc: (arm*-*-*): When building a-profile libraries, force
24899 the driver to pass through the default setting of -mfloat-abi.
24900 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
24902 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
24903 (all_feat_combs): New rule.
24904 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
24906 * config/arm/t-aprofile: Rewrite.
24908 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24910 * config/arm/arm.h (FPUTYPE_AUTO): Define.
24911 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
24912 fpu is not specified by the user/command-line.
24913 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
24914 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
24915 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
24916 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
24917 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
24918 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
24920 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24922 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
24923 * config/arm/t-arm-elf: Rewritten.
24925 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24927 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
24928 have some floating-point instructions.
24929 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
24930 (TARGET_MAYBE_HARD_FLOAT): New macro.
24931 * config/arm/arm-builtins.c (arm_init_builtins): Use
24932 TARGET_MAYBE_HARD_FLOAT.
24933 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
24935 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24937 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
24938 (configargs.h): Include it.
24939 (arm_print_hint_for_fpu_option): New function.
24940 (arm_parse_fpu_option): New function.
24941 (candidate_extension): New class.
24942 (arm_canon_for_multilib): New function.
24943 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
24944 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
24945 (ARCH_CANONICAL_SPECS): New macro.
24946 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
24948 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24950 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
24951 are set after handling multilib fragments. Set target_cpu_default2
24954 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24956 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
24958 (arm*-*-*): Set target_cpu_default2 to a quoted string.
24959 * config/arm/parsecpu.awk (check_cpu): Validate any extension
24961 (check_arch): Likewise.
24962 * config/arm/arm.c (arm_configure_build_target): Handle
24963 TARGET_CPU_DEFAULT being a string constant. Scan any feature
24964 options in the default.
24966 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24968 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
24969 when an option is an alias of another.
24970 * config/arm/parsecpu.awk (optalias): New parser token.
24971 (gen_comm_data): Mark non-alias options as such. Emit entries
24972 for extension aliases.
24973 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
24974 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
24975 (armv6kz, armv6zk, armv6t2): Likewise.
24976 (armv7): Make vfpv3-d16 an alias.
24977 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
24979 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
24980 Sort in canonical order.
24981 (armv8-a): Sort in canonical order.
24982 (armv8.1-a, armv8.2-a): Likewise.
24983 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
24985 (cortex-a9): Sort in canonical order.
24986 * config/arm/arm.c (selftests.h): Include it.
24987 (arm_test_cpu_arch_data): New function.
24988 (arm_run_self_tests): New function.
24989 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
24990 (targetm): Move declaration to the end of the file.
24991 * arm-cpu-cdata.h: Regenerated.
24993 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24995 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
24996 call to target_mode_check describing the type of option passed.
24997 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
24998 (arm_target_thumb_only): Use arm_parse_arch_option_name or
24999 arm_parse_cpu_option_name to match parameters against list of
25001 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
25002 arm_arch_core_flags data structure.
25003 * config/arm/arm-cpu_cdata.h: Regenerated.
25005 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25007 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
25009 (arm_print_hint_for_cpu_option): Likewise.
25010 (arm_print_hint_for_arch_option): Likewise.
25011 (arm_parse_cpu_option_name): Likewise.
25012 (arm_parse_arch_option_name): Likewise.
25013 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
25014 of entries in the all_fpus list.
25015 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
25016 (arm_parse_cpu_option_name): Declare.
25017 (arm_parse_arch_option_name): Declare.
25018 (arm_parse_option_features): Declare.
25019 (arm_intialize_isa): Declare.
25020 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
25022 (gen_comm_data): ... here. Make definitions non-static.
25023 * config/arm/arm-cpu-data.h: Regenerated.
25024 * config/arm/arm-cpu-cdata.h: Regenerated.
25026 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25028 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
25029 (cpu_arch_extension): New structure.
25030 (cpu_arch_option, arch_option, cpu_option): New structures.
25031 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
25032 architecture types.
25033 (gen_data): Generate new format data tables.
25034 * config/arm/arm.c (cpu_tune): New structure.
25035 (cpu_option, processors): Delete.
25036 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
25037 (arm_print_hint_for_cpu_option): ... this and ...
25038 (arm_print_hint_for_arch_option): ... this.
25039 (arm_parse_arch_cpu_name): Delete. Replace with ...
25040 (arm_parse_cpu_option_name): ... this and ...
25041 (arm_parse_arch_option_name): ... this.
25042 (arm_unrecognized_feature): Change type of target parameter to
25044 (arm_parse_arch_cpu_features): Delete. Replace with ...
25045 (arm_parse_option_features): ... this.
25046 (arm_configure_build_target): Rework to use new configuration data
25048 (arm_print_tune_info): Rework for new configuration data tables.
25049 * config/arm/arm-cpu-data.h: Regenerated.
25050 * config/arm/arm-cpu.h: Regenerated.
25052 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25054 * Makefile.in (OBJS): Move sbitmap.o from here ...
25055 (OBJS-libcommon): ... to here.
25057 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25059 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
25060 (ISA_ALL_CRYPTO): New macro.
25061 (ISA_ALL_SIMD): New macro
25062 (ISA_ALL_FP): New macro.
25063 * config/arm/arm.c (fpu_bitlist): Update initializer.
25064 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
25066 (arm9e): Add fpu. Add option for nofp
25067 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
25068 (arm926ej-s, arm1026ej-s): Likewise.
25069 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
25070 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
25071 neon-fp16, neon-vfpv4, nofp and nosimd.
25072 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
25073 (cortex-a8): Add fpu. Add option for nofp.
25074 (cortex-a9): Add fpu. Add options for nosimd and nofp.
25075 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
25076 (cortex-r4f): Add fpu.
25077 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
25078 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
25080 (cortex-r8): Likewise.
25081 (cortex-m4): Add fpu. Add option for nofp.
25082 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
25083 (cortex-a17.cortex-a7): Likewise.
25084 (cortex-a32): Add fpu. Add options for crypto and nofp.
25085 (cortex-a35, cortex-a53): Likewise.
25086 (cortex-a57): Add fpu. Add option for crypto.
25087 (cortex-a72, cortex-a73): Likewise.
25088 (exynos-m1): Likewise.
25089 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
25090 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
25091 (cortex-m33): Add fpu. Add option for nofp.
25092 * config/arm/arm-cpu-cdata.h: Regenerated
25093 * config/arm/arm-cpu-data.h: Regenerated.
25095 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25097 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
25098 (armv5te, armv5tej): Likewise.
25099 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
25100 (armv7): Add options fp and vfpv3-d16.
25101 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
25102 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
25104 (armv7ve): Likewise.
25105 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
25106 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
25107 (armv8-a): Add nocrypto option.
25108 (armv8.1-a, armv8.2-a): Likewise.
25109 (armv8-m.main): add options fp, fp.dp and nofp.
25111 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25113 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
25115 (armv8-a+crc): Delete.
25116 (armv8.1-a): Add options simd, crypto and nofp.
25117 (armv8.2-a): Add options fp16, simd, crypto and nofp.
25118 (armv8.2-a+fp16): Delete.
25119 (armv8-m.main): Add option dsp.
25120 (armv8-m.main+dsp): Delete.
25121 (cortex-a8): Add fpu. Add nofp option.
25122 (cortex-a9): Add fpu. Add nofp and nosimd options.
25123 * config/arm/parsecpu.awk (gen_data): Generate option tables and
25124 link to main cpu and architecture data structures.
25125 (gen_comm_data): Only put isa attributes from the main architecture
25127 (option): New statement for architecture and CPU entries.
25128 * arm.c (struct cpu_option): New structure.
25129 (struct processors): Add entry for options.
25130 (arm_unrecognized_feature): New function.
25131 (arm_parse_arch_cpu_name): Ignore any characters after the first
25133 (arm_parse_arch_cpu_feature): New function.
25134 (arm_configure_build_target): Separate out any CPU and architecture
25135 features and parse separately. Don't error out if -mfpu=auto is
25136 used with only an architecture string.
25137 (arm_print_asm_arch_directives): New function.
25138 (arm_file_start): Call it.
25139 * config/arm/arm-cpu-cdata.h: Regenerated.
25140 * config/arm/arm-cpu-data.h: Likewise.
25141 * config/arm/arm-tables.opt: Likewise.
25143 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25145 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
25146 assembler when it is not -mfpu=auto.
25148 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25150 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
25151 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
25152 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
25153 (ASM_CPU_SPEC): Rewrite.
25154 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
25155 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
25156 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
25157 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
25158 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
25159 copied string is NUL-terminated. Also strip any characters prefixed
25161 (arm_rewrite_selected_arch): New function.
25162 (arm_rewrite_march): New function.
25164 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25166 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
25167 (x_arm_cpu_string, x_arm_tune_string): Likewise.
25168 (march, mcpu, mtune): Convert to string-based options.
25169 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
25170 (arm_parse_arch_cpu_name): New function.
25171 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
25172 identify selected architecture or CPU.
25173 (arm_option_save): New function.
25174 (TARGET_OPTION_SAVE): Redefine.
25175 (arm_option_restore): Restore string options.
25176 (arm_option_print): Print string options.
25178 2017-06-16 Martin Sebor <msebor@redhat.com>
25180 PR tree-optimization/80933
25181 PR tree-optimization/80934
25182 * builtins.c (fold_builtin_3): Do not handle bcmp here.
25183 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
25184 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
25185 (gimple_fold_builtin): Call them.
25187 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25189 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
25190 as unlikely; update profile.
25192 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25194 * predict.c (force_edge_cold): Handle declaring edges impossible
25197 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25199 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
25201 (try_unroll_loop_completely): Fix reporting.
25203 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25205 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
25207 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
25210 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
25211 if given a non-constant argument for an intrinsic which requires a
25214 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25216 * profile.c (compare_freqs): New function.
25217 (branch_prob): Sort edge list.
25218 (find_spanning_tree): Assume that the list is priority sorted.
25220 2017-06-16 Richard Biener <rguenther@suse.de>
25222 PR tree-optimization/81090
25223 * passes.def (pass_record_bounds): Remove.
25224 * tree-pass.h (make_pass_record_bounds): Likewise.
25225 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
25226 make_pass_record_bounds): Likewise.
25227 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
25228 not free niter estimates at the beginning but at the end.
25229 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
25231 2017-06-16 Richard Biener <rguenther@suse.de>
25233 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
25234 initializer to workaround ICE in host GCC 4.8.
25236 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25238 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
25240 (clone_inlined_nodes): Update.
25242 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25244 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
25245 prefetch settings, and enable prefetching by default at -O3.
25247 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25249 * config/aarch64/aarch64.c (aarch64_override_options_internal):
25250 Set flag_prefetch_loop_arrays according to tuning data.
25252 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25254 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
25255 New tune structure.
25256 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
25257 [Unrelated to main purpose of the patch] Place the pointer field last
25258 to enable type checking errors when tune structure are wrongly merged.
25259 * config/aarch64/aarch64.c (generic_prefetch_tune,)
25260 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
25261 (thunderx2t99_prefetch_tune): New tune constants.
25262 (tune_params *_tunings): Update all tunings (no functional change).
25263 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
25264 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
25265 from tunings structures.
25267 2017-06-16 Jakub Jelinek <jakub@redhat.com>
25270 * ubsan.c (instrument_null): Add T argument, use it instead
25271 of computing it based on IS_LHS.
25272 (instrument_object_size): Likewise.
25273 (pass_ubsan::execute): Adjust instrument_null and
25274 instrument_object_size callers to pass gimple_get_lhs or
25275 gimple_assign_rhs1 result to it. Use instrument_null instead of
25276 calling get_base_address and instrument_mem_ref. Handle
25277 aggregate call arguments for object-size sanitization.
25279 2017-06-16 Yury Gribov <tetra2005@gmail.com>
25281 PR tree-optimization/81089
25282 * tree-vrp.c (is_masked_range_test): Validate operands of
25285 2017-06-15 Martin Sebor <msebor@redhat.com>
25288 * dumpfile.c (dump_register): Avoid calling memset to initialize
25289 a class with a default ctor.
25290 * gcc.c (struct compiler): Remove const qualification.
25291 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
25292 * hash-table.h: Ditto.
25293 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
25295 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
25296 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
25298 * params.h (struct param_info): Make struct members non-const.
25299 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
25300 with default initialization.
25301 * vec.h (vec_copy_construct, vec_default_construct): New helper
25303 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
25304 with vec_copy_construct.
25305 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
25306 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
25307 * doc/invoke.texi (-Wclass-memaccess): Document.
25309 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25311 * emit-rtl.h (is_leaf): Update comment about local
25312 register allocator.
25314 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
25317 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
25318 for a variable to have a section before checking if the section has a
25320 Set section to.persistent if persistent attribute is set.
25321 Warn if .persistent attribute is used on an automatic variable.
25323 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
25325 PR rtl-optimization/80474
25326 * reorg.c (update_block): Do not ignore instructions in a delay slot.
25328 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
25330 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
25333 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
25335 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
25336 (casesi): Emit bounds checking as RTL.
25337 (casesi_internal_mips16_<mode>): Remove bounds checking.
25339 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
25341 * config/xtensa/xtensa.c (xtensa_option_override): Append
25342 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
25343 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
25344 xtensa_doloop_hooks): Define unconditionally.
25345 (xtensa_reorg_loops): Only call reorg_loops in the presence of
25347 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
25348 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
25349 for it in xtensa_option_override.
25350 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
25351 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
25353 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
25355 * doc/cppopts.texi: Document '-' special value to -MF.
25357 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
25359 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
25360 (cortex_a53_fconst): Likewise.
25361 (cortex_a53_fpmul): Likewise.
25362 (cortex_a53_f_load_64): Likewise.
25363 (cortex_a53_f_load_many): Likewise.
25364 (cortex_a53_advsimd_alu): Likewise.
25365 (cortex_a53_advsimd_alu_q): Likewise.
25366 (cortex_a53_advsimd_mul): Likewise.
25367 (cortex_a53_advsimd_mul_q): Likewise.
25368 (fpmac bypass): Add new bypass for fpmac-fpmac case.
25369 Add missing fmul, r2f_cvt and fconst cases.
25371 2017-06-14 Richard Biener <rguenther@suse.de>
25373 PR middle-end/81088
25374 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
25376 (fold_binary_loc): When associating do not treat pre-existing
25377 TREE_OVERFLOW on literal constants as a reason to allow
25378 TREE_OVERFLOW on associated literal constants.
25380 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
25382 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
25383 (MASK_FEATURES): New macro.
25384 * config/sparc/sparc.c (sparc_option_override): Remove the special
25385 handling of -mfpu and generalize it to all MASK_FEATURES switches.
25387 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
25389 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
25390 a division of 0 if non-call exceptions are enabled.
25392 2017-06-14 Andrew Pinski <apinski@cavium.com>
25393 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25396 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
25397 Improve vector initialization code gen for only variable case.
25399 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
25401 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
25403 2017-06-14 Richard Biener <rguenther@suse.de>
25405 PR tree-optimization/81083
25406 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
25409 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25411 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
25412 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
25413 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
25414 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
25415 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
25416 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
25418 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25420 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
25421 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
25423 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25425 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
25427 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25429 * config/rs6000/t-rtems: Don't handle SPE.
25431 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25433 * config/rs6000/t-linux: Don't handle SPE.
25435 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25437 * config/rs6000/eabispe.h: Delete file.
25439 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25441 * config/rs6000/t-spe: Delete file.
25443 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25445 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
25446 (rs6000_legitimate_offset_address_p): Return false for anything in
25447 V2SImode or V2SFmode.
25449 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25451 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
25452 except V2SF and V2SI. Rearrange the vector modes, and add comments.
25453 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
25455 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
25456 (rs6000_legitimate_offset_address_p): Ditto.
25457 (rs6000_emit_move): Ditto.
25458 (rs6000_init_builtins): Remove V4HI_type_node.
25460 2017-06-13 Martin Liska <mliska@suse.cz>
25463 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
25464 (gate_asan): Likewise.
25465 * asan.h (asan_no_sanitize_address_p): Remove the function.
25466 (sanitize_flags_p): New function.
25467 * builtins.def: Fix coding style.
25468 * common.opt: Use renamed enum value.
25469 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
25470 * doc/extend.texi: Document no_sanitize attribute.
25471 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
25472 to SANITIZE_UNDEFINED_NONDEFAULT.
25473 * gcc.c (sanitize_spec_function): Use the renamed enum value.
25474 * gimple-fold.c (optimize_atomic_compare_exchange_p):
25475 Use sanitize_flags_p.
25476 * gimplify.c (gimplify_function_tree): Likewise.
25477 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
25478 * opts.c (parse_no_sanitize_attribute): New function.
25479 (common_handle_option): Use renamed enum value.
25480 * opts.h (parse_no_sanitize_attribute): Declare.
25481 * tree.c (sanitize_flags_p): New function.
25482 * tree.h: Declared here.
25483 * tsan.c: Use sanitize_flags_p.
25484 * ubsan.c (ubsan_expand_null_ifn): Likewise.
25485 (instrument_mem_ref): Likewise.
25486 (instrument_bool_enum_load): Likewise.
25487 (do_ubsan_in_current_function): Remove the function.
25488 (pass_ubsan::execute): Use sanitize_flags_p.
25489 * ubsan.h: Remove do_ubsan_in_current_function
25490 * tree-cfg.c (print_no_sanitize_attr_value): New function.
25491 (dump_function_to_file): Use it here.
25493 2017-06-13 Martin Jambor <mjambor@suse.cz>
25495 PR tree-optimization/80803
25496 PR tree-optimization/81063
25497 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
25498 (propagate_subaccesses_across_link): Enqueue subtree whenever
25499 necessary instead of relying on the caller.
25501 2017-06-13 Martin Jambor <mjambor@suse.cz>
25503 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
25504 that have a first_link.
25505 (sort_and_splice_var_accesses): Do not check first_link before
25507 (subtree_mark_written_and_enqueue): Likewise.
25508 (propagate_all_subaccesses): Likewise and do not stop at first
25509 parent with a first_link.
25511 2017-06-13 Martin Jambor <mjambor@suse.cz>
25513 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
25516 2017-06-13 Yury Gribov <tetra2005@gmail.com>
25518 * match.pd: New pattern.
25520 2017-06-13 Yury Gribov <tetra2005@gmail.com>
25522 * tree-vrp.c (is_masked_range_test): New function.
25523 (register_edge_assert_for): Determine ranges for
25526 2017-06-13 Yury Gribov <tetra2005@gmail.com>
25528 PR tree-optimization/67328
25529 * fold-const.c (maskable_range_p): New function.
25530 (build_range_check): Generate bittests if possible.
25532 2017-06-13 Martin Liska <mliska@suse.cz>
25534 * gimple-pretty-print.c (dump_probability): Add new argument.
25535 (dump_edge_probability): Dump both probability and count.
25536 (dump_gimple_label): Likewise.
25537 (dump_gimple_bb_header): Likewise.
25539 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
25542 * config/avr/avr-devices.c: Fix indentation.
25543 * config/avr/gen-avr-mmcu-specs.c: Dito.
25545 2017-06-13 Richard Biener <rguenther@suse.de>
25547 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
25548 instead get vector type from stmt_info.
25549 (vectorizable_reduction): Adjust. Remove dead code.
25551 2017-06-13 Richard Biener <rguenther@suse.de>
25553 PR middle-end/81065
25554 * fold-const.c (extract_muldiv_1): Remove bogus distribution
25555 case of C * (x * C2 + C3).
25556 (fold_addr_of_array_ref_difference): Properly fold index difference.
25558 2017-06-12 David S. Miller <davem@davemloft.net>
25561 * config/sparc/sparc.md (return expander): Emit frame blockage if
25562 function uses alloca.
25564 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
25566 * combine.c (make_field_assignment): Check len rather than the mode
25567 precision when calling force_to_mode.
25569 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
25571 Support multilibs and devices that see flash in RAM address range.
25574 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
25575 (avr_mcu_t) <flash_pm_offset>: New field.
25576 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
25577 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
25578 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
25579 (AVR_TINY_PM_OFFSET): Remove macro.
25580 * config/avr/avr.opt (-mshort-calls): New option.
25581 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
25582 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
25583 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
25584 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
25585 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
25586 instead of avr_arch->have_jmp_call.
25587 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
25588 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
25589 avr_arch->flash_pm_offset to define.
25590 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
25591 new field flash_pm_offset. Add entry for avrxmega3.
25592 (avr_texinfo): Add entry for avrxmega3.
25593 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
25594 attiny212, attiny214,
25595 attiny412, attiny414, attiny416, attiny417,
25596 attiny814, attiny816, attiny817,
25597 attiny1614, attiny1616, attiny1617,
25598 attiny3214, attiny3216, attiny3217.
25599 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
25600 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
25601 (avr_print_operand_address) [AVR_TINY]: Same.
25602 (avr_asm_init_sections) <readonly_data_section>: Only patch
25603 callback if avr_arch->flash_pm_offset = 0.
25604 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
25605 for rodata if avr_arch->flash_pm_offset != 0.
25606 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
25607 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
25608 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
25609 (m_options): Append opt_rcall.
25610 (m_dirnames): Append dir_rcall.
25611 * config/avr/t-multilib: Regenerate.
25613 * configure.ac [target=avr]: Check whether avrxmega3 default
25614 linker description file works as needed.
25615 * configure: Regenerate.
25616 * doc/avr-mmcu.texi: Regenerate.
25617 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
25618 <__AVR_ARCH__>: Document avrxmega3 and 103.
25619 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
25620 <__AVR_SHORT_CALLS__>: Document it.
25621 <__AVR_PM_BASE_ADDRESS__>: Document it.
25622 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
25623 (AVR Variable Attributes) <progmem>: Document this is
25624 not needed for avrxmega3.
25625 (AVR Named Address Spaces) <__flash>: Dito.
25627 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
25629 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
25631 2017-06-12 Doug Rupp <rupp@adacore.com>
25633 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
25634 Append vxworks-stdint.h to the tm_file list.
25635 * config/vxworks-stdint.h: New file.
25637 2017-06-12 Martin Liska <mliska@suse.cz>
25639 PR tree-optimization/81041
25640 * tree-profile.c (gimple_gen_ic_func_profiler):
25641 Create an extra BB in profile-generate
25642 (gimple_gen_time_profiler): Likewise.
25644 2017-06-12 Jakub Jelinek <jakub@redhat.com>
25646 PR tree-optimization/81003
25647 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
25648 (update_range_test): Use it instead of force_gimple_operand_gsi.
25650 2017-06-12 Richard Biener <rguenther@suse.de>
25652 PR tree-optimization/81053
25653 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
25654 with backedge value not defined in loop. Simplify def stmt
25657 2017-06-11 Tom de Vries <tom@codesourcery.com>
25660 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
25662 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
25663 nvptx_cannot_force_const_mem.
25665 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25667 * opts.c (finish_options): Move test for flag_split_stack after
25668 it has been initialized.
25670 2017-06-11 Jason Merrill <jason@redhat.com>
25672 * tree.h (id_equal): New.
25673 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
25674 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
25675 instead of strcmp of IDENTIFIER_POINTER.
25677 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25679 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
25680 (mark_all_inlined_calls_cdtor): Fix formating.
25681 (inline_transform): Rescale profile before inlining.
25683 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25685 * cgraph.h (cgraph_edge::clone): Update prototype.
25686 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
25687 (cgraph_node::create_clone): Update.
25688 (cgraph_node::create_version_clone): Update.
25689 * tree-inline.c (copy_bb): Update.
25690 (expand_call_inline): Update.
25692 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25694 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
25695 factored out from ...
25696 (rs6000_emit_prologue): ... here.
25698 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25700 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
25701 factored out from ...
25702 (rs6000_emit_prologue): ... here.
25704 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25706 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
25708 (handle_missing_profiles): Fix computation of tp_first_run.
25709 (counts_to_freqs): Do not touch freqs when count is 0.
25711 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25713 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
25716 2017-06-10 Tom de Vries <tom@codesourcery.com>
25718 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
25719 attributes): Document signal effective target.
25721 2017-06-10 Tom de Vries <tom@codesourcery.com>
25723 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
25724 Document effective target stack_size.
25726 2017-06-09 David Malcolm <dmalcolm@redhat.com>
25728 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
25729 to the edit_context if they can be auto-applied.
25731 2017-06-9 Ian Lance Taylor <iant@golang.org>
25733 * opts.c (finish_options): If -fsplit-stack, disable implicit
25734 -forder-blocks-and-partition.
25735 * doc/invoke.texi (Optimize Options): Document that when using
25736 -fsplit-stack -forder-blocks-and-partition is not implicitly
25739 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25741 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
25742 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
25743 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
25744 * builtins.def (abort, trap, unreachable): Declare cold.
25745 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
25746 * tree-core.h (ECF_COLD): New.
25747 * tree.c (set_call_expr_flags): Handle ECF_COLD.
25748 (build_common_builtin_nodes): Mark unreachable and abort as cold.
25750 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25752 * predict.c (unlikely_executed_stmt_p): Cleanup.
25754 2017-06-09 Richard Biener <rguenther@suse.de>
25756 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
25757 model if the ref is always written to.
25759 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25761 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
25763 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25765 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
25768 2017-06-09 Tom de Vries <tom@codesourcery.com>
25771 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
25772 "target cannot support label values" when encountering LABEL_REF.
25774 2017-06-09 Martin Liska <mliska@suse.cz>
25776 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
25777 (gimple_gen_ic_func_profiler): Emit direct comparison
25778 of __gcov_indirect_call_callee with NULL.
25779 (gimple_gen_time_profiler): Change probability from
25780 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
25782 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25784 * profile.c (edge_gcov_counts): Turn to pointer.
25785 (compute_branch_probabilities, compute_branch_probabilities): Update.
25786 (branch_prob): Do not clear edge_gcov_count.
25787 * profile.h (edge_gcov_counts): Turn to pointer.
25788 (edge_gcov_count): Update.
25790 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25792 * gimple.h (gimple_check_failed): Mark cold.
25794 2017-06-09 Richard Biener <rguenther@suse.de>
25796 PR tree-optimization/66623
25797 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
25798 refactor check_reduction into two parts, properly computing
25799 whether we have to check reduction validity for outer loop
25802 2017-06-09 Richard Biener <rguenther@suse.de>
25804 PR tree-optimization/79483
25805 * graphite-scop-detection.c (order): New global.
25806 (get_order): Compute bb to order mapping that satisfies code
25807 generation constraints.
25808 (cmp_pbbs): New helper.
25809 (build_scops): Start domwalk at entry block, sort generated
25812 2017-06-09 Richard Biener <rguenther@suse.de>
25814 PR middle-end/81007
25815 * ipa-polymorphic-call.c
25816 (ipa_polymorphic_call_context::restrict_to_inner_class):
25817 Skip FIELD_DECLs with error_mark_node type.
25818 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
25821 2017-06-09 Martin Liska <mliska@suse.cz>
25823 * predict.c (struct branch_predictor): New struct.
25824 (test_prediction_value_range): New test.
25825 (predict_c_tests): New function.
25826 * selftest-run-tests.c (selftest::run_tests): Run the function.
25827 * selftest.h: Declare new tests.
25829 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
25832 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
25833 gen_add3_insn did not fail.
25834 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
25835 r0, construct that number in a temporary reg and add that reg to r0.
25836 If asked to put the result in r0 as well, fail.
25838 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
25840 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25841 for early expansion of vec_eqv.
25843 2017-06-08 Jakub Jelinek <jakub@redhat.com>
25845 PR middle-end/81005
25846 * ubsan.c (instrument_null): Avoid pointless code temporary.
25847 (pass_ubsan::execute): Instrument aggregate arguments of calls.
25849 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
25853 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
25856 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
25859 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25861 * predict.c (unlikely_executed_edge_p): Move ahead.
25862 (probably_never_executed_edge_p): Use it.
25864 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25866 PR middle-end/79988
25867 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
25868 gimple_call_builtin_p call.
25870 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25872 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
25873 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
25874 rtl_check_failed_type2, rtl_check_failed_code1,
25875 rtl_check_failed_code2, rtl_check_failed_code_mode,
25876 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
25877 rtvec_check_failed_bounds, rtl_check_failed_flag,
25878 _fatal_insn_not_found, _fatal_insn): Likewise.
25879 * tree.h (tree_contains_struct_check_failed,
25880 tree_check_failed, tree_not_check_failed,
25881 tree_class_check_failed, tree_range_check_failed,
25882 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
25883 tree_vec_elt_check_failed, phi_node_elt_check_failed,
25884 tree_operand_check_failed, omp_clause_check_failed,
25885 omp_clause_operand_check_failed, omp_clause_range_check_failed):
25888 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25890 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
25891 flag_branch_probabilities.
25892 * ipa-inline.c (edge_badness): Likewise.
25893 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
25894 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
25895 * predict.c (maybe_hot_frequency_p): Likewise.
25896 (probably_never_executed): Likewise.
25897 * sched-ebb.c (schedule_ebbs): Likewise.
25898 * sched-rgn.c (find_single_block_region): Likewise.
25899 * tracer.c (tail_duplicate): Likewise.
25901 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25903 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
25904 longer requires x_flag_profile_use.
25906 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25908 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
25909 instead of flag_reorder_blocks_and_partition.
25910 * dbxout.c (dbxout_function_end): Likewise.
25911 * dwarf2out.c (gen_subprogram_die): Likewise.
25912 * haifa-sched.c (sched_create_recovery_edges): Likewise.
25913 * hw-doloop.c (reorg_loops): Likewise.
25914 * varasm.c (assemble_start_function,
25915 assemble_end_function): Likewise.
25916 (decide_function_section): Do not check for
25917 flag_reorder_blocks_and_partition.
25919 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25921 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
25923 (chkp_get_hard_register_fake_addr_expr): Ditto.
25924 (chkp_build_addr_expr): Add check for hard reg case.
25925 (chkp_parse_array_and_component_ref): Ditto.
25926 (chkp_find_bounds_1): Ditto.
25927 (chkp_process_stmt): Don't generate bounds store for
25930 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25932 * predict.c (maybe_hot_bb_p): Do not check profile status.
25933 (maybe_hot_edge_p): Likewise.
25934 (probably_never_executed): Check for zero counts even if profile
25936 (unlikely_executed_edge_p): New function.
25937 (unlikely_executed_stmt_p): New function.
25938 (unlikely_executed_bb_p): New function.
25939 (set_even_probabilities): Use unlikely predicates.
25940 (combine_predictions_for_bb): Likewise.
25941 (predict_paths_for_bb): Likewise.
25942 (predict_paths_leading_to_edge): Likewise.
25943 (determine_unlikely_bbs): New function.
25944 (estimate_bb_frequencies): Use it.
25945 (compute_function_frequency): Use zero counts even if profile is
25947 * profile-count.h: Fix typo.
25949 2017-08-08 Julia Koval <julia.koval@intel.com>
25951 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
25952 _mm512_mask_cvtsepi16_storeu_epi8,
25953 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
25954 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
25955 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
25956 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
25957 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
25958 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
25959 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
25960 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
25961 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
25962 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
25963 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
25964 __builtin_ia32_pmovuswb256mem_mask,
25965 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
25966 __builtin_ia32_pmovwb512mem_mask): New builtins.
25968 2017-08-08 Julia Koval <julia.koval@intel.com>
25970 PR target/73350,80862
25971 * config/i386/subst.md (round): Fix round pattern.
25972 * config/i386/i386.c (ix86_erase_embedded_rounding):
25973 Fix erasing rounding for the fixed pattern.
25975 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25977 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
25979 2017-06-08 Martin Liska <mliska@suse.cz>
25981 PR gcov-profile/80911
25982 * gcov.c (block_info::block_info): New constructor.
25984 2017-06-07 Carl Love <cel@us.ibm.com>
25986 * config/rs6000/rs6000-c: The return type of the following
25987 built-in functions was implemented as int not long long. Fix sign
25988 of return value for the unsigned version of vec_mulo and vec_mule.
25989 vector unsigned long long vec_bperm (vector unsigned long long,
25990 vector unsigned char)
25991 vector signed long long vec_mule (vector signed int,
25993 vector unsigned long long vec_mule (vector unsigned int,
25994 vector unsigned int)
25995 vector signed long long vec_mulo (vector signed int,
25997 vector unsigned long long vec_mulo (vector unsigned int,
25998 vector unsigned int)
25999 * doc/extend.texi: Fix the documentation for the built-in
26002 2017-06-07 Carl Love <cel@us.ibm.com>
26005 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
26008 2017-06-07 Carl Love <cel@us.ibm.com>
26010 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
26011 support, Generate doublehv for signed int/float for BE case only.
26013 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
26015 * doc/invoke.texi (mcx16): Rewrite.
26017 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26019 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
26020 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
26021 *mov<mode>_softfloat, and an anonymous splitter): Use
26022 nonimmediate_operand instead of rs6000_nonimmediate_operand.
26024 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26026 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
26028 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
26029 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
26030 (rs6000_debug_reg_global): Adjust.
26031 (rs6000_init_hard_regno_mode_ok): Adjust.
26032 (rs6000_dbx_register_number): Adjust.
26033 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
26034 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
26035 Remove SPE_ACC and SPEFSCR.
26036 (REG_ALLOC_ORDER): Ditto.
26037 (FRAME_POINTER_REGNUM): Change to 111.
26038 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
26039 (REG_CLASS_NAMES): Ditto.
26040 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
26041 (REGISTER_NAMES): Ditto.
26042 (ADDITIONAL_REG_NAMES): Ditto.
26043 (rs6000_reg_names): Ditto.
26044 * config/rs6000/rs6000.md: Renumber some register number
26047 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26049 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
26051 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
26052 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
26054 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
26055 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
26056 Delete the SPE high registers.
26057 (REG_ALLOC_ORDER): Ditto.
26058 (enum reg_class): Remove SPE_HIGH_REGS.
26059 (REG_CLASS_NAMES): Ditto.
26060 (REG_CLASS_CONTENTS): Delete the SPE high registers.
26061 (REGISTER_NAMES): Ditto.
26062 (rs6000_reg_names): Ditto.
26063 * doc/tm.texi.in: Remove SPE as example.
26064 * doc/tm.texi: Regenerate.
26066 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26068 * config/rs6000/8540.md (ppc8540_brinc): Delete.
26069 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
26070 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
26071 * config/rs6000/rs6000.md (type): Remove "brinc".
26073 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26075 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
26076 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
26077 * config/rs6000/linuxspe.h: Delete file.
26078 * config/rs6000/rs6000.md: Don't include spe.md.
26079 * config/rs6000/spe.h: Delete file.
26080 * config/rs6000/spe.md: Delete file.
26081 * config/rs6000/t-rs6000: Remove spe.md.
26083 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26085 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
26086 (reg_or_none500mem_operand): Delete.
26087 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
26088 instead of reg_or_none500mem_operand.
26090 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26092 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26093 handling of SPE flags.
26094 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
26096 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26098 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
26100 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
26101 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
26102 paired_divv2sf3): Similar.
26103 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
26104 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
26105 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
26107 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
26108 Rename the paired_* instruction patterns.
26109 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
26111 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
26112 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
26113 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
26114 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
26115 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
26116 PAIRED_VECTOR_MODE.
26117 (struct machine_function): Delete field spe_insn_chain_scanned_p.
26118 (spe_func_has_64bit_regs_p): Delete.
26119 (spe_expand_predicate_builtin): Delete.
26120 (spe_expand_evsel_builtin): Delete.
26121 (TARGET_DWARF_REGISTER_SPAN): Do not define.
26122 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
26123 (invalid_e500_subreg): Delete.
26124 (rs6000_legitimize_address): Always force_reg op2 as well, for
26125 paired single memory accesses.
26126 (rs6000_member_type_forces_blk): Delete.
26127 (rs6000_spe_function_arg): Delete.
26128 (rs6000_expand_unop_builtin): Delete SPE handling.
26129 (rs6000_expand_binop_builtin): Ditto.
26130 (spe_expand_stv_builtin): Delete.
26131 (bdesc_2arg_spe): Delete.
26132 (spe_expand_builtin): Delete.
26133 (spe_expand_predicate_builtin): Delete.
26134 (spe_expand_evsel_builtin): Delete.
26135 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
26136 (spe_init_builtins): Delete.
26137 (spe_func_has_64bit_regs_p): Delete.
26138 (savres_routine_name): Delete "info" parameter. Adjust callers.
26139 (rs6000_emit_stack_reset): Ditto.
26140 (rs6000_dwarf_register_span): Delete.
26141 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
26142 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
26143 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
26145 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
26147 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
26148 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
26149 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
26150 mulv2sf3, divv2sf3): Delete expanders.
26152 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26154 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
26156 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26158 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
26159 * config/rs6000/rs6000.c: Ditto.
26161 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26163 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
26164 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
26165 comparison_operator.
26167 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26169 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
26170 * config/rs6000/rs6000.opt: Ditto.
26171 * config/rs6000/t-rtems: Ditto.
26173 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26175 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
26176 TARGET_E500_SINGLE by 0, simplify.
26177 * config/rs6000/rs6000.c: Ditto.
26178 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
26179 (spe_build_register_parallel): Delete.
26180 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
26181 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
26182 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
26183 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
26184 (E500_CONVERT): Delete.
26185 * config/rs6000/spe.md: Remove many patterns and all define_constants.
26187 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26189 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
26190 * config/rs6000/dfp.md: Ditto.
26191 (negdd2, *negdd2_fpr): Merge.
26192 (absdd2, *absdd2_fpr): Merge.
26193 (negtd2, *negtd2_fpr): Merge.
26194 (abstd2, *abstd2_fpr): Merge.
26195 * config/rs6000/e500.h: Delete file.
26196 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
26197 TARGET_FPRS by 1 and simplify.
26198 * config/rs6000/rs6000-c.c: Ditto.
26199 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
26200 TARGET_DF_SPE by 0.
26201 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
26203 * config/rs6000/rs6000.md: Ditto.
26204 (floatdidf2, *floatdidf2_fpr): Merge.
26205 (move_from_CR_gt_bit): Delete.
26206 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
26207 (E500_CR_IOR_COMPARE): Delete.
26208 (All patterns that require !TARGET_FPRS): Delete.
26209 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
26211 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26213 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
26215 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26217 * graphds.c (add_edge): Intitialize edge's attached data.
26218 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
26219 pointer parameter. Call pointed function on each edge during
26220 graph traversing. Skip traversing the edge when the function
26222 (graphds_dfs, graphds_scc): Ditto.
26223 (for_each_edge): New parameter. Pass the new parameter to callback
26225 * graphds.h (skip_edge_callback): New function pointer type.
26226 (graphds_dfs, graphds_scc): New function pointer parameter.
26227 (graphds_edge_callback, for_each_edge): New parameter.
26229 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26231 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
26232 out code checking if runtime alias check is possible to below ...
26233 Call the new function.
26234 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
26235 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
26237 2017-06-07 Marek Polacek <polacek@redhat.com>
26240 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
26241 TYPE_OVERFLOW_WRAPS check.
26243 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26245 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
26246 if versioning is required.
26247 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
26248 peeling with the check for versioning.
26250 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26252 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
26253 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
26254 Set true to new parameter if new ssa variable is defined.
26255 (vect_gen_vector_loop_niters): Refactor. Set range information
26256 for the new vector loop bound variable.
26257 (vect_do_peeling): Ditto.
26259 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26261 * tree-affine.c (ssa.h): Include header file.
26262 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
26263 has wrapping overflow behavior.
26265 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26267 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
26269 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26271 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
26272 (tree_to_aff_combination): ... here.
26274 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26276 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
26277 reg_pressure model function.
26278 (ivopts_global_cost_for_size): Delete.
26279 (determine_set_costs, iv_ca_recount_cost): Call new model function
26280 ivopts_estimate_reg_pressure.
26282 2017-06-07 Tamar Christina <tamar.christina@arm.com>
26284 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
26285 expensive than udiv. Remove floating point cases from mod.
26287 2017-06-07 Tamar Christina <tamar.christina@arm.com>
26289 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
26290 Increase idiv cost.
26292 2017-06-07 Tamar Christina <tamar.christina@arm.com>
26294 * config/aarch64/aarch64.md
26295 (copysignsf3): Fix mask generation.
26297 2017-06-07 Jakub Jelinek <jakub@redhat.com>
26299 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
26301 (class dump_manager): Add register_dumps method.
26302 * dumpfile.c: Include langhooks.h.
26303 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
26304 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
26305 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
26306 (dump_manager::dump_register): Start with 512 entries instead of 32.
26307 (dump_manager::register_dumps): New method.
26308 * toplev.c (general_init): Instead of invoking register_dumps
26309 langhook, invoke register_dumps method on the dump manager.
26310 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
26313 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
26315 * doc/md.texi: Clarify the restrictions on a define_insn condition.
26316 Say that # requires an associated define_split to exist, and that
26317 the define_split must be suitable for use after register allocation.
26319 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26321 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
26322 (compute_outgoing_frequencies): Also initialize zero counts.
26323 (find_many_sub_basic_blocks): Do not produce uninitialized profile
26324 around loops; preserve more of profile when nothing changes.
26326 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
26328 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
26330 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
26331 * config/arm/arm-cpu-cdata.h: Regenerate.
26332 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
26333 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
26334 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
26335 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
26336 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
26338 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
26340 * config/arm/t-rmprofile: Likewise.
26341 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
26343 2017-06-06 David S. Miller <davem@davemloft.net>
26346 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
26347 blockage if function uses alloca.
26349 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26351 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
26352 New "uid" fields to hold pretty-print IDs of group and ref.
26353 Memory references are now identified as <group_id>:<ref_id>
26354 instead of using [random] addresses.
26355 (dump_mem_details): Simplify, no functional change.
26356 (dump_mem_ref): Simplify and make output more concise.
26357 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
26358 (find_or_create_group): Initialize group uid.
26359 (record_ref): Initialize ref uid. Improve debug output.
26360 (prune_group_by_reuse, should_issue_prefetch_p,)
26361 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
26362 (mark_nontemporal_store, determine_loop_nest_reuse):
26363 Improve debug output.
26365 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26367 * dbgcnt.def (prefetch): New debug counter.
26368 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
26369 (schedule_prefetches): Stop issueing prefetches if debug counter
26372 2017-06-06 Tom de Vries <tom@codesourcery.com>
26374 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
26375 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
26377 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26379 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
26380 Use aarch64_reg_or_zero predicate for operand 4.
26381 (aarch64_compare_and_swap<mode> define_insn_and_split):
26382 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
26383 (aarch64_store_exclusive<mode>): Likewise for operand 2.
26385 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
26387 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
26388 (arm_compute_save_core_reg_mask): This.
26389 (thumb1_compute_save_reg_mask): Rename into ...
26390 (thumb1_compute_save_core_reg_mask): This.
26391 (arm_compute_save_reg0_reg12_mask): Adapt comment.
26392 (arm_compute_frame_layout): Likewise.
26394 2017-06-06 Richard Biener <rguenther@suse.de>
26396 PR tree-optimization/80974
26397 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
26398 keep or clear leaders SSA info.
26400 2017-06-06 Tom de Vries <tom@codesourcery.com>
26402 * config/nvptx/nvptx.c (split_mode_p): New function.
26403 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
26405 2017-06-06 Tom de Vries <tom@codesourcery.com>
26407 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
26409 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26412 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
26415 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26417 * shrink-wrap.c (handle_simple_exit): Update profile.
26418 (try_shrink_wrapping): Upate profile.
26420 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26422 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
26423 (tree_guess_outgoing_edge_probabilities): New.
26424 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
26425 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
26427 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26429 * ipa-split.c (split_function): Initialize return bb profile.
26431 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26433 * profile.c (compute_branch_probabilities): Also initialize
26434 EXIT_BLOCK profile.
26436 2017-06-06 Richard Biener <rguenther@suse.de>
26438 PR tree-optimization/80928
26439 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
26440 (vect_analyze_loop_operations): Properly guard analysis for
26442 (vect_transform_loop): Likewise.
26443 (vect_analyze_loop_2): Also reset SLP type on PHIs.
26444 (vect_model_induction_cost): Do not cost for pure SLP.
26445 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
26446 of induction in inner loop vectorization.
26447 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
26448 (vect_get_and_check_slp_defs): Handle vect_induction_def.
26449 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
26451 (vect_analyze_slp_cost_1): Cost induction.
26452 (vect_detect_hybrid_slp_stmts): Handle PHIs.
26453 (vect_get_slp_vect_defs): Likewise.
26454 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
26455 (vect_transform_stmt): Handle SLP reductions.
26456 * tree-vectorizer.h (vectorizable_induction): Adjust.
26458 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
26460 * config/rs6000/rs6000.c (make_resolver_func): Update
26461 init_lowered_empty_function call.
26463 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
26465 * doc/invoke.texi: Document the -fprofile-abs-path option.
26466 * common.opt (fprofile-abs-path): New option.
26467 * gcov-io.h (gcov_write_filename): Declare.
26468 * gcov-io.c (gcov_write_filename): New function.
26469 * coverage.c (coverage_begin_function): Use gcov_write_filename.
26470 * profile.c (output_location): Likewise.
26472 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26474 * shring-wrap.c: Revert accidental commit.
26476 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
26478 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
26480 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26482 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
26484 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
26485 profile in callgraph edge.
26486 * profile-count.h (apply_probability): If THIS is 0, then result is 0
26487 (apply_scale): Likewise.
26488 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
26489 Also scale profile when inlining function with zero profile.
26490 (initialize_cfun): Update exit block profile even when it is zero.
26491 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
26492 when profile is read.
26494 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
26496 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
26497 (CLONE_*): New constants to define the processors we can generate
26498 code for with the target_clone attribute.
26499 (rs6000_clone_map): New array to identify which clone processors
26500 the current program is running on.
26501 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
26502 target_clone attribute.
26503 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
26504 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
26505 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
26506 (cpu_expand_builtin): Add support for target_clone attribute.
26507 (rs6000_valid_attribute_p): Allow "default" attribute.
26508 (get_decl_name): New debug function to simplify printing the
26509 current function name in debugging statements.
26510 (rs6000_clone_priority): New functions to support the target_clone
26511 attribute, and be able to generate code to switch between ISA 2.05
26512 through ISA 3.0 (power6 through power9).
26513 (rs6000_compare_version_priority): Likewise.
26514 (rs6000_get_function_versions_dispatcher): Likewise.
26515 (make_resolver_func): Likewise.
26516 (add_condition_to_bb): Likewise.
26517 (dispatch_function_versions): Likewise.
26518 (rs6000_generate_version_dispatcher_body): Likewise.
26519 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
26520 (fusion_gpr_load_p): Fix a spacing issue.
26521 * doc/extend.texi (Common Function Attributes): Document that the
26522 PowerPC supports the target_clone attribute.
26524 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
26526 * config/arm/arm.h: explain F symbol found in description of ARM
26527 register allocation in its legend.
26529 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26531 * config/mips/frame-header-opt.c: Include profile-count.h.
26532 * config/riscv/riscv.c: Include profile-count.h
26534 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26536 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
26538 (sm_set_flag_if_changed): Add bbs field.
26539 (execute_sm_if_changed_flag_set): Pass BBS.
26540 (execute_sm): Update.
26542 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26544 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
26547 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26549 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
26550 (peephole2): New peephole2 to emit the above.
26551 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
26553 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26555 * config/aarch64/aarch64.c (define_peephole2 above
26556 *sub_<shift>_<mode>): New peephole.
26558 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26560 * config/i386/i386.c (make_resolver_func): Update.
26561 * Makefile.in: Add profile-count.h and profile-count.o
26562 * auto-profile.c (afdo_indirect_call): Update to new API.
26563 (afdo_set_bb_count): Update.
26564 (afdo_propagate_edge): Update.
26565 (afdo_propagate_circuit): Update.
26566 (afdo_calculate_branch_prob): Update.
26567 (afdo_annotate_cfg): Update.
26568 * basic-block.h: Include profile-count.h
26569 (struct edge_def): Turn count to profile_count.
26570 (struct basic_block_def): Likewie.
26571 (REG_BR_PROB_BASE): Move to profile-count.h
26572 (RDIV): Move to profile-count.h
26573 * bb-reorder.c (max_entry_count): Turn to profile_count.
26574 (find_traces): Update.
26575 (rotate_loop):Update.
26576 (connect_traces):Update.
26577 (sanitize_hot_paths):Update.
26578 * bt-load.c (migrate_btr_defs): Update.
26579 * cfg.c (RDIV): Remove.
26580 (init_flow): Use alloc_block.
26581 (alloc_block): Uninitialize count.
26582 (unchecked_make_edge): Uninitialize count.
26583 (check_bb_profile): Update.
26584 (dump_edge_info): Update.
26585 (dump_bb_info): Update.
26586 (update_bb_profile_for_threading): Update.
26587 (scale_bbs_frequencies_int): Update.
26588 (scale_bbs_frequencies_gcov_type): Update.
26589 (scale_bbs_frequencies_profile_count): New.
26590 * cfg.h (update_bb_profile_for_threading): Update.
26591 (scale_bbs_frequencies_profile_count): Declare.
26592 * cfgbuild.c (compute_outgoing_frequencies): Update.
26593 (find_many_sub_basic_blocks): Update.
26594 * cfgcleanup.c (try_forward_edges): Update.
26595 (try_crossjump_to_edge): Update.
26596 * cfgexpand.c (expand_gimple_tailcall): Update.
26597 (construct_exit_block): Update.
26598 * cfghooks.c (verify_flow_info): Update.
26599 (dump_bb_for_graph): Update.
26600 (split_edge): Update.
26601 (make_forwarder_block): Update.
26602 (duplicate_block): Update.
26603 (account_profile_record): Update.
26604 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
26605 (get_estimated_loop_iterations): Update.
26606 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
26607 (single_likely_exit): Update.
26608 * cfgloopmanip.c (scale_loop_profile): Update.
26610 (set_zero_probability): Update.
26611 (lv_adjust_loop_entry_edge): Update.
26612 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
26613 (purge_dead_edges): Update.
26614 (rtl_account_profile_record): Update.
26615 * cgraph.c (cgraph_node::create): Uninitialize count.
26616 (symbol_table::create_edge): Uninitialize count.
26617 (cgraph_update_edges_for_call_stmt_node): Update.
26618 (cgraph_edge::dump_edge_flags): Update.
26619 (cgraph_node::dump): Update.
26620 (cgraph_edge::maybe_hot_p): Update.
26621 * cgraph.h: Include profile-count.h
26622 (create_clone), create_edge, create_indirect_edge): Update.
26623 (cgraph_node): Turn count to profile_count.
26624 (cgraph_edge0: Likewise.
26625 (make_speculative, clone): Update.
26626 (create_edge): Update.
26627 (init_lowered_empty_function): Update.
26628 * cgraphclones.c (cgraph_edge::clone): Update.
26629 (duplicate_thunk_for_node): Update.
26630 (cgraph_node::create_clone): Update.
26631 * cgraphunit.c (cgraph_node::analyze): Update.
26632 (cgraph_node::expand_thunk): Update.
26633 * final.c (dump_basic_block_info): Update.
26634 * gimple-streamer-in.c (input_bb): Update.
26635 * gimple-streamer-out.c (output_bb): Update.
26636 * graphite.c (print_global_statistics): Update.
26637 (print_graphite_scop_statistics): Update.
26638 * hsa-brig.c: Include basic-block.h.
26639 * hsa-dump.c: Include basic-block.h.
26640 * hsa-gen.c (T sum_slice): Update.
26641 (convert_switch_statements):Update.
26642 * hsa-regalloc.c: Include basic-block.h.
26643 * ipa-chkp.c (chkp_produce_thunks): Update.
26644 * ipa-cp.c (struct caller_statistics): Update.
26645 (init_caller_stats): Update.
26646 (gather_caller_stats): Update.
26647 (ipcp_cloning_candidate_p): Update.
26648 (good_cloning_opportunity_p): Update.
26649 (get_info_about_necessary_edges): Update.
26650 (dump_profile_updates): Update.
26651 (update_profiling_info): Update.
26652 (update_specialized_profile): Update.
26653 (perhaps_add_new_callers): Update.
26654 (decide_about_value): Update.
26655 (ipa_cp_c_finalize): Update.
26656 * ipa-devirt.c (struct odr_type_warn_count): Update.
26657 (struct decl_warn_count): Update.
26658 (struct final_warning_record): Update.
26659 (possible_polymorphic_call_targets): Update.
26660 (ipa_devirt): Update.
26661 * ipa-fnsummary.c (redirect_to_unreachable): Update.
26662 * ipa-icf.c (sem_function::merge): Update.
26663 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
26664 * ipa-inline.c (compute_uninlined_call_time): Update.
26665 (compute_inlined_call_time): Update.
26666 (want_inline_small_function_p): Update.
26667 (want_inline_self_recursive_call_p): Update.
26668 (edge_badness): Update.
26669 (lookup_recursive_calls): Update.
26670 (recursive_inlining): Update.
26671 (inline_small_functions): Update.
26672 (dump_overall_stats): Update.
26673 (dump_inline_stats): Update.
26674 * ipa-profile.c (ipa_profile_generate_summary): Update.
26675 (ipa_propagate_frequency): Update.
26676 (ipa_profile): Update.
26677 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26678 * ipa-utils.c (ipa_merge_profiles): Update.
26679 * loop-doloop.c (doloop_modify): Update.
26680 * loop-unroll.c (report_unroll): Update.
26681 (unroll_loop_runtime_iterations): Update.
26682 * lto-cgraph.c (lto_output_edge): Update.
26683 (lto_output_node): Update.
26684 (input_node): Update.
26685 (input_edge): Update.
26686 (merge_profile_summaries): Update.
26687 * lto-streamer-in.c (input_cfg): Update.
26688 * lto-streamer-out.c (output_cfg): Update.
26689 * mcf.c (create_fixup_graph): Update.
26690 (adjust_cfg_counts): Update.
26691 (sum_edge_counts): Update.
26692 * modulo-sched.c (sms_schedule): Update.
26693 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
26694 * predict.c (maybe_hot_count_p): Update.
26695 (probably_never_executed): Update.
26696 (dump_prediction): Update.
26697 (combine_predictions_for_bb): Update.
26698 (propagate_freq): Update.
26699 (handle_missing_profiles): Update.
26700 (counts_to_freqs): Update.
26701 (rebuild_frequencies): Update.
26702 (force_edge_cold): Update.
26703 * predict.h: Include profile-count.h
26704 (maybe_hot_count_p, counts_to_freqs): UPdate.
26705 * print-rtl-function.c: Do not include cfg.h
26706 * print-rtl.c: Include basic-block.h
26707 * profile-count.c: New file.
26708 * profile-count.h: New file.
26709 * profile.c (is_edge_inconsistent): Update.
26710 (correct_negative_edge_counts): Update.
26711 (is_inconsistent): Update.
26712 (set_bb_counts): Update.
26713 (read_profile_edge_counts): Update.
26714 (compute_frequency_overlap): Update.
26715 (compute_branch_probabilities): Update; Initialize and deinitialize
26717 (branch_prob): Update.
26718 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
26719 (edge_gcov_count): New.
26720 (bb_gcov_count): New.
26721 * shrink-wrap.c (try_shrink_wrapping): Update.
26722 * tracer.c (better_p): Update.
26723 * trans-mem.c (expand_transaction): Update.
26724 (ipa_tm_insert_irr_call): Update.
26725 (ipa_tm_insert_gettmclone_call): Update.
26726 * tree-call-cdce.c: Update.
26727 * tree-cfg.c (gimple_duplicate_sese_region): Update.
26728 (gimple_duplicate_sese_tail): Update.
26729 (gimple_account_profile_record): Update.
26730 (execute_fixup_cfg): Update.
26731 * tree-inline.c (copy_bb): Update.
26732 (copy_edges_for_bb): Update.
26733 (initialize_cfun): Update.
26734 (freqs_to_counts): Update.
26735 (copy_cfg_body): Update.
26736 (expand_call_inline): Update.
26737 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26738 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
26739 (try_unroll_loop_completely): Update.
26740 (try_peel_loop): Update.
26741 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26742 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
26743 * tree-ssa-loop-split.c (connect_loops): Update.
26744 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
26745 * tree-ssa-reassoc.c (branch_fixup): Update.
26746 * tree-ssa-tail-merge.c (replace_block_by): Update.
26747 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
26748 (compute_path_counts): Update.
26749 (update_profile): Update.
26750 (recompute_probabilities): Update.
26751 (update_joiner_offpath_counts): Update.
26752 (estimated_freqs_path): Update.
26753 (freqs_to_counts_path): Update.
26754 (clear_counts_path): Update.
26755 (ssa_fix_duplicate_block_edges): Update.
26756 (duplicate_thread_path): Update.
26757 * tree-switch-conversion.c (case_bit_test_cmp): Update.
26758 (struct switch_conv_info): Update.
26759 * tree-tailcall.c (decrease_profile): Update.
26760 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26761 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26762 * value-prof.c (check_counter): Update.
26763 (gimple_divmod_fixed_value): Update.
26764 (gimple_mod_pow2): Update.
26765 (gimple_mod_subtract): Update.
26766 (gimple_ic_transform): Update.
26767 (gimple_stringop_fixed_value): Update.
26768 * value-prof.h (gimple_ic): Update.
26770 2017-06-02 Carl Love <cel@us.ibm.com>
26772 * config/rs6000/rs6000-c: Add support for built-in functions
26773 vector double vec_doublee (vector signed int);
26774 vector double vec_doublee (vector unsigned int);
26775 vector double vec_doublee (vector float);
26776 vector double vec_doubleh (vector signed int);
26777 vector double vec_doubleh (vector unsigned int);
26778 vector double vec_doubleh (vector float);
26779 vector double vec_doublel (vector signed int);
26780 vector double vec_doublel (vector unsigned int);
26781 vector double vec_doublel (vector float);
26782 vector double vec_doubleo (vector signed int);
26783 vector double vec_doubleo (vector unsigned int);
26784 vector double vec_doubleo (vector float);.
26785 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
26786 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
26788 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
26789 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
26790 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
26792 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
26793 vec_doublel, vec_doubleh.
26794 * doc/extend.texi: Update the built-in documentation file for the
26795 new built-in functions.
26797 2017-06-02 David Malcolm <dmalcolm@redhat.com>
26800 * ipa-inline-analysis.c (free_growth_caches): Set
26801 edge_removal_hook_holder to NULL after removing it.
26803 2017-06-02 Sudakshina Das <sudi.das@arm.com>
26805 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
26806 comparision with zero.
26808 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26809 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26810 for early expansion of vec_min and vec_max builtins.
26811 (builtin_function_type): Add min/max unsigned variants to those
26812 identified as having unsigned arguments.
26814 2017-06-02 Olivier Hainque <hainque@adacore.com>
26816 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
26818 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26820 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
26821 Use VALL_F16 iterator rather than VALL.
26823 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26825 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
26826 Emit CBNZ inside loop when doing a strong exchange and comparing
26827 against zero. Generate the CC flags after the loop.
26829 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
26831 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
26832 (dl_section_ref): New.
26833 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
26834 On AIX, append an expression to subtract the size of the
26835 section length to dl_section_ref.
26837 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26839 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26840 for early expansion of vector absolute builtins.
26842 2017-06-02 Richard Biener <rguenther@suse.de>
26844 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
26845 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
26847 2017-06-02 Richard Biener <rguenther@suse.de>
26849 PR tree-optimization/80948
26850 * tree-tailcall.c (find_tail_calls): Track stmts to move in
26851 stmt order as well.
26853 2017-06-02 Richard Biener <rguenther@suse.de>
26855 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
26857 * tree-vect-stmts.c (process_use): Do not mark backedge defs
26858 for inductions as relevant.
26860 2017-06-02 Richard Biener <rguenther@suse.de>
26862 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
26863 (vectorizable_induction): ... this. Remove dead code.
26865 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
26867 * builtins. (expand_builtin_alloca): Remove second parameter and
26868 infer its value from the first parameter instead.
26869 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
26871 2017-06-02 Jakub Jelinek <jakub@redhat.com>
26873 PR rtl-optimization/80903
26874 * loop-doloop.c (add_test): Unshare sequence.
26876 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
26878 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
26880 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
26882 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
26884 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
26885 xlogue_layout::get_instance, logue_layout::xlogue_layout,
26886 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
26887 (xlogue_layout::get_stub_rtx): Make static.
26888 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
26889 (xlogue_layout::compute_stub_managed_regs): Rename to...
26890 (xlogue_layout::count_stub_managed_regs): ...this.
26891 (xlogue_layout::is_stub_managed_reg): New function.
26892 (xlogue_layout::m_stub_names): Rename to...
26893 (xlogue_layout::s_stub_names): ...this, make static.
26894 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
26895 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
26896 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
26897 xlogue_layout::s_stub_names): Instantiate statics.
26898 (stub_managed_regs): Remove.
26899 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
26900 (disable_call_ms2sysv_xlogues): Rename to...
26901 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
26902 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
26904 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
26905 change after reload_completed.
26906 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
26908 (ix86_expand_prologue): Likewise.
26909 (ix86_expand_epilogue): Likewise.
26910 (ix86_expand_split_stack_prologue): Likewise.
26911 (ix86_compute_frame_layout): Remove frame parameter ...
26912 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
26913 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
26915 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
26916 (ix86_frame): Move from here ...
26917 * config/i386/i386.h (ix86_frame): ... to here.
26918 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
26919 complete ix86_frame data structure instead. Remove some_ld_name.
26921 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26923 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
26924 symbols that hold a DECL_VALUE_EXPR.
26926 2017-06-01 Martin Jambor <mjambor@suse.cz>
26928 PR tree-optimization/80898
26929 * tree-sra.c (process_subtree_disqualification): Removed.
26930 (disqualify_candidate): Do not acll
26931 process_subtree_disqualification.
26932 (subtree_mark_written_and_enqueue): New function.
26933 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
26934 RHS has been disqualified and re-queue LHS if necessary. Apart
26935 from that, ignore disqualified RHS.
26937 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26939 * config/s390/s390.c (s390_emit_epilogue): Disable early return
26940 address fetch for z10 or later.
26942 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26944 * config/arc/arc.md (tst_movb): Add guard when splitting.
26946 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26948 * config/arc/arc.c (arc_can_eliminate): Test against
26949 arc_frame_pointer_needed.
26951 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26953 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
26954 to prevent store reordering.
26955 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
26956 (type): Add block type.
26957 (stack_tie): Define special instruction to be used in
26960 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26962 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
26963 constraint. It is not valid for the pattern.
26964 (noncommutative_binary_comparison): Likewise.
26966 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26968 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
26971 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26973 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
26974 be used by the reg-alloc.
26976 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26978 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
26979 reg-alloc when having mul64 or mul32x16 instructions.
26980 (mulsidi3): Likewise.
26981 (umulsidi3): Likewise.
26982 (mulsi32x16): New pattern.
26983 (mulsi64): Likewise.
26984 (mulsidi64): Likewise.
26985 (umulsidi64): Likewise.
26986 (MUL32x16_REG): Define.
26987 (mul64_600): Use MUL32x16_REG.
26988 (mac64_600): Likewise.
26989 (umul64_600): Likewise.
26990 (umac64_600): Likewise.
26992 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26994 * config/arc/arc.md (mulsi3_700): Make it commutative.
26996 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
26998 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
27000 (*sign_extendsidi2_insn): Likewise for movstosw.
27002 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
27004 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
27005 the type of the input discriminant value. Convert the
27006 discriminant value of signedness vary.
27008 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
27010 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
27011 Add to -Wall section.
27013 2017-06-01 Richard Biener <rguenther@suse.de>
27015 PR middle-end/66313
27016 * fold-const.c (fold_plusminus_mult_expr): If the factored
27017 factor may be zero use a wrapping type for the inner operation.
27018 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
27019 and handle moved defs.
27020 (process_assignment): Properly guard the unary op case. Return a
27021 tri-state indicating that moving the stmt before the call may allow
27022 to continue. Pass through to_move.
27023 (find_tail_calls): Handle moving unrelated defs before
27026 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
27029 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
27030 splitter result in the canonical way.
27032 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
27034 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
27035 also for 32bit target. Update insn attributes.
27036 (zero-extendsidi2 splitter): Allow all registers for operand 1.
27038 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
27040 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
27041 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
27042 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
27043 (_mm_maskz_min_ss): New intrinsics.
27045 2017-05-31 Martin Liska <mliska@suse.cz>
27047 * tree-vect-loop.c (vect_create_epilog_for_reduction):
27048 Change comment style to one we normally use.
27049 (vectorizable_reduction): Likewise.
27050 (vectorizable_induction): Likewise.
27051 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
27052 (vectorizable_call): Likewise.
27053 (vectorizable_simd_clone_call): Likewise.
27054 (vectorizable_conversion): Likewise.
27055 (vectorizable_assignment): Likewise.
27056 (vectorizable_shift): Likewise.
27057 (vectorizable_operation): Likewise.
27058 (vectorizable_store): Likewise.
27059 (vectorizable_load): Likewise.
27060 * tree-vectorizer.h: Likewise.
27062 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
27064 * passes.c (emergency_dump_function): New.
27065 * tree-pass.h (emergency_dump_function): Declare.
27066 * plugin.c (plugins_internal_error_function): Remove.
27067 * plugin.h (plugins_internal_error_function): Remove declaration.
27068 * toplev.c (internal_error_function): New static function. Use it...
27069 (general_init): ...here.
27071 2017-05-31 Graham Markall <graham.markall@embecosm.com>
27073 * config/arc/arc.c (arc_print_operand): Handle constant operands.
27074 (arc_rtx_costs): Add costs for new patterns.
27075 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
27076 * config/arc/predicates.md: Add _1_2_3_operand predicate.
27078 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
27080 * tree-ssa-strlen.c (get_next_strinfo): New function.
27081 (get_stridx_plus_constant): Use it.
27082 (zero_length_string): Likewise.
27083 (adjust_related_strinfos): Likewise.
27084 (adjust_last_stmt): Likewise.
27086 2017-05-31 Richard Biener <rguenther@suse.de>
27089 * config/i386/i386.c (ix86_expand_builtin): Remove assert
27090 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
27092 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
27094 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
27095 loop_vinfo argument and use of dependence distance vectors.
27096 Check instead whether the two references differ only in their
27097 initial value and assume that they have the same alignment if the
27098 difference is a multiple of the vector alignment.
27099 (vect_analyze_data_refs_alignment): Update call accordingly.
27101 2017-05-31 Martin Liska <mliska@suse.cz>
27104 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
27106 2017-05-31 Bin Cheng <bin.cheng@arm.com>
27108 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
27109 (create_intersect_range_checks): Move from ...
27110 * tree-data-ref.c (create_intersect_range_checks_index)
27111 (create_intersect_range_checks): ... to here.
27112 (create_runtime_alias_checks): New function factored from ...
27113 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
27114 here. Call above function.
27115 * tree-data-ref.h (create_runtime_alias_checks): New function.
27117 2017-05-31 Bin Cheng <bin.cheng@arm.com>
27119 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
27120 segment length for dr_b and compute it in wide_int.
27122 2017-05-31 Richard Biener <rguenther@suse.de>
27124 PR tree-optimization/80906
27125 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
27126 and pass through iv_map.
27127 (copy_bb_and_scalar_dependences): Adjust.
27128 (translate_pending_phi_nodes): Likewise.
27129 (copy_loop_close_phi_args): Handle code-generating IVs instead
27132 2017-05-30 David Malcolm <dmalcolm@redhat.com>
27134 * diagnostic-color.c (color_dict): Add "type-diff".
27135 (parse_gcc_colors): Update comment.
27136 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
27137 -fdiagnostics-show-template-tree and -fno-elide-type.
27138 (GCC_COLORS): Add type-diff to example.
27140 (-fdiagnostics-show-template-tree): New.
27141 (-fno-elide-type): New.
27142 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
27143 the pp_format_decoder callback. Call any m_format_postprocessor's
27145 (pretty_printer::pretty_printer): Initialize
27146 m_format_postprocessor.
27147 (pretty_printer::~pretty_printer): Delete any
27148 m_format_postprocessor.
27149 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
27150 (class format_postprocessor): New class.
27151 (struct pretty_printer::format_decoder): Document the new parameters.
27152 (struct pretty_printer::m_format_postprocessor): New field.
27153 * tree-diagnostic.c (default_tree_printer): Update for new
27154 bool and const char ** params.
27155 * tree-diagnostic.h (default_tree_printer): Likewise.
27157 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
27159 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
27160 (lwa_operand): Delete rs6000_gen_cell_microcode test.
27161 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27162 rs6000_gen_cell_microcode code.
27163 (rs6000_final_prescan_insn): Delete.
27164 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
27165 "warn-cell-microcode" entries.
27166 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
27167 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
27168 throughout. Change cc_reg_not_micro_cr0_operand to
27169 cc_reg_not_cr0_operand throughout.
27170 (*extendhi<mode>2_noload): Delete.
27171 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
27172 (mwarn-cell-microcode): Delete.
27173 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
27174 -mgen-cell-microcode and -mwarn-cell-microcode.
27176 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
27179 * config/i386/constraints.md (Yd): New constraint.
27181 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
27182 and (?Yd, r) alternatives. Update insn attributes.
27183 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
27184 and (?*Yd, r) alternatives. Update insn attributes.
27185 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
27187 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
27189 * gimplify.c (gimplify_modify_expr): Don't create a
27190 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
27193 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
27195 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
27197 2017-05-30 Richard Biener <rguenther@suse.de>
27199 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
27200 and reduc_def fields.
27201 (STMT_VINFO_REDUC_TYPE): New define.
27202 (STMT_VINFO_REDUC_DEF): Likewise.
27203 (vect_force_simple_reduction): Adjust prototype.
27204 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
27205 (vect_is_simple_reduction): Remove check_reduction argument.
27206 (vect_force_simple_reduction): Adjust and set
27207 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
27208 (vectorizable_reduction): Do not re-do reduction analysis
27209 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
27210 * tree-parloops.c (gather_scalar_reductions): Adjust.
27212 2017-05-30 Richard Biener <rguenther@suse.de>
27214 PR middle-end/80901
27215 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
27218 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
27220 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
27221 Introduce unknown_misalignment parameter and remove vf.
27222 (vect_peeling_hash_get_lowest_cost):
27223 Pass unknown_misalignment parameter.
27224 (vect_enhance_data_refs_alignment):
27225 Fix unsupportable data ref treatment.
27227 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27229 * tree-vect-data-refs.c (vect_get_data_access_cost):
27230 Workaround for SLP handling.
27231 (vect_enhance_data_refs_alignment):
27232 Compute costs for doing no peeling at all, compare to the best
27233 peeling costs so far and avoid peeling if cheaper.
27235 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27237 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
27238 Return peeling info and set costs to zero for unlimited cost
27240 (vect_enhance_data_refs_alignment): Also inspect all datarefs
27241 with unknown misalignment. Compute and costs for unknown
27242 misalignment, compare them to the costs for known misalignment
27243 and choose the cheapest for peeling.
27245 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27247 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
27248 (vect_get_peeling_costs_all_drs): Create function.
27249 (vect_peeling_hash_get_lowest_cost):
27250 Use vect_get_peeling_costs_all_drs.
27251 (vect_peeling_supportable): Create function.
27252 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
27254 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27256 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
27257 DR_HAS_NEGATIVE_STEP.
27258 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
27259 (vect_enhance_data_refs_alignment): Use.
27260 (vect_duplicate_ssa_name_ptr_info): Use.
27261 * tree-vectorizer.h (dr_misalignment): Use.
27262 (known_alignment_for_access_p): Use.
27264 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27267 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
27269 (has_section_name): New function.
27271 2017-05-30 Martin Liska <mliska@suse.cz>
27274 * auto-profile.c (get_function_decl_from_block): Fix
27277 2017-05-30 Richard Biener <rguenther@suse.de>
27279 PR middle-end/80876
27280 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
27282 2017-05-30 Martin Liska <mliska@suse.cz>
27284 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
27285 * dumpfile.h (struct dump_file_info): Remove ctors.
27287 2017-05-30 Martin Liska <mliska@suse.cz>
27289 * predict.def: Fix GNU coding style.
27291 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
27293 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
27294 Mark 'to' argument with ATTRIBUTE_UNUSED.
27296 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
27298 * config/xtensa/xtensa.c (xtensa_emit_call): Use
27299 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
27300 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
27303 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
27305 * doc/install.texi (Options specification): Restore entry of
27306 --enable-sjlj-exceptions.
27308 2017-05-27 Michael Eager <eager@eagercon.com>
27311 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
27313 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
27315 * config/microblaze/microblaze.h
27316 (FIXED_REGISTERS): Update in macro.
27317 (CALL_USED_REGISTERS): Update in macro.
27319 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
27321 * doc/install.texi: Add links to macOS binary distributions.
27323 2017-05-27 Jakub Jelinek <jakub@redhat.com>
27327 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27329 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
27331 2017-05-26 Martin Liska <mliska@suse.cz>
27333 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
27335 2017-05-26 Martin Liska <mliska@suse.cz>
27337 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
27338 always leading ';; '.
27339 (dump_bb_info): Likewise.
27340 (brief_dump_cfg): Likewise.
27341 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
27342 * dumpfile.c: Remove usage of TDF_VERBOSE.
27343 * dumpfile.h (enum dump_kind): Likewise.
27344 (dump_gimple_bb_header): Do not use TDF_COMMENT.
27345 * print-tree.c (debug_verbose): Remove.
27346 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
27347 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
27348 * tree-diagnostic.c (default_tree_printer): Replace
27349 TDF_DIAGNOSTIC with TDF_SLIM.
27351 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27353 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
27354 in parameter loop, rather than loop_vinfo.
27355 (create_intersect_range_checks): Ditto.
27356 (vect_create_cond_for_alias_checks): Update call to above functions.
27358 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27360 PR tree-optimization/80815
27361 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
27362 for merging runtime alias checks. Handle negative DR_STEPs.
27364 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27366 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
27368 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
27369 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
27370 out code pruning runtime alias checks.
27371 * tree-data-ref.c (prune_runtime_alias_test_list): New function
27372 factored out from above.
27373 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
27375 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
27377 (prune_runtime_alias_test_list): New decalaration.
27379 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27381 * tree-vect-data-refs.c (compare_tree): Rename and move ...
27382 * tree-data-ref.c (data_ref_compare_tree): ... to here.
27383 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
27384 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
27385 (operator==, comp_dr_with_seg_len_pair): Ditto.
27386 (vect_prune_runtime_alias_test_list): Ditto.
27388 2017-05-26 Martin Liska <mliska@suse.cz>
27391 * params.def: Bound partial-inlining-entry-probability param.
27393 2017-05-26 Marek Polacek <polacek@redhat.com>
27396 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
27399 2017-05-26 Richard Biener <rguenther@suse.de>
27401 PR tree-optimization/80842
27402 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
27405 2017-05-26 Richard Biener <rguenther@suse.de>
27407 PR tree-optimization/80844
27408 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
27410 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
27412 * doc/md.texi (Machine Constraints): Update x86 family
27413 machine constraints section to match 'config/i386/constraints.md'.
27415 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
27417 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
27419 2017-05-25 Nathan Sidwell <nathan@acm.org>
27421 * doc/invoke.texi (--enable-languages): Update documentation.
27423 2017-05-25 Martin Liska <mliska@suse.cz>
27425 * dumpfile.c: Add TDF_FOLDING.
27426 * dumpfile.h (enum dump_kind): Likewise.
27427 * genmatch.c (dt_simplify::gen_1): Use it.
27429 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27431 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
27433 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27435 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
27436 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
27438 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27440 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
27441 * match.pd (X == C): Rewrite it here.
27442 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
27443 with_certain_nonzero_bits2): New predicates.
27444 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
27446 2017-05-24 Nathan Sidwell <nathan@acm.org>
27448 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
27451 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
27454 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
27456 * config/powerpcspe: New port. Files are copied from the rs6000
27457 port, with "rs6000" in filenames replaced by "powerpcspe".
27459 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
27461 PR rtl-optimization/80754
27462 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
27464 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
27466 * config/sparc/sparc.md (length): Return the correct value for -mflat
27467 sibcalls to match output_sibcall.
27469 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
27473 * config/rs6000/rs6000.c (struct machine_function): Add new field
27475 (rs6000_get_separate_components): Init that field, use it.
27476 (rs6000_components_for_bb): Use the field.
27478 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
27480 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
27482 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
27484 PR middle-end/80823
27485 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
27487 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27490 * config/s390/s390.c (s390_check_qrst_address): Check incoming
27491 address against address_operand predicate.
27492 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
27494 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
27496 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
27497 parameters passed indirectly.
27499 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
27501 * config/i386/i386.md (*movdi_internal): Remove SSE4
27502 alternative 18 (?r, *v). Update insn attributes.
27503 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
27504 Update insn attributes.
27505 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
27506 Update insn attributes.
27507 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
27508 alternative 1 (r, v). Remove isa attribute.
27509 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
27510 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
27511 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
27513 2017-05-23 Tom de Vries <tom@codesourcery.com>
27515 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
27518 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27520 * cgraphunit.c (symbol_table::process_new_functions): Update.
27521 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
27522 (inline_generate_summary): Rename to ...
27523 (ipa_fn_summary_generate): ... this one.
27524 (inline_read_summary): Rename to ...
27525 (ipa_fn_summary_read): ... this one.
27526 (inline_write_summary): Rename to ...
27527 (ipa_fn_summary_write): ... this one.
27528 (inline_free_summary): Rename to ...
27529 (ipa_free_fn_summary): ... this one.
27530 (pass_data_local_fn_summary, pass_local_fn_summary,
27531 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
27532 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
27533 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
27534 make_pass_ipa_fn_summary): New.
27535 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
27536 inline_write_summary, inline_free_summary): Remove.
27537 (ipa_free_fn_summary) : New.
27538 * ipa-inline.c (ipa_inline): Update.
27539 (pass_ipa_inline): Do not generate summaries.
27540 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
27542 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
27543 and add pass_ipa_fn_summary.
27544 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
27546 (make_pass_inline_parameters): Remove.
27548 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
27550 * omp-low.c (struct omp_context): Remove "default_kind" member.
27553 * omp-offload.c (execute_oacc_device_lower): Remove the
27554 parallelism dimensions function attributes for unparallelized
27555 OpenACC kernels constructs.
27557 2017-05-23 Martin Liska <mliska@suse.cz>
27559 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
27561 (cgraph_edge::make_speculative): Likewise.
27562 (cgraph_edge::resolve_speculation): Likewise.
27563 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
27564 (cgraph_node::dump): Likewise.
27565 * cgraph.h: Likewise.
27566 * cgraphunit.c (analyze_functions): Likewise.
27567 (symbol_table::compile): Likewise.
27568 * ipa-cp.c (print_all_lattices): Likewise.
27569 (determine_versionability): Likewise.
27570 (initialize_node_lattices): Likewise.
27571 (ipcp_verify_propagated_values): Likewise.
27572 (estimate_local_effects): Likewise.
27573 (update_profiling_info): Likewise.
27574 (create_specialized_node): Likewise.
27575 (perhaps_add_new_callers): Likewise.
27576 (decide_about_value): Likewise.
27577 (decide_whether_version_node): Likewise.
27578 (identify_dead_nodes): Likewise.
27579 (ipcp_store_bits_results): Likewise.
27580 * ipa-devirt.c (dump_targets): Likewise.
27581 (ipa_devirt): Likewise.
27582 * ipa-icf.c (sem_item::dump): Likewise.
27583 (sem_function::equals): Likewise.
27584 (sem_variable::equals): Likewise.
27585 (sem_item_optimizer::read_section): Likewise.
27586 (sem_item_optimizer::execute): Likewise.
27587 (congruence_class::dump): Likewise.
27588 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
27589 (dump_inline_summary): Likewise.
27590 (estimate_node_size_and_time): Likewise.
27591 (inline_analyze_function): Likewise.
27592 * ipa-inline-transform.c (inline_call): Likewise.
27593 * ipa-inline.c (report_inline_failed_reason): Likewise.
27594 (want_early_inline_function_p): Likewise.
27595 (edge_badness): Likewise.
27596 (update_edge_key): Likewise.
27597 (inline_small_functions): Likewise.
27598 * ipa-profile.c (ipa_profile): Likewise.
27599 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
27600 (ipa_make_edge_direct_to_target): Likewise.
27601 (remove_described_reference): Likewise.
27602 (ipa_impossible_devirt_target): Likewise.
27603 (propagate_controlled_uses): Likewise.
27604 (ipa_print_node_params): Likewise.
27605 (ipcp_transform_function): Likewise.
27606 * ipa-pure-const.c (pure_const_read_summary): Likewise.
27607 (propagate_pure_const): Likewise.
27608 * ipa-reference.c (generate_summary): Likewise.
27609 (read_write_all_from_decl): Likewise.
27610 (propagate): Likewise.
27611 (ipa_reference_read_optimization_summary): Likewise.
27612 * ipa-utils.c (ipa_merge_profiles): Likewise.
27613 * ipa.c (walk_polymorphic_call_targets): Likewise.
27614 (symbol_table::remove_unreachable_nodes): Likewise.
27615 (ipa_single_use): Likewise.
27616 * passes.c (execute_todo): Likewise.
27617 * predict.c (drop_profile): Likewise.
27618 * symtab.c (symtab_node::get_dump_name): New function.
27619 (symtab_node::dump_name): Likewise.
27620 (symtab_node::dump_asm_name): Likewise.
27621 (symtab_node::dump_references): Likewise.
27622 (symtab_node::dump_referring): Likewise.
27623 (symtab_node::dump_base): Likewise.
27624 (symtab_node::debug_symtab): Likewise.
27625 * tree-sra.c (convert_callers_for_node): Likewise.
27626 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
27627 * value-prof.c (init_node_map): Likewise.
27629 2017-05-23 Martin Liska <mliska@suse.cz>
27631 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
27632 and symtab_node::debug_symtab to symbol_table::debug.
27633 * cgraphunit.c (analyze_functions): Use the renamed function.
27634 (symbol_table::compile): Likewise.
27635 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
27636 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
27637 * passes.c (execute_todo): Likewise.
27638 * symtab.c (symbol_table::dump): New function.
27639 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
27641 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27643 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
27644 that nonconst implies exec.
27646 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27648 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27649 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27650 (inline_edge_summary_vec): Turn into ...
27651 (ipa_call_summaries): ... this one.
27652 (redirect_to_unreachable, edge_set_predicate,
27653 evaluate_properties_for_edge, inline_summary_alloc,
27654 reset_ipa_call_summary, reset_inline_summary,
27655 inline_summary_t::duplicate): Update.
27656 (inline_edge_duplication_hook): Turn to ...
27657 (ipa_call_summary_t::duplicate): ... this one.
27658 (inline_edge_removal_hook): Turn to ...
27659 (ipa_call_summary_t::remove): ... this one.
27660 (dump_inline_edge_summary): Turn to ...
27661 (dump_ipa_call_summary): ... this one.
27662 (estimate_function_body_sizes): Update.
27663 (inline_update_callee_summaries): Update.
27664 (remap_edge_change_prob): Update.
27665 (remap_edge_summaries): Update.
27666 (inline_merge_summary): Update.
27667 (do_estimate_edge_time): Update.
27668 (inline_generate_summary): Update.
27669 (inline_read_section): Update.
27670 (inline_read_summary): Update.
27671 (inline_free_summary): Update.
27672 * ipa-inline.c (can_inline_edge_p): Update.
27673 (compute_inlined_call_time): Update.
27674 (want_inline_small_function_p): Update.
27675 (edge_badness): Update.
27676 (early_inliner): Update.
27677 * ipa-inline.h (inline_edge_summary): Turn to ...
27678 (ipa_call_summary): ... this one.
27679 (ipa_call_summary_t): New class.
27680 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27681 (ipa_call_summaries): New.
27682 (inline_edge_summary): Remove.
27683 (estimate_edge_growth): Update.
27684 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27685 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27686 * ipa-split.c (execute_split_functions): Update.
27687 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27689 2017-05-23 Tom de Vries <tom@codesourcery.com>
27691 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27692 attributes): Document rdrand effective target.
27694 2017-05-23 Tom de Vries <tom@codesourcery.com>
27696 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27697 attributes): Sort alphabetically.
27699 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
27701 * config/avr/genmultilib.awk: Use gsub instead of gensub.
27703 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
27706 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
27707 V2DF/V2DI splat into two separate patterns, one that handles
27708 registers, and the other that only handles memory. Drop support
27709 for splatting from a GPR on ISA 2.07 and then splitting the
27710 splat into direct move and splat.
27711 (vsx_splat_<mode>_reg): Likewise.
27712 (vsx_splat_<mode>_mem): Likewise.
27714 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
27716 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
27718 2017-05-22 Jakub Jelinek <jakub@redhat.com>
27720 PR middle-end/80809
27721 * omp-low.c (finish_taskreg_remap): New function.
27722 (finish_taskreg_scan): If unit size of ctx->record_type
27723 is non-constant, unshare the size expression and replace
27724 decls in it with possible outer var refs.
27726 PR middle-end/80809
27727 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
27728 GOVD_SHARED rather than GOVD_PRIVATE with it.
27729 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
27730 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
27732 PR middle-end/80853
27733 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
27734 as last argument to build_outer_var_ref for pointer bases of array
27735 section reductions.
27737 2017-05-19 Martin Sebor <msebor@redhat.com>
27739 * print-tree.c (print_node): Print DECL_READ_P flag.
27741 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27743 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
27744 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
27745 * cgraph.c: Likewise.
27746 * cgraphunit.c: Likewise.
27747 * gengtype.c: Likewise.
27748 * ipa-cp.c: Likewise.
27749 * ipa-devirt.c: Likewise.
27750 * ipa-icf.c: Likewise.
27751 * ipa-predicate.c: Likewise.
27752 * ipa-profile.c: Likewise.
27753 * ipa-prop.c: Likewise.
27754 * ipa-split.c: Likewise.
27756 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
27757 edge_predicate_pool, dump_inline_hints,
27758 inline_summary::account_size_time, redirect_to_unreachable,
27759 edge_set_predicate, set_hint_predicate,
27760 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
27761 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
27762 inline_summary_t::remove, remap_hint_predicate_after_duplication,
27763 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
27764 ipa_call_summary_t::remove, initialize_growth_caches,
27765 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
27766 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
27767 mark_modified, unmodified_parm_1, unmodified_parm,
27768 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
27769 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
27770 compute_bb_predicates, will_be_nonconstant_expr_predicate,
27771 will_be_nonconstant_predicate, record_modified_bb_info,
27772 get_minimal_bb, record_modified, param_change_prob,
27773 phi_result_unknown_predicate, predicate_for_phi_result,
27774 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
27775 estimate_function_body_sizes, compute_inline_parameters,
27776 compute_inline_parameters_for_curren, pass_data_inline_parameters,
27777 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
27778 inline_update_callee_summaries, remap_edge_change_prob,
27779 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
27780 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
27781 inline_analyze_function, inline_summary_t::insert,
27782 inline_generate_summary, read_ipa_call_summary, inline_read_section,
27783 inline_read_summary, write_ipa_call_summary, inline_write_summary,
27784 inline_free_summary): Move to ipa-fnsummary.h
27785 (predicate_t): Remove.
27786 * ipa-fnsummary.c: New file.
27787 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
27788 (enum inline_hints_vals, inline_hints, agg_position_info,
27789 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
27790 inline_summaries, ipa_call_summary, ipa_call_summary_t,
27791 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
27792 dump_inline_summary, dump_inline_hints, inline_generate_summary,
27793 inline_read_summary, inline_write_summary, inline_free_summary,
27794 inline_analyze_function, initialize_inline_failed,
27795 inline_merge_summary, inline_update_overall_summary,
27796 compute_inline_parameters): Move to ipa-fnsummary.h
27797 * ipa-fnsummary.h: New file.
27798 * ipa-inline-transform.h: Include ipa-inline.h.
27799 * ipa-inline.c: LIkewise.
27801 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27803 * ipa-inline.c (edge_badness): Use inlined_time instead of
27804 inline_summaries->get.
27806 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27808 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
27810 2017-05-22 Nathan Sidwell <nathan@acm.org>
27812 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
27813 (fdump-lang): Document 'raw' option.
27814 * dumpfile.h (TDI_tu): Delete.
27815 * dumpfile.c (dump_files): Remove translation-unit.
27816 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
27818 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
27820 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
27821 command option from $(AWK) call.
27822 * config/avr/genmultilib.awk: Simplify and rewrite so that it
27823 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
27824 [FORMAT]: Remove handling of variable.
27825 * config/avr/t-multilib: Regenerate.
27827 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27829 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
27831 (dump_inline_summary): Do not print self_time.
27832 (estimate_function_body_sizes): Do not set self_time.
27833 (compute_inline_parameters): Likewise.
27834 (inline_read_section, inline_write_summary): Do not stream self_time.
27835 * ipa-inline.h (inline_summary): Drop self_time.
27837 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27839 * ipa-inline-analysis.c (account_size_time): Rename to ...
27840 (inline_summary::account_size_time): ... this one.
27841 (reset_ipa_call_summary): Turn to ...
27842 (ipa_call_summary::reset): ... this one.
27843 (reset_inline_summary): Turn to ...
27844 (inline_summary::reset): ... this one.
27845 (inline_summary_t::remove): Update.
27846 (inline_summary_t::duplicate): Update.
27847 (ipa_call_summary_t::remove): Update.
27848 (dump_inline_summary): Update.
27849 (estimate_function_body_sizes): Update.
27850 (compute_inline_parameters): Update.
27851 (estimate_node_size_and_time): Update.
27852 (inline_merge_summary): Update.
27853 (inline_update_overall_summary): Update.
27854 (inline_read_section): Update.
27855 (inline_write_summary): Update.
27856 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
27857 add account_size_time and reset member functions.
27858 (ipa_call_summary): Add reset function.
27859 * ipa-predicate.h (predicate::operator &): Constify.
27861 2017-05-22 Richard Biener <rguenther@suse.de>
27863 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
27865 2017-05-19 Jason Merrill <jason@redhat.com>
27867 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
27869 2017-05-19 Marek Polacek <polacek@redhat.com>
27872 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
27873 TYPE_OVERFLOW_WRAPS checks.
27875 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
27877 * tree-core.h (enum omp_clause_default_kind): Add
27878 "OMP_CLAUSE_DEFAULT_PRESENT".
27879 * tree-pretty-print.c (dump_omp_clause): Handle it.
27880 * gimplify.c (enum gimplify_omp_var_data): Add
27881 "GOVD_MAP_FORCE_PRESENT".
27882 (gimplify_adjust_omp_clauses_1): Map it to
27883 "GOMP_MAP_FORCE_PRESENT".
27884 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
27886 * gimplify.c (oacc_default_clause): Clarify.
27888 2017-05-19 Nathan Sidwell <nathan@acm.org>
27890 LANG_HOOK_REGISTER_DUMPS
27891 * toplev.c (general_init): Call register dump lang hook.
27892 * doc/invoke.texi: Document -fdump-lang option family.
27893 * dumpfile.c (dump_files): Remove class dump here.
27894 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
27895 * dumpfile.h (tree_dump_index): Remove TDI_class.
27896 * langhooks-def.h (lhd_register_dumps): Declare.
27897 (LANG_HOOKS_REGISTER_DUMPS): Define.
27898 (LANG_HOOKS_INITIALIZER): Add it.
27899 * langhooks.c (lhd_register_dumps): Define.
27900 * langhooks.h (struct lang_hooks): Add register_dumps.
27902 2017-05-19 Nathan Sidwell <nathan@acm.org>
27904 * context.h (context::set_passes): New.
27905 * context.c (context::context): Do not create pass manager.
27906 * toplev.c (general_init): Create pass manager here.
27908 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
27910 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
27911 use this splitter if two add or or instructions would also work for
27912 the constant we want to generate.
27914 2017-05-19 Richard Biener <rguenther@suse.de>
27917 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
27918 predicate evaluation.
27920 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27922 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
27924 * ipa-inline.c (want_inline_small_function_p): Do not cast to
27927 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27929 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27930 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27931 (inline_edge_summary_vec): Turn into ...
27932 (ipa_call_summaries): ... this one.
27933 (redirect_to_unreachable, edge_set_predicate,
27934 evaluate_properties_for_edge, inline_summary_alloc,
27935 reset_ipa_call_summary, reset_inline_summary,
27936 inline_summary_t::duplicate): Update.
27937 (inline_edge_duplication_hook): Turn to ...
27938 (ipa_call_summary_t::duplicate): ... this one.
27939 (inline_edge_removal_hook): Turn to ...
27940 (ipa_call_summary_t::remove): ... this one.
27941 (dump_inline_edge_summary): Turn to ...
27942 (dump_ipa_call_summary): ... this one.
27943 (estimate_function_body_sizes): Update.
27944 (inline_update_callee_summaries): Update.
27945 (remap_edge_change_prob): Update.
27946 (remap_edge_summaries): Update.
27947 (inline_merge_summary): Update.
27948 (do_estimate_edge_time): Update.
27949 (inline_generate_summary): Update.
27950 (inline_read_section): Update.
27951 (inline_read_summary): Update.
27952 (inline_free_summary): Update.
27953 * ipa-inline.c (can_inline_edge_p): Update.
27954 (compute_inlined_call_time): Update.
27955 (want_inline_small_function_p): Update.
27956 (edge_badness): Update.
27957 (early_inliner): Update.
27958 * ipa-inline.h (inline_edge_summary): Turn to ...
27959 (ipa_call_summary): ... this one.
27960 (ipa_call_summary_t): New class.
27961 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27962 (ipa_call_summaries): New.
27963 (inline_edge_summary): Remove.
27964 (estimate_edge_growth): Update.
27965 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27966 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27967 * ipa-split.c (execute_split_functions): Update.
27968 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27970 2017-05-19 Richard Biener <rguenther@suse.de>
27972 PR middle-end/80764
27973 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
27975 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
27977 * config/rs6000/rs6000.c (struct machine_function): Add field
27978 fpr_is_wrapped_separately.
27979 (rs6000_get_separate_components): Use 64 components. Handle the
27980 new FPR components.
27981 (rs6000_components_for_bb): Handle the FPR components.
27982 (rs6000_emit_prologue_components): Handle the FPR components.
27983 (rs6000_emit_epilogue_components): Handle the FPR components.
27984 (rs6000_set_handled_components): Handle the FPR components.
27985 (rs6000_emit_prologue): Don't output prologue code for those FPRs
27986 that are already separately shrink-wrapped.
27987 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
27988 that are already separately shrink-wrapped.
27990 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27993 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
27996 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
27997 (define_peephole2 for Altivec d-form load): Add peepholes to catch
27998 cases where the register allocator uses a move and an offsettable
27999 memory operation to/from a FPR register on ISA 2.06/2.07.
28000 (define_peephole2 for Altivec d-form store): Likewise.
28002 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
28005 * config/i386/mmx.md (*mov<mode>_internal): Enable
28006 alternatives 11, 12, 13 and 14 also for 32bit targets.
28007 Remove alternatives 15, 16, 17 and 18.
28008 * config/i386/sse.md (vec_concatv2di): Change
28009 alternative (!x, *y) to (x, ?!*Yn).
28011 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
28013 * dumpfile.h (enum dump_kind): Remove stray comma.
28015 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
28017 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
28018 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
28019 predicate::num_conditions
28020 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
28021 (CHANGED): turn into predicate::changed.
28022 (agg_position_info): Move to ipa-predicate.h
28023 (add_condition, predicate::add_clause, predicate::operator &=,
28024 predicate::or_with, predicate::evaluate, predicate::probability,
28025 dump_condition, dump_clause, predicate::dump,
28026 predicate::remap_after_duplication, predicate::remap_after_inlining,
28027 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
28028 (evaluate_conditions_for_known_args): Update.
28029 (set_cond_stmt_execution_predicate): Update.
28030 * ipa-inline.h: Include ipa-predicate.h
28031 (condition, inline_param_summary, conditions, agg_position_info,
28032 predicate): Move to ipa-predicate.h
28033 * ipa-predicate.c: New file.
28034 * ipa-predicate.h: New file.
28036 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
28038 * final.c (leaf_function_p): Check we are not in a sequence.
28040 2017-05-18 Martin Liska <mliska@suse.cz>
28042 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
28043 * dumpfile.c (dump_register): Use new enum dump_kind.
28044 (get_dump_file_name): Likewise.
28045 (dump_enable_all): Likewise.
28046 (dump_switch_p_1): Likewise.
28047 (enable_rtl_dump_file): Remove usage of TDF_RTL.
28048 * dumpfile.h (enum dump_kind): New enum type.
28049 (struct dump_file_info): Create constructor and
28050 format fields and comments.
28051 * passes.c (pass_manager::register_one_dump_file):
28053 * statistics.c (statistics_early_init): Likewise.
28054 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
28055 TDF_TREE with TDF_SLIM.
28056 (gather_memory_references_ref): Likewise.
28058 2017-05-18 Martin Liska <mliska@suse.cz>
28060 * vec.h (struct vnull): Use it.
28062 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
28064 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
28065 (true_predicate, false_predicate, true_predicate_p,
28066 false_predicate_p): Remove.
28067 (single_cond_predicate, not_inlined_predicate): Turn to member function
28069 (add_condition): Update.
28070 (add_clause): Turn to...
28071 (predicate::add_clause): ... this one; update; allow passing NULL
28073 (and_predicates): Turn to ...
28074 (predicate::operator &=): ... this one.
28075 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
28076 (or_predicates): Turn to ...
28077 (predicate::or_with): ... this one.
28078 (evaluate_predicate): Turn to ...
28079 (predicate::evaluate): ... this one.
28080 (predicate_probability): Turn to ...
28081 (predicate::probability): ... this one.
28082 (dump_condition): Update.
28083 (dump_predicate): Turn to ...
28084 (predicate::dump): ... this one.
28085 (account_size_time): Update.
28086 (edge_set_predicate): Update.
28087 (set_hint_predicate): UPdate.
28088 (evaluate_conditions_for_known_args): Update.
28089 (evaluate_properties_for_edge): Update.
28090 (remap_predicate_after_duplication): Turn to...
28091 (predicate::remap_after_duplication): ... this one.
28092 (remap_hint_predicate_after_duplication): Update.
28093 (inline_summary_t::duplicate): UPdate.
28094 (dump_inline_edge_summary): Update.
28095 (dump_inline_summary): Update.
28096 (set_cond_stmt_execution_predicate): Update.
28097 (set_switch_stmt_execution_predicate): Update.
28098 (compute_bb_predicates): Update.
28099 (will_be_nonconstant_expr_predicate): Update.
28100 (will_be_nonconstant_predicate): Update.
28101 (phi_result_unknown_predicate): Update.
28102 (predicate_for_phi_result): Update.
28103 (array_index_predicate): Update.
28104 (estimate_function_body_sizes): Update.
28105 (estimate_node_size_and_time): Update.
28106 (estimate_ipcp_clone_size_and_time): Update.
28107 (remap_predicate): Rename to ...
28108 (predicate::remap_after_inlining): ... this one.
28109 (remap_hint_predicate): Update.
28110 (inline_merge_summary): Update.
28111 (inline_update_overall_summary): Update.
28112 (estimate_size_after_inlining): Update.
28113 (read_predicate): Rename to ...
28114 (predicate::stream_in): ... this one.
28115 (read_inline_edge_summary): Update.
28116 (write_predicate): Rename to ...
28117 (predicate::stream_out): ... this one.
28118 (write_inline_edge_summary): Update.
28119 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
28120 (clause_t): Turn to uint32_t
28121 (predicate): Turn to class; implement constructor and operators
28123 (size_time_entry): Update.
28124 (inline_summary): Update.
28125 (inline_edge_summary): Update.
28127 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
28129 * fold-const.c (fold_binary_loc): Move transformation...
28130 * match.pd (C - X CMP X): ... here.
28132 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
28134 * config/sparc/sparc.c (sparc_option_override): Set function
28135 alignment for -mcpu=niagara7 to 64 to match the I$ line.
28136 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
28138 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
28140 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
28142 2017-05-18 Marek Polacek <polacek@redhat.com>
28145 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
28146 (pass_ubsan::execute): Call gimple_assign_single_p instead of
28147 gimple_assign_load_p.
28149 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
28151 PR middle-end/80692
28152 * real.c (do_compare): Give decimal_do_compare preference over
28153 comparing just the signs.
28155 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
28157 * doc/md.texi (Canonicalization of Instructions): Describe the
28158 canonical form of instructions that inherently set a condition
28161 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
28163 PR middle-end/80775
28164 * tree-cfg.c: Move deletion of unreachable case statements to after
28165 the merging of consecutive case labels.
28167 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
28169 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
28170 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
28171 restoring of callee-saved registers.
28173 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
28175 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
28176 * config/visium/visium.c (single_set_and_flags): Likewise.
28177 * config/visium/visium.md (Substitutions): Likewise.
28179 2017-05-17 Martin Liska <mliska@suse.cz>
28181 * cfg.c: Introduce dump_flags_t type and
28182 use it instead of int type.
28184 * cfghooks.c: Likewise.
28185 * cfghooks.h (struct cfg_hooks): Likewise.
28186 * cfgrtl.c: Likewise.
28187 * cfgrtl.h: Likewise.
28188 * cgraph.c (cgraph_node::get_body): Likewise.
28189 * coretypes.h: Likewise.
28190 * domwalk.c: Likewise.
28191 * domwalk.h: Likewise.
28192 * dumpfile.c (struct dump_option_value_info): Likewise.
28193 (dump_enable_all): Likewise.
28194 (dump_switch_p_1): Likewise.
28195 (opt_info_switch_p): Likewise.
28196 * dumpfile.h (enum tree_dump_index): Likewise.
28197 (struct dump_file_info): Likewise.
28198 * genemit.c: Likewise.
28199 * generic-match-head.c: Likewise.
28200 * gengtype.c (open_base_files): Likewise.
28201 * gimple-pretty-print.c: Likewise.
28202 * gimple-pretty-print.h: Likewise.
28203 * graph.c (print_graph_cfg): Likewise.
28204 * graphite-scop-detection.c (dot_all_sese): Likewise.
28205 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
28206 * loop-unroll.c (report_unroll): Likewise.
28207 * passes.c (pass_manager::register_one_dump_file): Likewise.
28208 * print-tree.c: Likewise.
28209 * statistics.c: Likewise.
28210 * tree-cfg.c: Likewise.
28211 * tree-cfg.h: Likewise.
28212 * tree-dfa.c: Likewise.
28213 * tree-dfa.h: Likewise.
28214 * tree-dump.c (dump_function): Likewise.
28215 * tree-dump.h (struct dump_info): Likewise.
28216 * tree-pretty-print.c: Likewise.
28217 * tree-pretty-print.h: Likewise.
28218 * tree-ssa-live.c: Likewise.
28219 * tree-ssa-live.h: Likewise.
28220 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
28221 * tree-vect-loop.c: Likewise.
28222 * tree-vect-slp.c: Likewise.
28224 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
28225 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28227 PR tree-optimization/80457
28228 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
28229 of all arguments to a statement as scalar_to_vec operations.
28230 (vectorizable_call): Adjust call to vect_model_simple_cost for
28232 (vectorizable_conversion): Likewise.
28233 (vectorizable_assignment): Likewise.
28234 (vectorizable_shift): Likewise.
28235 (vectorizable_operation): Likewise.
28236 (vectorizable_comparison): Likewise.
28237 (vect_is_simple_cond): Record the def types for operands.
28238 (vectorizable_condition): Likewise, call vect_model_simple_cost.
28239 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
28240 for statement argument count.
28242 2017-05-16 Carl Love <cel@us.ibm.com>
28244 * config/rs6000/rs6000-c: Add support for built-in functions
28245 vector unsigned long long vec_bperm (vector unsigned long long,
28246 vector unsigned char)
28247 vector signed long long vec_mule (vector signed int,
28249 vector unsigned long long vec_mule (vector unsigned int,
28250 vector unsigned int)
28251 vector signed long long vec_mulo (vector signed int,
28253 vector unsigned long long vec_mulo (vector unsigned int,
28254 vector unsigned int)
28255 vector signed char vec_sldw (vector signed char,
28256 vector signed char,
28258 vector unsigned char vec_sldw (vector unsigned char,
28259 vector unsigned char,
28261 vector signed short vec_sldw (vector signed short,
28262 vector signed short,
28264 vector unsigned short vec_sldw (vector unsigned short,
28265 vector unsigned short,
28267 vector signed int vec_sldw (vector signed int,
28270 vector unsigned int vec_sldw (vector unsigned int,
28271 vector unsigned int,
28273 vector signed long long vec_sldw (vector signed long long,
28274 vector signed long long,
28276 vector unsigned long long vec_sldw (vector unsigned long long,
28277 vector unsigned long long,
28279 * config/rs6000/rs6000-c: Add support for built-in functions
28280 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
28281 * config/rs6000/altivec.h: Add defintion for vec_sldw.
28282 * doc/extend.texi: Update the built-in documentation for the
28283 new built-in functions.
28285 2017-05-16 Marek Polacek <polacek@redhat.com>
28289 * tree.c (save_expr): Don't fold the expression.
28291 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
28293 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
28294 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
28295 and (?*y,m). Update insn attributes.
28297 2017-05-16 Martin Liska <mliska@suse.cz>
28299 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
28300 flags argument of print_gimple_stmt, print_gimple_expr,
28301 print_generic_stmt and print_generic_expr.
28302 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
28303 * coretypes.h: Likewise.
28304 * except.c (dump_eh_tree): Likewise.
28305 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
28306 * gimple-pretty-print.h: Likewise.
28307 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
28308 (backprop::push_to_worklist): Likewise.
28309 (backprop::pop_from_worklist): Likewise.
28310 (backprop::process_use): Likewise.
28311 (backprop::intersect_uses): Likewise.
28312 (note_replacement): Likewise.
28313 * gimple-ssa-store-merging.c
28314 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
28315 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
28316 (pass_store_merging::execute): Likewise.
28317 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
28318 (ssa_base_cand_dump_callback): Likewise.
28319 (dump_incr_vec): Likewise.
28320 (replace_refs): Likewise.
28321 (replace_mult_candidate): Likewise.
28322 (create_add_on_incoming_edge): Likewise.
28323 (create_phi_basis): Likewise.
28324 (insert_initializers): Likewise.
28325 (all_phi_incrs_profitable): Likewise.
28326 (introduce_cast_before_cand): Likewise.
28327 (replace_one_candidate): Likewise.
28328 * gimplify.c (gimplify_expr): Likewise.
28329 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
28330 (set_rename): Likewise.
28331 (rename_uses): Likewise.
28332 (copy_loop_phi_nodes): Likewise.
28333 (add_close_phis_to_merge_points): Likewise.
28334 (copy_loop_close_phi_args): Likewise.
28335 (copy_cond_phi_args): Likewise.
28336 (graphite_copy_stmts_from_block): Likewise.
28337 (translate_pending_phi_nodes): Likewise.
28338 * graphite-poly.c (print_pdr): Likewise.
28339 (dump_gbb_cases): Likewise.
28340 (dump_gbb_conditions): Likewise.
28341 (print_scop_params): Likewise.
28342 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
28343 (build_cross_bb_scalars_use): Likewise.
28344 (gather_bbs::before_dom_children): Likewise.
28345 * hsa-dump.c (dump_hsa_immed): Likewise.
28346 * ipa-cp.c (print_ipcp_constant_value): Likewise.
28347 (get_replacement_map): Likewise.
28348 * ipa-inline-analysis.c (dump_condition): Likewise.
28349 (estimate_function_body_sizes): Likewise.
28350 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
28351 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
28352 * ipa-prop.c (ipa_dump_param): Likewise.
28353 (ipa_print_node_jump_functions_for_edge): Likewise.
28354 (ipa_modify_call_arguments): Likewise.
28355 (ipa_modify_expr): Likewise.
28356 (ipa_dump_param_adjustments): Likewise.
28357 (ipa_dump_agg_replacement_values): Likewise.
28358 (ipcp_modif_dom_walker::before_dom_children): Likewise.
28359 * ipa-pure-const.c (check_stmt): Likewise.
28360 (pass_nothrow::execute): Likewise.
28361 * ipa-split.c (execute_split_functions): Likewise.
28362 * omp-offload.c (dump_oacc_loop_part): Likewise.
28363 (dump_oacc_loop): Likewise.
28364 * trans-mem.c (tm_log_emit): Likewise.
28365 (tm_memopt_accumulate_memops): Likewise.
28366 (dump_tm_memopt_set): Likewise.
28367 (dump_tm_memopt_transform): Likewise.
28368 * tree-cfg.c (gimple_verify_flow_info): Likewise.
28369 (print_loop): Likewise.
28370 * tree-chkp-opt.c (chkp_print_addr): Likewise.
28371 (chkp_gather_checks_info): Likewise.
28372 (chkp_get_check_result): Likewise.
28373 (chkp_remove_check_if_pass): Likewise.
28374 (chkp_use_outer_bounds_if_possible): Likewise.
28375 (chkp_reduce_bounds_lifetime): Likewise.
28376 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
28377 (chkp_mark_completed_bounds): Likewise.
28378 (chkp_register_incomplete_bounds): Likewise.
28379 (chkp_mark_invalid_bounds): Likewise.
28380 (chkp_maybe_copy_and_register_bounds): Likewise.
28381 (chkp_build_returned_bound): Likewise.
28382 (chkp_get_bound_for_parm): Likewise.
28383 (chkp_build_bndldx): Likewise.
28384 (chkp_get_bounds_by_definition): Likewise.
28385 (chkp_generate_extern_var_bounds): Likewise.
28386 (chkp_get_bounds_for_decl_addr): Likewise.
28387 * tree-chrec.c (chrec_apply): Likewise.
28388 * tree-data-ref.c (dump_data_reference): Likewise.
28389 (dump_subscript): Likewise.
28390 (dump_data_dependence_relation): Likewise.
28391 (analyze_overlapping_iterations): Likewise.
28392 * tree-inline.c (expand_call_inline): Likewise.
28393 (tree_function_versioning): Likewise.
28394 * tree-into-ssa.c (dump_defs_stack): Likewise.
28395 (dump_currdefs): Likewise.
28396 (dump_names_replaced_by): Likewise.
28397 (dump_update_ssa): Likewise.
28398 (update_ssa): Likewise.
28399 * tree-object-size.c (pass_object_sizes::execute): Likewise.
28400 * tree-parloops.c (build_new_reduction): Likewise.
28401 (try_create_reduction_list): Likewise.
28402 (ref_conflicts_with_region): Likewise.
28403 (oacc_entry_exit_ok_1): Likewise.
28404 (oacc_entry_exit_single_gang): Likewise.
28405 * tree-pretty-print.h: Likewise.
28406 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
28407 (get_scalar_evolution): Likewise.
28408 (add_to_evolution): Likewise.
28409 (get_loop_exit_condition): Likewise.
28410 (analyze_evolution_in_loop): Likewise.
28411 (analyze_initial_condition): Likewise.
28412 (analyze_scalar_evolution): Likewise.
28413 (instantiate_scev): Likewise.
28414 (number_of_latch_executions): Likewise.
28415 (gather_chrec_stats): Likewise.
28416 (final_value_replacement_loop): Likewise.
28417 (scev_const_prop): Likewise.
28418 * tree-sra.c (dump_access): Likewise.
28419 (disqualify_candidate): Likewise.
28420 (create_access): Likewise.
28421 (reject): Likewise.
28422 (maybe_add_sra_candidate): Likewise.
28423 (create_access_replacement): Likewise.
28424 (analyze_access_subtree): Likewise.
28425 (analyze_all_variable_accesses): Likewise.
28426 (sra_modify_assign): Likewise.
28427 (initialize_constant_pool_replacements): Likewise.
28428 (find_param_candidates): Likewise.
28429 (decide_one_param_reduction): Likewise.
28430 (replace_removed_params_ssa_names): Likewise.
28431 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
28432 * tree-ssa-copy.c (dump_copy_of): Likewise.
28433 (copy_prop_visit_cond_stmt): Likewise.
28434 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
28435 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
28436 (record_equivalences_from_stmt): Likewise.
28437 * tree-ssa-dse.c (compute_trims): Likewise.
28438 (delete_dead_call): Likewise.
28439 (delete_dead_assignment): Likewise.
28440 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
28441 (forward_propagate_into_cond): Likewise.
28442 (pass_forwprop::execute): Likewise.
28443 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
28444 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
28446 (move_computations_worker): Likewise.
28447 (execute_sm): Likewise.
28448 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
28449 (remove_exits_and_undefined_stmts): Likewise.
28450 (remove_redundant_iv_tests): Likewise.
28451 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
28452 (adjust_iv_update_pos): Likewise.
28453 * tree-ssa-math-opts.c (bswap_replace): Likewise.
28454 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
28455 (value_replacement): Likewise.
28456 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
28457 * tree-ssa-pre.c (print_pre_expr): Likewise.
28458 (get_representative_for): Likewise.
28459 (create_expression_by_pieces): Likewise.
28460 (insert_into_preds_of_block): Likewise.
28461 (eliminate_insert): Likewise.
28462 (eliminate_dom_walker::before_dom_children): Likewise.
28463 (eliminate): Likewise.
28464 (remove_dead_inserted_code): Likewise.
28465 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
28466 * tree-ssa-reassoc.c (get_rank): Likewise.
28467 (eliminate_duplicate_pair): Likewise.
28468 (eliminate_plus_minus_pair): Likewise.
28469 (eliminate_not_pairs): Likewise.
28470 (undistribute_ops_list): Likewise.
28471 (eliminate_redundant_comparison): Likewise.
28472 (update_range_test): Likewise.
28473 (optimize_range_tests_var_bound): Likewise.
28474 (optimize_vec_cond_expr): Likewise.
28475 (rewrite_expr_tree): Likewise.
28476 (rewrite_expr_tree_parallel): Likewise.
28477 (linearize_expr): Likewise.
28478 (break_up_subtract): Likewise.
28479 (linearize_expr_tree): Likewise.
28480 (attempt_builtin_powi): Likewise.
28481 (attempt_builtin_copysign): Likewise.
28482 (transform_stmt_to_copy): Likewise.
28483 (transform_stmt_to_multiply): Likewise.
28484 (dump_ops_vector): Likewise.
28485 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
28486 (print_scc): Likewise.
28487 (set_ssa_val_to): Likewise.
28488 (visit_reference_op_store): Likewise.
28489 (visit_use): Likewise.
28490 (sccvn_dom_walker::before_dom_children): Likewise.
28491 (run_scc_vn): Likewise.
28492 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
28494 (expr_hash_elt::print): Likewise.
28495 (const_and_copies::pop_to_marker): Likewise.
28496 (const_and_copies::record_const_or_copy_raw): Likewise.
28497 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
28498 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
28499 (dump_predicates): Likewise.
28500 (find_uninit_use): Likewise.
28501 (warn_uninitialized_phi): Likewise.
28502 (pass_late_warn_uninitialized::execute): Likewise.
28503 * tree-ssa.c (verify_vssa): Likewise.
28504 (verify_ssa): Likewise.
28505 (maybe_optimize_var): Likewise.
28506 * tree-vrp.c (dump_value_range): Likewise.
28507 (dump_all_value_ranges): Likewise.
28508 (dump_asserts_for): Likewise.
28509 (register_edge_assert_for_2): Likewise.
28510 (vrp_visit_cond_stmt): Likewise.
28511 (vrp_visit_switch_stmt): Likewise.
28512 (vrp_visit_stmt): Likewise.
28513 (vrp_visit_phi_node): Likewise.
28514 (simplify_cond_using_ranges_1): Likewise.
28515 (fold_predicate_in): Likewise.
28516 (evrp_dom_walker::before_dom_children): Likewise.
28517 (evrp_dom_walker::push_value_range): Likewise.
28518 (evrp_dom_walker::pop_value_range): Likewise.
28519 (execute_early_vrp): Likewise.
28521 2017-05-16 Richard Biener <rguenther@suse.de>
28523 * dwarf2out.c (loc_list_from_tree_1): Do not create
28524 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
28526 2017-05-16 Richard Biener <rguenther@suse.de>
28528 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
28530 (note_variable_value_in_expr): If we resolved the decl ref
28531 do not push to the stack.
28533 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
28535 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
28536 operations in fast-math mode.
28537 (vaddq_f16): Likewise.
28538 (vmul_f16): Likewise.
28539 (vmulq_f16): Likewise.
28540 (vsub_f16): Likewise.
28541 (vsubq_f16): Likewise.
28542 * config/arm/neon.md (add<mode>3): New.
28544 (fma:<VH:mode>3): New. Also remove outdated comment.
28547 2017-05-16 Martin Liska <mliska@suse.cz>
28551 * ipa-devirt.c (warn_types_mismatch): Fix typo.
28552 (odr_types_equivalent_p): Likewise.
28554 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
28556 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
28558 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
28561 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
28562 non-interunit SSE move alternatives with '?'.
28563 (zero-extendsidi peephole2): New peephole to skip intermediate
28564 general register in SSE zero-extend sequence.
28566 2017-05-15 Jeff Law <law@redhat.com>
28568 * reorg.c (relax_delay_slots): Create a new variable to hold
28569 the temporary target rather than clobbering TARGET_LABEL.
28571 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
28572 missing argument to extract_bit_field call.
28573 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
28575 2017-05-15 Martin Liska <mliska@suse.cz>
28578 * gcc.c (process_command): Do not allow empty argument of -o option.
28580 2017-05-15 Renlin Li <renlin.li@arm.com>
28582 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
28583 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
28584 * config/aarch64/constraints.md (Usf): Add long call check.
28585 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
28586 (call_value): Likewise.
28587 (sibcall): Likewise.
28588 (sibcall_value): Likewise.
28590 (call_value_insn): New.
28591 (sibcall_insn): Update rtx pattern.
28592 (sibcall_value_insn): Likewise.
28593 (call_internal): Remove.
28594 (call_value_internal): Likewise.
28595 (sibcall_internal): Likewise.
28596 (sibcall_value_internal): Likewise.
28597 (call_reg): Likewise.
28598 (call_symbol): Likewise.
28599 (call_value_reg): Likewise.
28600 (call_value_symbol): Likewise.
28602 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
28605 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
28607 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
28609 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
28610 compatible with CCGOCmode and with CCZmode.
28612 2017-05-14 Martin Sebor <msebor@redhat.com>
28614 PR middle-end/77671
28615 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
28616 (gimple_fold_builtin_snprintf): Same.
28617 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
28618 (gimple_fold_builtin_snprintf): Same.
28619 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
28620 of character types.
28621 (is_call_safe): New function.
28622 (try_substitute_return_value): Call it.
28623 (try_simplify_call): New function.
28624 (pass_sprintf_length::handle_gimple_call): Call it.
28626 2017-05-14 Martin Sebor <msebor@redhat.com>
28628 PR middle-end/80669
28629 * builtins.c (expand_builtin_stpncpy): Simplify.
28631 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
28633 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
28634 * config/i386/i386.h
28635 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
28636 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
28637 (struct machine_function): Add new members call_ms2sysv,
28638 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
28639 (struct machine_frame_state): New fields sp_realigned and
28640 sp_realigned_offset.
28641 * config/i386/i386.c
28642 (enum xlogue_stub): New enum.
28643 (enum xlogue_stub_sets): New enum.
28644 (class xlogue_layout): New class.
28645 (struct ix86_frame): New fields stack_realign_allocate_offset,
28646 stack_realign_offset and outlined_save_offset. Modify comments to
28647 detail stack layout when using out-of-line stubs.
28648 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
28649 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
28650 -mcall-ms2sysv-xlogues.
28651 (stub_managed_regs): New static variable.
28652 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
28653 registers managed by out-of-line stub.
28654 (disable_call_ms2sysv_xlogues): New function.
28655 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
28656 m->call_ms2sysv when appropriate and compute frame layout for
28658 (sp_valid_at, fp_valid_at): New inline functions.
28659 (choose_basereg): New function.
28660 (choose_baseaddr): Add align parameter, use choose_basereg and modify
28662 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
28663 Use align parameter of choose_baseaddr to generated aligned SSE movs
28665 (pro_epilogue_adjust_stack): Modify to track
28666 machine_frame_state::sp_realigned.
28667 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
28668 (ix86_nsaved_sseregs): Likewise.
28669 (ix86_emit_save_regs): Likewise.
28670 (ix86_emit_save_regs_using_mov): Likewise.
28671 (ix86_emit_save_sse_regs_using_mov): Likewise.
28672 (get_scratch_register_on_entry): Likewise.
28673 (gen_frame_set): New function.
28674 (gen_frame_load): Likewise.
28675 (gen_frame_store): Likewise.
28676 (emit_outlined_ms2sysv_save): Likewise.
28677 (emit_outlined_ms2sysv_restore): Likewise.
28678 (ix86_expand_prologue): Modify stack re-alignment code and call
28679 emit_outlined_ms2sysv_save when appropriate.
28680 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
28681 parameter rtx_insn *insn, which allows the function to be used to only
28682 generate the notes.
28683 (ix86_expand_epilogue): Modify validity checks of frame and stack
28684 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
28685 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
28686 * config/i386/predicates.md
28687 (save_multiple): New predicate.
28688 (restore_multiple): Likewise.
28689 * config/i386/sse.md
28690 (save_multiple<mode>): New pattern.
28691 (save_multiple_realign<mode>): Likewise.
28692 (restore_multiple<mode>): Likewise.
28693 (restore_multiple_and_return<mode>): Likewise.
28694 (restore_multiple_leave_return<mode>): Likewise.
28695 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
28697 2017-05-14 Julia Koval <julia.koval@intel.com>
28699 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
28700 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
28701 (__builtin_ia32_xsetbv): New builtins.
28702 * config/i386/i386.c (ix86_expand_special_args_builtin):
28704 (ix86_expand_builtin): Special expand for new intrinsics.
28705 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
28706 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
28707 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
28709 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28711 * cfganal.c (inverted_post_order_compute): Change argument type
28713 * cfganal.h (inverted_post_order_compute): Adjust prototype.
28714 * df-core.c (rest_of_handle_df_initialize): Adjust.
28715 (rest_of_handle_df_finish): Likewise.
28716 (df_analyze_1): Likewise.
28717 (df_analyze): Likewise.
28718 (loop_inverted_post_order_compute): Change argument to be a vec *.
28719 (df_analyze_loop): Adjust.
28720 (df_get_n_blocks): Likewise.
28721 (df_get_postorder): Likewise.
28722 * df.h (struct df_d): Change field to be a vec.
28723 * lcm.c (compute_laterin): Adjust.
28724 (compute_available): Likewise.
28725 * lra-lives.c (lra_create_live_ranges_1): Likewise.
28726 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
28727 * tree-ssa-pre.c (compute_antic): Likewise.
28729 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28731 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
28732 (depth_first_search::depth_first_search): Change structure init
28733 function to this constructor.
28734 (depth_first_search::add_bb): Rename function to this member.
28735 (depth_first_search::execute): Likewise.
28736 (flow_dfs_compute_reverse_finish): Adjust.
28738 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28740 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
28741 (longest_simple_path): Likewise.
28742 * shrink-wrap.c (spread_components): Likewise.
28743 (disqualify_problematic_components): Likewise.
28744 (emit_common_heads_for_components): Likewise.
28745 (emit_common_tails_for_components): Likewise.
28746 (insert_prologue_epilogue_for_components): Likewise.
28748 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28750 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
28753 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28755 * df-core.c (df_set_blocks): Start using auto_bitmap.
28756 (df_compact_blocks): Likewise.
28757 * df-problems.c (df_rd_confluence_n): Likewise.
28758 * df-scan.c (df_insn_rescan_all): Likewise.
28759 (df_process_deferred_rescans): Likewise.
28760 (df_update_entry_block_defs): Likewise.
28761 (df_update_exit_block_uses): Likewise.
28762 (df_entry_block_bitmap_verify): Likewise.
28763 (df_exit_block_bitmap_verify): Likewise.
28764 (df_scan_verify): Likewise.
28765 * lra-constraints.c (lra_constraints): Likewise.
28766 (undo_optional_reloads): Likewise.
28767 (lra_undo_inheritance): Likewise.
28768 * lra-remat.c (calculate_gen_cands): Likewise.
28769 (do_remat): Likewise.
28770 * lra-spills.c (assign_spill_hard_regs): Likewise.
28771 (spill_pseudos): Likewise.
28772 * tree-ssa-pre.c (bitmap_set_and): Likewise.
28773 (bitmap_set_subtract_values): Likewise.
28775 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28777 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
28778 management with auto_bitmap.
28779 (fix_inter_tick): Likewise.
28780 (fix_recovery_deps): Likewise.
28781 * ira.c (add_store_equivs): Likewise.
28782 (find_moveable_pseudos): Likewise.
28783 (split_live_ranges_for_shrink_wrap): Likewise.
28784 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
28785 (rtx_reuse_manager::seen_def_p): Likewise.
28786 (rtx_reuse_manager::set_seen_def): Likewise.
28787 * print-rtl.h (class rtx_reuse_manager): Likewise.
28789 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28791 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
28793 (migrate_btr_def): Likewise.
28794 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
28795 * df-core.c (loop_post_order_compute): Likewise.
28796 (loop_inverted_post_order_compute): Likewise.
28797 * hsa-common.h: Likewise.
28798 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
28799 * init-regs.c (initialize_uninitialized_regs): Likewise.
28800 * ipa-inline.c (resolve_noninline_speculation): Likewise.
28801 (inline_small_functions): Likewise.
28802 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
28803 * ira.c (combine_and_move_insns): Likewise.
28804 (build_insn_chain): Likewise.
28805 * loop-invariant.c (find_invariants): Likewise.
28806 * lower-subreg.c (propagate_pseudo_copies): Likewise.
28807 * predict.c (tree_predict_by_opcode): Likewise.
28808 (predict_paths_leading_to): Likewise.
28809 (predict_paths_leading_to_edge): Likewise.
28810 (estimate_loops_at_level): Likewise.
28811 (estimate_loops): Likewise.
28812 * shrink-wrap.c (try_shrink_wrapping): Likewise.
28813 (spread_components): Likewise.
28814 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
28815 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
28816 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
28817 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
28818 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
28819 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
28820 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
28821 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
28822 (mark_threaded_blocks): Likewise.
28823 (thread_through_all_blocks): Likewise.
28824 * tree-ssa.c (verify_ssa): Likewise.
28825 (execute_update_addresses_taken): Likewise.
28826 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
28828 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28830 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
28832 (post_order_compute): Likewise.
28833 (inverted_post_order_compute): Likewise.
28834 (pre_and_rev_post_order_compute_fn): Likewise.
28836 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28838 * genrecog.c (int_set::int_set): Explicitly construct our
28839 auto_vec base class.
28840 * vec.h (auto_vec::auto_vec): New constructor.
28842 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28844 * bitmap.h (class auto_bitmap): New constructor taking
28845 bitmap_obstack * argument.
28847 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28849 * bitmap.h (class auto_bitmap): Change type of m_bits to
28850 bitmap_head, and adjust ctor / dtor and member operators.
28852 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
28854 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
28855 when returned register mode doesn't match original mode.
28857 2017-05-12 Jeff Law <law@redhat.com>
28858 Jakub Jelinek <jakub@redhat.com>
28860 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
28861 we look for cc setter after the compare-elim changes.
28862 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
28863 within the vector to match what compare-elim now expects.
28864 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
28865 (xorsi3_flags, one_cmplsi2_flags): Likewise.
28867 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
28868 after the compare-elim changes.
28869 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
28870 the vector to match what compare-elim now expects.
28871 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
28872 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
28873 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
28874 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
28875 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
28877 * config/visium/visium.c (single_set_and_flags): Fix where
28878 we look for cc setter after the compare-elim changes.
28879 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
28880 with the vector to match what compare-elim now expects.
28881 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
28882 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
28883 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
28884 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
28885 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
28886 (neg<mode>2_insn_set_overflow): Likewise.
28888 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
28890 PR middle-end/79794
28891 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
28892 maybe_expand_insn call, set ops[0].target. If still set after call,
28893 set alt_rtl. Add extra arg to recursive calls.
28894 (extract_bit_field): Add alt_rtl argument. Pass to
28896 * expmed.h (extract_bit_field): Fix prototype.
28897 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
28898 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
28899 to extract_bit_field_calls.
28900 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
28901 Pass alt_rtl to extract_bit_field calls.
28902 * calls.c (store_unaligned_arguments_into_psuedos)
28903 load_register_parameters): Pass extra NULL to extract_bit_field calls.
28904 * optabs.c (maybe_legitimize_operand): Clear op->target when call
28906 * optabs.h (struct expand_operand): Add target bitfield.
28908 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28910 * compare-elim.c (try_eliminate_compare): Canonicalize
28911 operation with embedded compare to
28912 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
28913 (set (reg) (operation)].
28915 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
28917 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28920 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
28921 cost of adding a carry flag for ADC instruction.
28922 [case MINUS]: Ignore the cost of subtracting a carry flag
28923 for SBB instruction.
28925 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
28927 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
28929 * config/rs6000/bmiintrin.h: New file.
28930 * config/rs6000/bmi2intrin.h: New file.
28931 * config/rs6000/x86intrin.h: New file.
28933 2017-05-12 Jeff Law <law@redhat.com>
28935 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
28938 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
28940 PR middle-end/80707
28941 * tree-cfg.c: Remove cfg edges of unreachable case statements.
28943 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28945 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28946 early expansion of vector divide builtins.
28947 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
28948 builtins identified as having unsigned arguments.
28950 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28952 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
28953 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
28954 expansion of vector logical operations (and, andc, or, xor,
28957 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28959 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
28960 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
28962 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28964 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28965 early GIMPLE expansion of vector multiplies.
28967 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28969 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
28970 TARGET_HAVE_MOVT conditional.
28971 (movt splitter): Likewise.
28973 2017-05-12 Richard Biener <rguenther@suse.de>
28975 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
28976 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28977 Fold all stmts not inplace.
28979 2017-05-12 Richard Biener <rguenther@suse.de>
28981 PR tree-optimization/80713
28982 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
28983 inserted_exprs bit for not removed stmts.
28985 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
28987 PR middle-end/69921
28988 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
28989 parallelized" attribute for parallelized OpenACC kernels.
28990 * omp-offload.c (execute_oacc_device_lower): Use it.
28992 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
28993 Set "oacc kernels" attribute.
28994 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
28995 parameter. Adjust all users.
28996 (oacc_fn_attrib_kernels_p): Remove function.
28997 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
28998 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
28999 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
29000 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
29001 assert "oacc kernels" attribute is set.
29003 2017-05-11 Carl Love <cel@us.ibm.com>
29005 * config/rs6000/rs6000-c: Add support for built-in functions
29006 vector unsigned char vec_popcnt (vector signed char)
29007 vector unsigned char vec_popcnt (vector unsigned char)
29008 vector unsigned short vec_popcnt (vector signed short)
29009 vector unsigned short vec_popcnt (vector unsigned short)
29010 vector unsigned int vec_popcnt (vector signed int)
29011 vector unsigned int vec_popcnt (vector unsigned int)
29012 vector unsigned long long vec_popcnt (vector signed long long)
29013 vector unsigned long long vec_popcnt (vector unsigned long long)
29014 vector signed long long vec_slo (vector signed long long,
29015 vector signed char)
29016 vector signed long long vec_slo (vector signed long long,
29017 vector unsigned char)
29018 vector unsigned long long vec_slo (vector unsigned long long,
29019 vector signed char)
29020 vector unsigned long long vec_slo (vector unsigned long long,
29021 vector unsigned char)
29022 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
29023 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
29024 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
29025 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
29026 * doc/extend.texi: Update the built-in documentation file for the
29027 new built-in functions.
29029 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
29031 * attribs.h (sorted_attr_string): Move machine independent
29032 functions for target clone support from the i386 port to common
29033 code. Rename ix86_function_versions to common_function_versions.
29034 Rename make_name to make_unique_name.
29035 (common_function_versions): Likewise.
29036 (make_unique_name): Likewise.
29037 (make_dispatcher_decl): Likewise.
29038 (is_function_default_version): Likewise.
29039 * attribs.c (attr_strcmp): Likewise.
29040 (sorted_attr_string): Likewise.
29041 (common_function_versions): Likewise.
29042 (make_unique_name): Likewise.
29043 (make_dispatcher_decl): Likewise.
29044 (is_function_default_version): Likewise.
29045 * config/i386/i386.c (attr_strcmp): Likewise.
29046 (sorted_attr_string): Likewise.
29047 (ix86_function_versions): Likewise.
29048 (make_name): Likewise.
29049 (make_dispatcher_decl): Likewise.
29050 (is_function_default_version): Likewise.
29051 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
29053 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29056 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
29057 Account for direct move costs for vec_construct of integer
29060 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
29063 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
29064 (UNSPEC_STX_ATOMIC): Ditto.
29065 (loaddi_via_sse): New insn.
29066 (storedi_via_sse): Ditto.
29067 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
29068 Update corresponding peephole2 patterns.
29069 (atomic_storedi_fpu): Ditto.
29071 2017-05-11 Julia Koval <julia.koval@intel.com>
29073 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
29074 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
29076 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
29077 (__builtin_ia32_rsqrt14ss_mask): New builtins.
29078 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
29080 2017-05-11 Nathan Sidwell <nathan@acm.org>
29082 * graphite-poly.c: Include dumpfile.h.
29084 * dumpfle.h (dump_function): Declare here ...
29085 * tree-dump.h (dump_function): ... not here.
29086 * dumpfile.c: #include tree-cfg.h.
29087 (dump_function): Move here from ...
29088 * tree-dump.c (dump_function): ... here.
29089 * gimplify.c: #include splay-tree.h, not tree-dump.h.
29090 * graphite-poly.c: Don't include tree-dump.h.
29091 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
29092 * print-tree.c: Likewise.
29093 * stor-layout.c: Likewise.
29094 * tree-nested.c: Likewise.
29096 * dumpfile.c (dump_start): Use TDF_FLAGS.
29097 (dump_enable_all): Fix TDF_KIND check thinko.
29099 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
29101 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
29102 array entries to represent two legal parameterizations of the
29103 overloaded __builtin_cmpb function, as represented by the
29104 P6_OV_BUILTIN_CMPB constant.
29105 (altivec_resolve_overloaded_builtin): Add special case handling
29106 for the __builtin_cmpb function, as represented by the
29107 P6_OV_BUILTIN_CMPB constant.
29108 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
29109 (BU_P6_64BIT_2): New macro.
29110 (BU_P6_OVERLOAD_2): New macro
29111 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
29112 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
29113 (CMPB): Add overload support to represent both 32-bit and 64-bit
29114 compare-bytes function.
29115 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
29116 support for TARGET_CMPB.
29117 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
29118 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
29119 documentation of the __builtin_cmpb overloaded built-in function.
29121 2017-05-11 Richard Biener <rguenther@suse.de>
29123 PR tree-optimization/80705
29124 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
29125 bases are not vectorizable.
29127 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29129 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
29130 when counting register pressure.
29132 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29134 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
29135 (iv_ca_more_deps): Renamed to ...
29136 (iv_ca_compare_deps): ... this.
29137 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
29139 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29141 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
29143 (determine_group_iv_costs): ... here.
29144 (find_inv_vars_cb): Record inv var if it's not recorded before.
29146 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29148 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
29149 (get_shiftadd_cost): Ditto.
29151 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29153 * tree-ssa-address.c: Include header file.
29154 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
29156 (add_to_parts): Refactor.
29157 (addr_to_parts): New parameter. Update use of move_hint_to_base.
29158 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
29161 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29163 PR tree-optimization/53090
29164 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
29166 (extract_cond_operands): Detect condition with IV on both sides
29167 and return COMP_IV_EXPR_2.
29168 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
29169 (rewrite_use_compare): Simplify by removing call to function
29170 extract_cond_operands.
29172 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29174 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
29175 (extract_cond_operands): Detect condition comparing against non-
29176 invariant bound and return appropriate enum value.
29177 (find_interesting_uses_cond): Update use of extract_cond_operands.
29178 Handle its return value accordingly.
29179 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
29181 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29183 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
29184 nonlinear iv_use computation in loop invariant sensitive way.
29186 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29188 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
29189 (find_iv_candidates): Call relate_compare_use_with_all_cands.
29191 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29193 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
29194 (dump_cand): Support iv_cand.inv_exprs.
29195 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
29197 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
29200 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29202 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
29204 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
29205 as local function. Include necessary header files.
29206 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
29208 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29210 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
29212 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29214 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
29215 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
29216 RSHIFT_EXPR and BIT_NOT_EXPR.
29218 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29220 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
29221 (adjust_setup_cost): New parameter supporting round up adjustment.
29222 (struct address_cost_data): Delete.
29223 (force_expr_to_var_cost): Don't bound cost with spill_cost.
29224 (split_address_cost, ptr_difference_cost): Delete.
29225 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
29226 (struct ainc_cost_data): New struct.
29227 (get_address_cost_ainc): New function.
29228 (get_address_cost, get_computation_cost): Reimplement.
29229 (determine_group_iv_cost_address): Record inv_expr for all uses of
29231 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
29232 (iv_ca_has_deps): Reimplemented to ...
29233 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
29235 (iv_ca_extend): Call iv_ca_more_deps.
29237 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29239 * tree-ssa-address.c (struct mem_address): Move to header file.
29240 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
29241 * tree-ssa-address.h (struct mem_address): Move from C file.
29242 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
29244 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29246 * tree-affine.h (aff_combination_type): New interface.
29247 (aff_combination_zero_p): Remove static.
29248 (aff_combination_const_p): New interface.
29249 (aff_combination_singleton_var_p): New interfaces.
29251 2017-05-11 Richard Biener <rguenther@suse.de>
29253 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
29254 Skip unreachable blocks and destinations.
29255 (eliminate): Move stmt removal and fixup ...
29256 (fini_eliminate): ... here. Skip inserted exprs.
29257 (pass_pre::execute): Move fini_pre after fini_eliminate.
29258 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
29259 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
29260 PRE to get rid of dead code that has invalid SSA form and
29261 split critical edges again.
29263 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29265 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29267 2017-05-11 Richard Biener <rguenther@suse.de>
29269 * passes.c (execute_function_todo): Verify loops if they are
29270 said to be up-to-date.
29271 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
29272 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
29274 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
29277 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
29278 handle calling assemble_external ourself.
29281 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
29282 modes with zero size. Enhance comment.
29284 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29286 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
29287 built-ins for vec_xl and vec_xst with short and char pointer
29290 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
29292 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
29293 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
29294 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
29295 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
29296 (_mm_maskz_min_round_ss): New intrinsics.
29297 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
29298 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
29299 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
29300 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
29301 (__builtin_ia32_minss_mask_round): New builtins.
29302 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29303 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29304 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
29306 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
29307 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
29309 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
29312 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
29314 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
29315 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
29316 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
29317 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
29318 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
29319 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
29320 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
29321 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29322 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
29323 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
29324 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
29325 (__builtin_ia32_mulss_mask_round): New builtins.
29326 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29327 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29328 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
29330 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
29331 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
29333 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
29336 2017-05-10 Julia Koval <julia.koval@intel.com>
29338 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
29339 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
29340 (_mm256_setr_m128i): New intrinsics.
29342 2017-05-10 Julia Koval <julia.koval@intel.com>
29344 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
29345 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
29346 (_mm_maskz_rcp14_ss): New intrinsics.
29347 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
29348 (__builtin_ia32_rcp14ss_mask): New builtins.
29349 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
29351 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
29353 PR tree-optimization/51513
29354 * tree-cfg.c (gimple_seq_unreachable_p): New function.
29355 (assert_unreachable_fallthru_edge_p): Use it.
29356 (group_case_labels_stmt): Likewise.
29357 * tree-cfg.h: Prototype it.
29358 * stmt.c: Include cfghooks.h and tree-cfg.h.
29359 (emit_case_dispatch_table) <gap_label>: New local variable.
29360 Use it to fill dispatch table gaps.
29361 Test for default_label before updating probabilities.
29362 (expand_case) <default_label>: Remove unneeded initialization.
29363 Test for unreachable default case statement and remove its edge.
29364 Set default_label accordingly.
29365 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
29367 2017-05-10 Carl Love <cel@us.ibm.com>
29369 * config/rs6000/rs6000-c: Add support for built-in functions
29370 vector signed char vec_neg (vector signed char)
29371 vector signed short int vec_neg (vector short int)
29372 vector signed int vec_neg (vector signed int)
29373 vector signed long long vec_neg (vector signed long long)
29374 vector float vec_neg (vector float)
29375 vector double vec_neg (vector double)
29376 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
29378 * config/rs6000/altivec.h: Add define for vec_neg
29379 * doc/extend.texi: Update the built-in documentation for the
29380 new built-in functions.
29382 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29384 PR tree-optimization/77644
29385 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
29387 2017-05-10 Nathan Sidwell <nathan@acm.org>
29389 * dumpfile.h (TDI_lang_all): New.
29390 (TDF_KIND): New. Renumber others
29391 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
29393 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
29395 (get_dump_file_name): Adjust suffix generation.
29396 (dump_enable_all): Use TDF_KIND.
29397 * doc/invoke.texi (-fdump-lang-all): Document.
29399 * dumpfile.h: Tabify.
29401 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
29404 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
29405 Move member access before delete.
29407 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
29409 * tree-inline.c (expand_call_inline): Split block at stmt
29412 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
29415 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
29416 are now unused after splitting mov{sf,sd}_hardfloat.
29417 (f32_lr2): Likewise.
29418 (f32_lm): Likewise.
29419 (f32_lm2): Likewise.
29420 (f32_li): Likewise.
29421 (f32_li2): Likewise.
29422 (f32_lv): Likewise.
29423 (f32_sr): Likewise.
29424 (f32_sr2): Likewise.
29425 (f32_sm): Likewise.
29426 (f32_sm2): Likewise.
29427 (f32_si): Likewise.
29428 (f32_si2): Likewise.
29429 (f32_sv): Likewise.
29430 (f32_dm): Likewise.
29431 (f32_vsx): Likewise.
29432 (f32_av): Likewise.
29433 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
29434 For movsf, order stores so the VSX stores occur before the GPR
29435 store which encourages the register allocator to use a traditional
29436 FPR instead of a GPR. For movsd, order the stores so that the GPR
29437 store comes before the VSX stores to allow the power6 to work.
29438 This is due to the power6 not having a 32-bit integer store
29439 instruction from a FPR.
29440 (movsf_hardfloat): Likewise.
29441 (movsd_hardfloat): Likewise.
29443 2017-05-09 Martin Sebor <msebor@redhat.com>
29445 PR translation/80280
29446 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
29449 PR translation/80280
29450 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
29451 data member added in r247778.
29452 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
29454 2017-05-09 Nathan Sidwell <nathan@acm.org>
29456 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
29458 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
29461 2017-05-09 Marek Polacek <polacek@redhat.com>
29463 * doc/invoke.texi: Fix typo.
29465 2017-05-09 Richard Biener <rguenther@suse.de>
29467 * tree-vrp.c (vrp_val_is_max): Adjust comment.
29468 (vrp_val_is_min): Likewise.
29469 (set_value_range_to_value): Likewise.
29470 (set_value_range_to_nonnegative): Likewise.
29471 (gimple_assign_nonzero_p): Likewise.
29472 (gimple_stmt_nonzero_p): Likewise.
29473 (vrp_int_const_binop): Likewise. Remove unreachable case.
29474 (adjust_range_with_scev): Adjust comments.
29475 (compare_range_with_value): Likewise.
29476 (extract_range_from_phi_node): Likewise.
29477 (test_for_singularity): Likewise.
29479 2017-05-09 Richard Biener <rguenther@suse.de>
29481 * tree-vrp.c (get_single_symbol): Add assert that we don't
29482 get overflowed constants as invariant part.
29483 (compare_values_warnv): Add comment before the TREE_NO_WARNING
29484 checks. Use wi::cmp instead of recursing for integer constants.
29485 (compare_values): Just ignore whether we assumed undefined
29486 overflow instead of failing the compare.
29487 (extract_range_for_var_from_comparison_expr): Add comment before the
29488 TREE_NO_WARNING sets.
29489 (test_for_singularity): Likewise.
29490 (extract_range_from_comparison): Do not disable optimization
29491 when we assumed undefined overflow.
29492 (extract_range_basic): Remove init of unused var.
29494 2017-05-09 Richard Biener <rguenther@suse.de>
29496 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
29497 (extract_range_from_multiplicative_op_1): Adjust.
29498 (extract_range_from_binary_expr_1): Use int_const_binop.
29500 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
29503 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
29504 rs6000_store_data_bypass_p in seven define_bypass directives and
29505 in several comments.
29506 * config/rs6000/rs6000-protos.h: Add prototype for
29507 rs6000_store_data_bypass_p function.
29508 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
29509 function implements slightly different (rs6000-specific) semantics
29510 than store_data_bypass_p, returning false rather than aborting
29511 with assertion error when arguments do not satisfy the
29512 requirements of store data bypass.
29513 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
29514 rs6000_store_data_bypass_p.
29516 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
29518 * config/xtensa/xtensa-protos.h
29519 (xtensa_initial_elimination_offset): New declaration.
29520 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29521 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
29522 macro definition, add case for FRAME_POINTER_REGNUM when
29523 FRAME_GROWS_DOWNWARD.
29524 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
29525 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
29526 xtensa_initial_elimination_offset.
29528 2017-05-08 Nathan Sidwell <nathan@acm.org>
29530 * doc/invoke.texi: Alphabetize -fdump options.
29532 2017-05-08 Martin Sebor <msebor@redhat.com>
29534 PR translation/80280
29535 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
29537 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
29539 * target.def (compute_frame_layout): New optional target hook.
29540 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
29541 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
29542 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
29544 * reload1.c (verify_initial_elim_offsets): Likewise.
29545 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
29546 (use_simple_return_p): Call arm_compute_frame_layout if needed.
29547 (arm_get_frame_offsets): Split up into this ...
29548 (arm_compute_frame_layout): ... and this function.
29550 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
29552 * config/aarch64/constraints.md (Usa): New constraint.
29553 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
29555 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
29557 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
29558 with_multilib_list after it has been checked.
29560 2017-05-08 Richard Biener <rguenther@suse.de>
29562 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
29563 (bitmap_set_subtract_values): Likewise.
29565 2017-05-08 Richard Biener <rguenther@suse.de>
29567 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
29568 (gimple_assign_nonzero): ... this and remove strict_overflow_p
29570 (gimple_stmt_nonzero_warnv_p): Rename to ...
29571 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
29573 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
29574 (extract_range_basic): Adjust, do not disable propagation on
29575 strict overflow sensitive simplification.
29576 (vrp_visit_cond_stmt): Likewise.
29578 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
29580 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
29581 body size unconditionally.
29583 2017-05-07 Jeff Law <law@redhat.com>
29586 2017-05-06 Jeff Law <law@redhat.com>
29587 PR tree-optimization/78496
29588 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
29591 PR tree-optimization/78496
29592 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
29593 (simplify_stmt_using_ranges): Call it.
29594 (vrp_dom_walker::before_dom_children): Extract equivalences
29595 from an ASSERT_EXPR with an equality comparison against a
29598 2017-05-06 Jeff Law <law@redhat.com>
29600 PR tree-optimization/78496
29601 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
29604 PR tree-optimization/78496
29605 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
29606 (simplify_stmt_using_ranges): Call it.
29607 (vrp_dom_walker::before_dom_children): Extract equivalences
29608 from an ASSERT_EXPR with an equality comparison against a
29611 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
29613 * lra-constraints.c (lra_copy_reg_equiv): New function.
29614 (split_reg): Use it to copy equivalence information from the
29615 original register to the spill register.
29617 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
29619 PR rtl-optimization/75964
29620 * simplify-rtx.c (simplify_const_relational_operation): Remove
29621 invalid handling of comparisons of integer ABS.
29623 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
29625 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
29626 initialize to zero.
29627 (init_regs): Remove declaration.
29628 (function_arg_advance_32): Initialize error_p as boolean variable.
29630 2017-05-05 Nathan Sidwell <nathan@acm.org>
29632 * store-motion.c (remove_reachable_equiv_notes): Reformat long
29633 lines. Use for (;;).
29635 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29637 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
29638 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
29639 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
29640 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
29641 VF=2 that require versioning.
29643 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29645 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
29648 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29650 * diagnostic.h (diagnostic_override_option_index): Convert from
29651 macro to inline function.
29653 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29655 * diagnostic.c (last_module_changed_p): New function.
29656 (set_last_module): New function.
29657 (diagnostic_report_current_module): Convert macro usage to
29658 the above functions.
29659 * diagnostic.h (diagnostic_context::last_module): Strengthen
29660 from const line_map * to const line_map_ordinary *.
29661 (diagnostic_last_module_changed): Delete macro.
29662 (diagnostic_set_last_module): Delete macro.
29664 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29666 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
29667 with diagnostic_report_diagnostic.
29668 (diagnostic_n_impl_richloc): Likewise.
29669 * diagnostic.h (report_diagnostic): Delete macro.
29670 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
29671 with diagnostic_report_diagnostic.
29672 * substring-locations.c (format_warning_va): Likewise.
29674 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29676 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
29677 save/restor of format_spec. Move option-printing code to...
29678 (print_option_information): ...this new function, and
29679 reimplement by simply printing to the pretty_printer,
29680 rather than appending to the format string.
29682 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29684 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
29685 handling logic into...
29686 (update_effective_level_from_pragmas): ...this new function.
29688 2017-05-04 Andrew Waterman <andrew@sifive.com>
29690 * config/riscv/riscv.opt (mstrict-align): New option.
29691 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
29692 (SLOW_UNALIGNED_ACCESS): Define.
29693 (riscv_slow_unaligned_access): Declare.
29694 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
29696 (riscv_slow_unaligned_access): New variable.
29697 (rocket_tune_info): Set slow_unaligned_access to true.
29698 (optimize_size_tune_info): Set slow_unaligned_access to false.
29699 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
29700 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
29701 (riscv_option_override): Set riscv_slow_unaligned_access.
29702 * doc/invoke.texi: Add -mstrict-align to RISC-V.
29704 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
29706 * config/riscv/riscv.md: Unify indentation.
29708 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29713 * config/rs6000/rs6000.md (u code attribute): Add FIX and
29715 (extendsi<mode>2): Add support for doing sign extension via
29716 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
29717 don't have ISA 3.0 instructions.
29718 (extendsi<mode>2 splitter): Likewise.
29719 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
29720 generate the normal insns since SImode can now go in vector
29721 registers. Disallow the special UNSPECs needed for previous
29722 machines to hide SImode being used. Add new insns
29723 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
29724 (fix_trunc<mode>si2_stfiwx): Likewise.
29725 (fix_trunc<mode>si2_internal): Likewise.
29726 (fixuns_trunc<mode>si2): Likewise.
29727 (fixuns_trunc<mode>si2_stfiwx): Likewise.
29728 (fctiw<u>z_<mode>_smallint): Likewise.
29729 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
29730 of floating point to 32-bit integer from doing a direct move to
29731 the GPR registers to do a store.
29732 (fctiwz_<mode>): Break long line.
29734 2017-05-05 Bin Cheng <bin.cheng@arm.com>
29736 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
29737 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
29738 (addr_list, addr_offset_valid_p): New.
29739 (split_address_groups): Check offset validity with above function.
29740 (gt-tree-ssa-loop-ivopts.h): Include header file.
29742 2017-05-05 Nathan Sidwell <nathan@acm.org>
29744 * config.gcc (arm*-*-*): Add missing 'fi'.
29746 2017-05-05 Steve Ellcey <sellcey@cavium.com>
29748 * doc/invoke.texi (-fopt-info): Explicitly say order of options
29749 included in -fopt-info does not matter.
29750 * doc/optinfo.texi (-fopt-info): Fix description of default
29751 behavour. Explicitly say order of options included in -fopt-info
29754 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29756 * config.gcc: Allow combinations of aprofile and rmprofile values for
29757 --with-multilib-list.
29758 * config/arm/t-multilib: New file.
29759 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
29760 variables. Remove setting of ISA and floating-point ABI in
29761 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
29762 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
29763 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
29764 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
29766 * config/arm/t-rmprofile: Likewise except for the matches changes.
29767 * doc/install.texi (--with-multilib-list): Document the combination of
29768 aprofile and rmprofile values and warn about pitfalls in doing that.
29770 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29772 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
29773 (movdi_aarch64): Likewise.
29775 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29777 PR tree-optimization/80632
29778 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
29780 (build_arrays): Initialize it for virtual phis.
29781 (fix_phi_nodes): Use it for virtual phis.
29783 PR tree-optimization/80558
29784 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
29785 [x, y] op z into [x op, y op z] for op & or | if conditions
29788 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
29789 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29792 * config/arm/arm.md (use_literal_pool): Remove.
29793 (64-bit immediate split): No longer takes cost into consideration
29794 if arm_disable_literal_pool is enabled.
29795 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
29796 used when arm_disable_literal_pool is enabled.
29797 (arm_max_const_double_inline_cost): Remove use of
29798 arm_disable_literal_pool.
29799 (push_minipool_fix): Add assert.
29800 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
29801 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
29802 (no_literal_pool_sf_immediate): New.
29804 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29806 PR tree-optimization/80613
29807 * tree-ssa-dce.c (propagate_necessity): Remove cases for
29808 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
29810 2017-05-05 Richard Biener <rguenther@suse.de>
29812 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
29814 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
29816 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
29817 of this flag from insn conditions due to removal from r247495.
29819 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29821 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
29823 (arm_early_store_addr_dep_ptr): Likewise.
29824 * config/arm/aarch-common-protos.h
29825 (arm_early_load_addr_dep_ptr): Add prototype.
29826 (arm_early_store_addr_dep_ptr): Likewise.
29827 * config/arm/cortex-a53.md: Add new bypasses.
29829 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29831 * tree.c (next_type_uid): Change type to unsigned.
29832 (type_hash_canon): Decrement back next_type_uid if
29833 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
29834 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
29837 2017-05-04 Martin Sebor <msebor@redhat.com>
29839 * builtins.c: Fix a trivial typo in a comment.
29841 PR middle-end/79234
29842 * builtins.c (check_sizes): Adjust to handle reading past the end.
29843 Avoid printing excessive upper bound of ranges. Use %E to print
29844 tree nodes instead of converting them to %wu.
29845 (expand_builtin_memchr): New function.
29846 (compute_dest_size): Rename...
29847 (compute_objsize): ...to this.
29848 (expand_builtin_memcpy): Adjust.
29849 (expand_builtin_mempcpy): Adjust.
29850 (expand_builtin_strcat): Adjust.
29851 (expand_builtin_strcpy): Adjust.
29852 (check_strncat_sizes): Adjust.
29853 (expand_builtin_strncat): Adjust.
29854 (expand_builtin_strncpy): Adjust and simplify.
29855 (expand_builtin_memset): Adjust.
29856 (expand_builtin_bzero): Adjust.
29857 (expand_builtin_memcmp): Adjust.
29858 (expand_builtin): Handle memcmp.
29859 (maybe_emit_chk_warning): Check strncat just once.
29861 2017-05-04 Martin Sebor <msebor@redhat.com>
29863 PR preprocessor/79214
29864 PR middle-end/79222
29865 PR middle-end/79223
29866 * builtins.c (check_sizes): Add inlining context and issue
29867 warnings even when -Wno-system-headers is set.
29868 (check_strncat_sizes): Same.
29869 (expand_builtin_strncat): Same.
29870 (expand_builtin_memmove): New function.
29871 (expand_builtin_stpncpy): Same.
29872 (expand_builtin): Handle memmove and stpncpy.
29874 2017-05-04 Bin Cheng <bin.cheng@arm.com>
29876 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
29877 which is not used any more.
29879 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29881 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
29883 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29885 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
29886 (cortexa53_tunings): Likewise.
29887 (cortexa57_tunings): Likewise.
29888 (cortexa72_tunings): Likewise.
29889 (cortexa73_tunings): Likewise.
29891 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29893 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
29894 Set loop alignment to 8.
29896 2017-05-04 Martin Sebor <msebor@redhat.com>
29898 PR translation/80280
29899 * builtins.c (expand_builtin_object_size): Add missing quoting to
29900 %D and like directives.
29901 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
29902 (hsa_type_for_tree_type): Same.
29903 (verify_function_arguments): Same.
29904 * symtab.c (symbol_table::change_decl_assembler_name): Same.
29905 * varasm.c (get_section): Same.
29908 2017-05-04 Martin Sebor <msebor@redhat.com>
29910 PR translation/80280
29911 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
29913 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29915 * config/aarch64/aarch64.c (generic_addrcost_table):
29916 Change HI/TI mode setting.
29918 2017-05-04 Martin Jambor <mjambor@suse.cz>
29920 PR tree-optimization/80622
29921 * tree-sra.c (comes_initialized_p): New function.
29922 (build_accesses_from_assign): Only set write lazily when
29923 comes_initialized_p is false.
29924 (analyze_access_subtree): Use comes_initialized_p.
29925 (propagate_subaccesses_across_link): Assert !comes_initialized_p
29926 instead of testing for PARM_DECL.
29928 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29930 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
29931 constraint on operand 0 to allow more general addressing modes.
29932 Adjust output template.
29933 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
29935 * config/aarch64/aarch64-protos.h
29936 (aarch64_address_valid_for_prefetch_p): Declare prototype.
29937 * config/aarch64/constraints.md (Dp): New address constraint.
29938 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
29941 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
29943 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
29944 update use of estimate_ipcp_clone_size_and_time.
29945 (estimate_local_effects): Update use of
29946 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
29947 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
29948 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
29949 Return nonspecialized time.
29951 2017-05-04 Richard Biener <rguenther@suse.de>
29953 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
29954 for the last VUSE which def dominates the PHI. Directly call
29956 (get_continuation_for_phi_1): Remove.
29958 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
29960 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
29961 to explain the use of truncating division. Cap the number of
29962 iterations to the maximum given by nb_iterations_upper_bound,
29965 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
29967 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
29968 * configure: Regenerate.
29969 * config.in: Regenerate.
29970 * config/i386/driver-mingw32.c: new file.
29971 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
29972 * config.host: Link driver-mingw32.o on MinGW host.
29973 * doc/install.texi: Document new --enable-mingw-wildcard configure
29976 2017-05-04 Marek Polacek <polacek@redhat.com>
29978 PR tree-optimization/80612
29979 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
29981 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29982 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
29984 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
29985 (movt splitter): Likewise.
29986 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
29987 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
29988 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
29989 block for Thumb-1 with MOVT.
29990 (thumb2_legitimate_address_p): Move code block ...
29991 (can_avoid_literal_pool_for_label_p): ... into this new function.
29992 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
29994 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
29995 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
29996 "M-profile targets with the MOVT instruction".
29998 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30000 * config/arm/arm-builtins.c (arm_init_builtins): Rename
30001 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
30002 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
30004 2017-05-04 Martin Liska <mliska@suse.cz>
30006 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
30007 variable cond_code.
30009 2017-05-04 Richard Biener <rguenther@suse.de>
30011 * tree.c (array_at_struct_end_p): Handle arrays at struct
30012 end with flexarrays more conservatively. Refactor and treat
30013 arrays of arrays or aggregates more strict. Fix
30014 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
30015 * tree.c (array_at_struct_end_p): Adjust prototype.
30016 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
30017 * gimple-fold.c (get_range_strlen): Likewise.
30018 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
30020 2017-05-04 Richard Biener <rguenther@suse.de>
30022 PR tree-optimization/31130
30023 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
30025 (supports_overflow_infinity): Likewise.
30026 (is_negative_overflow_infinity): Likewise.
30027 (is_positive_overflow_infinity): Likewise.
30028 (is_overflow_infinity): Likewise.
30029 (stmt_overflow_infinity): Likewise.
30030 (overflow_infinity_range_p): Likewise.
30031 (usable_range_p): Remove as always returning true.
30032 (make_overflow_infinity): Remove.
30033 (negative_overflow_infinity): Likewise.
30034 (positive_overflow_infinity): Likewise.
30035 (avoid_overflow_infinity): Likewise.
30036 (set_value_range): Adjust accordingly.
30037 (set_value_range_to_nonnegative): Likewise, remove now unused
30038 overflow_infinity arg.
30039 (vrp_operand_equal_p): Adjust.
30040 (update_value_range): Likewise.
30041 (range_int_cst_singleton_p): Likewise.
30042 (operand_less_p): Likewise.
30043 (compare_values_warnv): Likewise.
30044 (extract_range_for_var_from_comparison_expr): Likewise.
30045 (vrp_int_const_binop): Likewise.
30046 (zero_nonzero_bits_from_vr): Likewise.
30047 (extract_range_from_multiplicative_op_1): Likewise.
30048 (extract_range_from_binary_expr_1): Likewise.
30049 (extract_range_from_unary_expr): Likewise.
30050 (extract_range_from_comparison): Likewise.
30051 (extract_range_basic): Likewise.
30052 (adjust_range_with_scev): Likewise.
30053 (compare_ranges): Likewise.
30054 (compare_range_with_value): Likewise.
30055 (dump_value_range): Likewise.
30056 (test_for_singularity): Likewise, remove strict_overflow_p parameter
30058 (simplify_cond_using_ranges): Adjust.
30060 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
30062 * brig-builtins.def: Added a builtin for class_f64.
30063 * builtin-types.def: Added a builtin type needed by class_f64.
30065 2017-05-03 Jason Merrill <jason@redhat.com>
30067 * timevar.def: Add TV_CONSTEXPR.
30069 2017-05-03 David Malcolm <dmalcolm@redhat.com>
30071 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
30073 2017-05-03 Martin Jambor <mjambor@suse.cz>
30075 * ipa-prop.c (ipa_update_after_lto_read): Removed.
30076 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
30077 * ipa-cp.c (ipcp_propagate_stage): Do not call
30078 ipa_update_after_lto_read.
30079 * ipa-inline.c (ipa_inline): Likewise.
30081 2017-05-03 Martin Jambor <mjambor@suse.cz>
30083 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
30084 tag. Added a default constructor and a destructor.
30085 (ipa_edge_args_sum_t): New class;
30086 (ipa_edge_args_sum): Declare.
30087 (ipa_edge_args_vector): Remove declaration.
30088 (IPA_EDGE_REF): Use ipa_edge_args_sum.
30089 (ipa_free_edge_args_substructures): Remove declaration.
30090 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
30091 (ipa_edge_args_info_available_for_edge_p): Likewise.
30092 * ipa-prop.c (ipa_edge_args_vector): Removed.
30093 (edge_removal_hook_holder): Likewise.
30094 (edge_duplication_hook_holder): Likewise.
30095 (ipa_edge_args_sum): New variable.
30096 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
30097 ipa_edge_args_vector.
30098 (ipa_free_edge_args_substructures): Likewise.
30099 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
30100 ipa_edge_args_vector.
30101 (ipa_edge_removal_hook): Turned into method
30102 ipa_edge_args_sum_t::remove.
30103 (ipa_edge_duplication_hook): Turned into method
30104 ipa_edge_args_sum_t::duplicate.
30105 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
30106 registering edge hooks.
30107 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
30108 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
30109 ipa_edge_args_sum instead of ipa_edge_args_vector.
30110 * ipa-profile.c (ipa_profile): Likewise.
30112 2017-05-03 Martin Jambor <mjambor@suse.cz>
30114 * symbol-summary.h (function_summary): New method exists.
30115 (function_summary::symtab_removal): Deallocate through release.
30116 (call_summary): New class.
30117 (gt_ggc_mx): New overload.
30118 (gt_pch_nx): Likewise.
30119 (gt_pch_nx): Likewise.
30121 2017-05-03 Jeff Law <law@redhat.com>
30123 PR tree-optimization/78496
30124 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
30125 from simplify_cond_using_ranges. Split off code to walk
30126 backwards through casts into ...
30127 (simplify_cond_using_ranges_2): New function.
30128 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
30129 (execute_vrp): After identifying jump threads, call
30130 simplify_cond_using_ranges_2.
30132 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
30135 * ipa-inline.h (inline_summary): Add ctor.
30136 (create_ggc): Do not use ggc_cleared_alloc.
30138 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
30139 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
30141 * gcc.c (handle_braces): Support escaping in switch matching
30143 * doc/invoke.texi (Spec Files): Document it.
30144 Remove superfluous @code markup in items.
30146 2017-05-03 David Malcolm <dmalcolm@redhat.com>
30148 * diagnostic-show-locus.c (struct column_range): New struct.
30149 (get_affected_columns): New function.
30150 (get_printed_columns): New function.
30151 (struct correction): New struct.
30152 (correction::ensure_capacity): New function.
30153 (correction::ensure_terminated): New function.
30154 (struct line_corrections): New struct.
30155 (line_corrections::~line_corrections): New dtor.
30156 (line_corrections::add_hint): New function.
30157 (layout::print_trailing_fixits): Reimplement in terms of the new
30159 (selftest::test_overlapped_fixit_printing): New function.
30160 (selftest::diagnostic_show_locus_c_tests): Call it.
30162 2017-05-03 Nathan Sidwell <nathan@acm.org>
30164 Canonicalize canonical type hashing
30165 * tree.h (type_hash_canon_hash): Declare.
30166 * tree.c (type_hash_list, attribute_hash_list): Move into
30167 type_hash_canon_hash.
30168 (build_type_attribute_qual_variant): Break out hash code calc into
30169 type_hash_canon_hash.
30170 (type_hash_canon_hash): New. Generic type hash computation.
30171 (build_range_type_1, build_array_type_1, build_function_type,
30172 build_method_type_directly, build_offset_type, build_complex_type,
30173 make_vector_type): Call it.
30175 2017-05-03 Richard Biener <rguenther@suse.de>
30177 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
30178 When all DRs have unknown misaligned do not always peel
30179 when there is a store but apply the same costing model as if
30180 there were only loads.
30182 2017-05-03 Richard Biener <rguenther@suse.de>
30185 PR tree-optimization/80492
30186 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30187 compare_base_decls returning dont-know properly.
30189 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
30191 * config/arm/iterators.md (CCSI): New mode iterator.
30192 (arch): New mode attribute.
30193 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
30194 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
30195 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
30196 code iterator for success result mode.
30197 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
30198 the corresponding new insn generators.
30200 2017-05-03 Bin Cheng <bin.cheng@arm.com>
30203 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30204 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30206 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
30208 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
30209 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
30210 (DDR_A): Wrap DDR argument in brackets.
30211 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
30212 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
30213 (DDR_REVERSED_P): Likewise.
30215 2017-05-03 Jakub Jelinek <jakub@redhat.com>
30217 PR tree-optimization/79472
30218 * tree-switch-conversion.c (struct switch_conv_info): Add
30219 contiguous_range and default_case_nonstandard fields.
30220 (collect_switch_conv_info): Compute contiguous_range and
30221 default_case_nonstandard fields, don't clear final_bb if
30222 contiguous_range and only the default case doesn't have the required
30224 (check_all_empty_except_final): Set default_case_nonstandard instead
30225 of failing if contiguous_range and the default case doesn't have empty
30227 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
30228 and only the default case doesn't have the required constants. Skip
30230 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
30231 if default_case_nonstandard.
30232 (build_constructors): Build constant 1 just once. Assert that default
30233 values aren't inserted in between cases if contiguous_range. Skip
30235 (build_arrays): Skip virtual phis.
30236 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
30237 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
30238 Handle virtual phis.
30239 (gen_inbound_check): Handle default_case_nonstandard case.
30240 (process_switch): Adjust check_final_bb caller. Call
30241 gather_default_values with the first non-default case instead of
30242 default case if default_case_nonstandard.
30244 2017-05-02 Nathan Sidwell <nathan@acm.org>
30246 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
30247 check. Fix formatting.
30249 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
30251 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
30252 errors when comparing specialized and unspecialized times.
30254 2017-05-02 David Malcolm <dmalcolm@redhat.com>
30256 * diagnostic-show-locus.c
30257 (layout::should_print_annotation_line_p): Make private.
30258 (layout::print_annotation_line): Make private.
30259 (layout::annotation_line_showed_range_p): Make private.
30260 (layout::show_ruler): Make private.
30261 (layout::print_source_line): Make private. Pass in line and
30262 line_width, rather than calling location_get_source_line. Drop
30264 (layout::print_leading_fixits): New method.
30265 (layout::print_any_fixits): Rename to...
30266 (layout::print_trailing_fixits): ...this, and make private.
30267 Don't print newline fixits.
30268 (diagnostic_show_locus): Move logic for printing one row into...
30269 (layout::print_line): ...this new function. Move the
30270 location_get_source_line call and error-handling from
30271 print_source_line to here. Call print_leading_fixits, and rename
30272 print_any_fixits to print_trailing_fixits.
30273 (selftest::test_fixit_insert_containing_newline): Update now that
30274 newlines are partially supported.
30275 (selftest::test_fixit_insert_containing_newline_2): New test.
30276 (selftest::test_fixit_replace_containing_newline): Update comments.
30277 (selftest::diagnostic_show_locus_c_tests): Call the new test.
30278 * edit-context.c (class added_line): New class.
30279 (class edited_line): Describe newline handling in comment.
30280 (edited_line::actually_edited_p): New method.
30281 (edited_line::print_content): Delete redundant decl.
30282 (edited_line::m_predecessors): New field.
30283 (edited_file::print_content): Call edited_line::print_content.
30284 (edited_file::print_diff): Update to support newlines.
30285 (edited_file::print_diff_hunk): Likewise.
30286 (edited_file::print_run_of_changed_lines): New function.
30287 (edited_file::print_diff_line): Convert to...
30288 (print_diff_line): ...this.
30289 (edited_file::get_effective_line_count): New function.
30290 (edited_line::edited_line): Initialize new field m_predecessors.
30291 (edited_line::~edited_line): Clean up m_predecessors.
30292 (edited_line::apply_fixit): Handle newlines.
30293 (edited_line::get_effective_line_count): New function.
30294 (edited_line::print_content): New function.
30295 (edited_line::print_diff_lines): New function.
30296 (selftest::test_applying_fixits_insert_containing_newline): New
30298 (selftest::test_applying_fixits_replace_containing_newline): New
30300 (selftest::insert_line): New function.
30301 (selftest::test_applying_fixits_multiple_lines): Add example of
30303 (selftest::edit_context_c_tests): Call the new tests.
30305 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30307 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
30308 parameter cand. Update dump information.
30309 (get_computation_cost): Update uses.
30311 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30313 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
30314 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
30315 (get_computation_at, rewrite_use_address): Update use of
30316 get_computation_aff.
30318 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30320 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
30321 (get_computation): Delete.
30322 (get_computation_cost): Implement like get_computation_cost_at.
30323 Use get_computation_at.
30324 (get_computation_cost_at): Delete.
30325 (rewrite_use_nonlinear_expr): Use get_computation_at.
30326 (rewrite_use_compare, remove_unused_ivs): Ditto.
30328 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30330 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
30332 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30334 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
30335 (ivopts_global_cost_for_size): Rename parameter and update uses.
30336 (iv_ca_recount_cost): Update uses.
30337 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
30338 candidates seperately in n_invs and n_cands.
30339 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
30341 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30343 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
30344 (find_inv_vars_cb): New.
30345 (find_depends): Renamed to ...
30346 (find_inv_vars): ... this.
30347 (add_candidate_1, force_var_cost): Call find_inv_vars.
30348 (split_address_cost, determine_group_iv_cost_cond): Ditto.
30350 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30352 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
30353 inv_vars. Add inv_exprs.
30354 (struct iv_cand): Rename depends_on to inv_vars.
30355 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
30356 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
30357 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
30358 (dump_cand): Dump inv_vars.
30359 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
30360 (record_invariant, find_depends, add_candidate_1): Ditto.
30361 (set_group_iv_cost, force_var_cost): Ditto.
30362 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
30363 (get_computation_cost_at, get_computation_cost): Ditto.
30364 (determine_group_iv_cost_generic): Ditto.
30365 (determine_group_iv_cost_address): Ditto.
30366 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
30367 (determine_group_iv_costs): Ditto.
30368 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
30369 (iv_ca_set_remove_invariants): Renamed to ...
30370 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
30371 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
30372 (iv_ca_set_add_invariants): Renamed to ...
30373 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
30374 (iv_ca_set_cp): Use iv_ca_set_add_invs.
30375 (iv_ca_has_deps): Support inv_vars and inv_exprs.
30376 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
30377 (create_new_ivs): Remove useless dump.
30379 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30381 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
30383 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
30384 (iv_ca_set_no_cp, create_new_iv): Ditto.
30386 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30388 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30390 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
30392 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
30393 function tree_check2.
30395 2017-05-02 Martin Liska <mliska@suse.cz>
30397 * doc/gcov.texi: Add missing preposition.
30398 * gcov.c (function_info::function_info): Properly fill up
30399 all member variables.
30401 2017-05-02 Tamar Christina <tamar.christina@arm.com>
30403 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
30405 2017-05-02 Tamar Christina <tamar.christina@arm.com>
30407 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
30409 2017-05-02 Martin Liska <mliska@suse.cz>
30412 * lto-streamer-in.c (lto_read_tree_1): Remove
30413 LTO_STREAMER_DEBUG.
30414 * lto-streamer.c (struct tree_hash_entry): Likewise.
30415 (struct tree_entry_hasher): Likewise.
30416 (tree_entry_hasher::hash): Likewise.
30417 (tree_entry_hasher::equal): Likewise.
30418 (lto_streamer_init): Likewise.
30419 (lto_orig_address_map): Likewise.
30420 (lto_orig_address_get): Likewise.
30421 (lto_orig_address_remove): Likewise.
30422 * lto-streamer.h: Likewise.
30423 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
30424 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
30426 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
30428 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
30429 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
30430 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
30431 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
30432 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
30433 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
30434 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
30435 (mm_maskz_sub_ss): New intrinsics.
30436 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30437 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
30438 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
30439 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
30440 (__builtin_ia32_subss_mask_round): New builtins.
30441 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30442 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30443 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
30445 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
30446 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
30448 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
30451 2017-05-02 Martin Jambor <mjambor@suse.cz>
30453 PR tree-optimization/78687
30454 * tree-sra.c (access): New field parent.
30455 (process_subtree_disqualification): New function.
30456 (disqualify_candidate): Call it.
30457 (build_accesses_from_assign): Reset write flag if creating an
30459 (build_access_subtree): Fill in parent field and also prpagate
30460 down grp_write flag.
30461 (create_artificial_child_access): New parameter set_grp_write, set
30462 grp_write to its value.
30463 (propagate_subaccesses_across_link): Also propagate grp_write flag
30465 (propagate_all_subaccesses): Push the closest parent back to work
30466 queue if add_access_to_work_queue returned true.
30468 2017-05-02 Richard Biener <rguenther@suse.de>
30470 * common.opt (fstrict-overflow): Alias negative to fwrapv.
30471 * doc/invoke.texi (fstrict-overflow): Remove all traces of
30472 -fstrict-overflow documentation.
30473 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
30474 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
30475 flag_strict_overflow.
30476 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
30477 * lto-opts.c (lto_write_options): Do not stream it.
30478 * lto-wrapper.c (merge_and_complain): Do not handle it.
30479 * opts.c (default_options_table): Do not set -fstrict-overflow.
30480 (finish_options): Likewise do not clear it when sanitizing.
30481 * simplify-rtx.c (simplify_const_relational_operation): Do not
30482 test flag_strict_overflow.
30484 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
30486 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
30487 using enabled attribute.
30488 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
30489 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
30490 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
30491 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
30492 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
30493 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
30494 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
30495 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
30496 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
30497 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
30499 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
30501 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
30503 2017-05-02 Richard Biener <rguenther@suse.de>
30505 PR tree-optimization/80591
30507 2017-04-10 Richard Biener <rguenther@suse.de>
30509 * tree-ssa-structalias.c (find_func_aliases): Properly handle
30512 2017-05-02 Richard Biener <rguenther@suse.de>
30514 PR tree-optimization/80549
30515 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
30516 (cleanup_tree_cfg_noloop): Create forwarders to known loop
30517 headers if they do not have a preheader.
30519 2017-05-02 Martin Liska <mliska@suse.cz>
30522 * common.opt: Fix typo.
30523 * doc/invoke.texi: Likewise.
30525 2017-05-01 Jan Beulich <jbeulich@suse.com>
30527 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
30528 swapping, add (x,x,m,x,n) alternative.
30530 2017-05-01 Nathan Sidwell <nathan@acm.org>
30532 * calls.c (combine_pending_stack_adjustment_and_call): Remove
30533 unnecessary unadjusted_alignment check.
30535 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
30538 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
30540 * gimplify.c (gimplify_cilk_detach): New function.
30541 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
30542 * tree-core.h: Document EXPR_CILK_SPAWN.
30543 * tree.h (EXPR_CILK_SPAWN): Define.
30545 2017-05-01 David Malcolm <dmalcolm@redhat.com>
30547 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
30548 to use new fixit_hint representation, using the "replace" logic.
30549 (get_line_span_for_fixit_hint): Likewise.
30550 (layout::print_any_fixits): Likewise.
30551 (selftest::test_one_liner_many_fixits): Rename to...
30552 (selftest::test_one_liner_many_fixits_1): ...this, and update
30553 comment and expected output to reflect that the multiple fix-it
30554 hints are now consolidated into one insertion.
30555 (selftest::test_one_liner_many_fixits_2): New test.
30556 (selftest::test_diagnostic_show_locus_one_liner): Update for
30558 (selftest::test_fixit_consolidation): Update for fix-it API
30560 * diagnostic.c (print_parseable_fixits): Likewise.
30561 * edit-context.c (edited_line::m_line_events): Convert from
30562 auto_vec <line_event *> to auto_vec <line_event>.
30563 (class line_event): Convert from abstract base class to a concrete
30564 class, taking over the role of replace_event.
30565 (class insert_event): Delete.
30566 (class replace_event): Rename to class line_event. Convert to
30568 (edit_context::add_fixits): Reimplement.
30569 (edit_context::apply_insert): Delete.
30570 (edit_context::apply_replace): Rename to...
30571 (edit_context::apply_fixit): ...this. Convert to half-open range.
30572 (edited_file::apply_insert): Delete.
30573 (edited_file::apply_replace): Rename to...
30574 (edited_file::apply_fixit): ...this.
30575 (edited_line::~edited_line): Drop deletion of events.
30576 (edited_line::apply_insert): Delete.
30577 (edited_line::apply_replace): Rename to...
30578 (edited_line::apply_fixit): ...this. Convert to half-open range.
30579 Update for change to type of m_line_events.
30580 * edit-context.h (edit_context::apply_insert): Delete.
30581 (edit_context::apply_replace): Rename to...
30582 (edit_context::apply_fixit): ...this.
30584 2017-05-01 Martin Sebor <msebor@redhat.com>
30586 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
30589 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
30592 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
30593 __get_cpuid_max returns 0.
30594 (__get_cpuid_count): Ditto.
30596 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
30598 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
30599 replacement expression is another instance of one of its arguments.
30601 2017-05-01 Jakub Jelinek <jakub@redhat.com>
30604 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
30605 check for stack push/pop autoinc.
30606 * config/i386/i386.c (ix86_agi_dependent): Return false
30607 if the only reason why modified_in_p returned true is that
30608 addr is SP based and set_insn is a push or pop.
30610 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
30612 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
30615 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
30618 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
30619 (account_size_time): Use two predicates - exec_pred and
30621 (evaluate_conditions_for_known_args): Compute both clause and
30623 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
30624 (inline_summary_t::duplicate): Update.
30625 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
30627 (compute_inline_parameters): Likewise.
30628 (estimate_edge_size_and_time): Update caluclation of time.
30629 (estimate_node_size_and_time): Compute both time and nonspecialized
30631 (estimate_ipcp_clone_size_and_time): Update.
30632 (inline_merge_summary): Update.
30633 (do_estimate_edge_time): Update.
30634 (do_estimate_edge_size): Update.
30635 (do_estimate_edge_hints): Update.
30636 (inline_read_section, inline_write_summary): Stream both new predicates.
30637 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
30639 (compute_inlined_call_time): Cleanup.
30640 (big_speedup_p): Update.
30641 (edge_badness): Update.
30642 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
30643 (size_time_entry): Replace predicate by exec_predicate and
30644 nonconst_predicate.
30645 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
30646 (estimate_edge_time): Return also nonspec_time.
30647 (reset_edge_growth_cache): Update.
30649 2017-04-29 Jakub Jelinek <jakub@redhat.com>
30651 PR rtl-optimization/80491
30652 * ifcvt.c (noce_process_if_block): When looking for x setter
30653 with missing else_bb, don't check only the insn right before
30654 cond_earliest, but look for the last insn that x is modified in
30655 within the same bb.
30657 PR rtl-optimization/80491
30658 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
30660 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
30662 PR tree-optimization/80487
30663 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
30665 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30667 PR tree-optimization/79697
30668 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
30669 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
30670 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
30672 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
30673 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
30675 2017-04-28 Martin Sebor <msebor@redhat.com>
30677 PR tree-optimization/80523
30678 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
30679 (init_target_to_host_charmap, target_to_host, target_strtol10): New
30681 (maybe_warn, format_directive, parse_directive): Use new functions.
30682 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
30684 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
30686 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
30688 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
30690 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
30691 target_header_dir): Set correctly.
30692 * configure: Regenerated.
30693 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30694 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30695 instead of SYSTEM_HEADER_DIR.
30697 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
30699 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
30700 (estimate_local_effects): Likewise.
30701 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
30702 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
30703 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
30704 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
30705 do_estimate_edge_time, estimate_edge_time): Likewise.
30706 * ipa-inline-analysis.c (estimate_node_size_and_time,
30707 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
30708 (estimate_time_after_inlining): Remove.
30710 2017-04-28 Martin Liska <mliska@suse.cz>
30712 * doc/gcov.texi: Enhance documentation of gcov.
30714 2017-04-28 Martin Liska <mliska@suse.cz>
30716 * doc/gcov.texi: Sort options in alphabetic order.
30717 * doc/gcov-dump.texi: Likewise.
30718 * doc/gcov-tool.texi: Likewise.
30719 * gcov.c (print_usage): Likewise.
30720 * gcov-dump.c (print_usage): Likewise.
30721 * gcov-tool.c (print_merge_usage_message): Likewise.
30722 (print_rewrite_usage_message): Likewise.
30723 (print_overlap_usage_message): Likewise.
30725 2017-04-28 Martin Liska <mliska@suse.cz>
30727 PR gcov-profile/53915
30728 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
30730 2017-04-28 Martin Liska <mliska@suse.cz>
30732 PR gcov-profile/79891
30733 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
30734 is marked by compiler as living on a line.
30735 (get_cycles_count): Remove usage of the union.
30736 (output_intermediate_file): Likewise.
30737 (find_source): Fix GNU coding style.
30738 (accumulate_line_counts): Remove old non-all block mode.
30739 (output_lines): Remove usage of the union.
30740 * profile.c (output_location): Include all BBs, even if
30741 belonging to a same line (and file) as a previous BB.
30743 2017-04-28 Martin Liska <mliska@suse.cz>
30745 * gcov.c (process_args): Handle new argument 'w'.
30746 (read_graph_file): Assign ID to BBs.
30747 (output_branch_count): Display BB # if verbose flag is set.
30748 (output_lines): Likewise for arcs.
30749 (print_usage): Add '--verbose' option help.
30750 * doc/gcov.texi: Document --verbose (-w) option.
30752 2017-04-28 Martin Liska <mliska@suse.cz>
30754 * gcov.c (struct block_location_info): New struct.
30755 (process_file): Fill up the new structure.
30756 (read_graph_file): Replace usage of encoding by the newly added
30758 (add_line_counts): Likewise.
30759 (accumulate_line_counts): Remove usage of the union.
30760 (function_info::function_info): New function.
30761 (function_info::~function_info): Likewise.
30762 (process_file): Call delete instead of release_function.
30763 (release_function): Release the function.
30764 (release_structures): Call delete instead of release_function.
30765 (solve_flow_graph): Replace usage of num_blocks.
30766 (find_exception_blocks): Likewise.
30767 (output_lines): Fix GNU coding style.
30769 2017-04-28 Martin Liska <mliska@suse.cz>
30772 * coverage.c (coverage_remove_note_file): New function.
30773 * coverage.h: Declare the function.
30774 * toplev.c (finalize): Clean if an error has been seen.
30776 2017-04-28 Martin Liska <mliska@suse.cz>
30778 PR gcov-profile/80031
30779 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
30780 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
30781 * gcov.c (read_graph_file): Read just number of blocks.
30782 * profile.c (branch_prob): Do not stream 0 flags per a basic
30785 2017-04-28 Martin Liska <mliska@suse.cz>
30787 * gcov-dump.c (tag_*): Add new argument to declarations.
30788 (dump_gcov_file): Likewise.
30789 (tag_blocks): Add and use new argument depth.
30790 (tag_arcs): Likewise.
30791 (tag_lines): Likewise.
30792 (tag_counters): Likewise.
30793 (tag_summary): Likewise.
30794 (dump_working_sets): Use depth to do a proper indentation.
30796 2017-04-28 Jakub Jelinek <jakub@redhat.com>
30799 * cgraph.h (symtab_node::debug_symtab): No longer inline.
30800 * symtab.c (symtab_node::debug_symtab): Move definition here.
30802 2017-04-28 Richard Biener <rguenther@suse.de>
30804 * lto-streamer.h (LTO_major_version): Bump to 7.
30806 2017-04-28 Richard Biener <rguenther@suse.de>
30808 * tree-vrp.c (assert_info): New struct.
30809 (add_assert_info): New helper.
30810 (register_edge_assert_for_2): Refactor to add asserts to a vector
30812 (register_edge_assert_for_1): Likewise.
30813 (register_edge_assert_for): Likewise.
30814 (finish_register_edge_assert_for): New helper actually registering
30815 asserts where live on edge.
30816 (find_conditional_asserts): Adjust.
30817 (find_switch_asserts): Likewise.
30818 (evrp_dom_walker::try_find_new_range): Generalize.
30819 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
30821 2017-04-27 Marek Polacek <polacek@redhat.com>
30824 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
30825 arg10 and arg11 to itype.
30827 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
30829 * doc/extend.texi (Object Size Checking): Improve grammar.
30831 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
30834 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
30835 that the logic for permitting reciprocal estimates matches that
30838 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30841 * tree.c (type_cache_hasher::equal): Only compare
30842 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
30843 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
30844 non-aggregate element types.
30845 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
30846 about the flag on ARRAY_TYPEs in the comment, formatting fix.
30848 2017-04-27 Richard Biener <rguenther@suse.de>
30850 PR middle-end/80533
30851 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
30852 stripping ARRAY_REFs from MEM_EXPR make sure we're not
30853 keeping a reference to a trailing array.
30855 2017-04-27 Richard Biener <rguenther@suse.de>
30857 PR middle-end/80539
30858 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
30859 being in loop-closed SSA form conservatively.
30860 (chrec_fold_multiply_poly_poly): Likewise.
30862 2017-04-27 Tamar Christina <tamar.christina@arm.com>
30864 PR middle-end/79665
30865 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
30866 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
30868 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30871 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
30872 (aarch64_function_arg_alignment): Return unsigned int again, but still
30873 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
30874 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30875 Don't emit -Wpsabi note.
30876 (aarch64_function_arg_boundary): Likewise.
30877 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30880 2017-04-26 Nathan Sidwell <nathan@acm.org>
30882 * tree.h (crc32_unsigned_n): Declare.
30883 (crc32_unsigned, crc32_unsigned): Make inline.
30884 * tree.c (crc32_unsigned_bits): Replace with ...
30885 (crc32_unsigned_n): ... this.
30886 (crc32_unsigned, crc32_byte): Remove.
30887 (crc32_string): Remove unnecessary braces.
30889 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30891 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
30892 * ipa-inline-analysis.c (MAX_TIME): Remove.
30893 (account_size_time): Use sreal for time.
30894 (dump_inline_summary): Update.
30895 (estimate_function_body_sizes): Update.
30896 (estimate_edge_size_and_time): Update.
30897 (estimate_calls_size_and_time): Update.
30898 (estimate_node_size_and_time): Update.
30899 (inline_merge_summary): Update.
30900 (inline_update_overall_summary): Update.
30901 (estimate_time_after_inlining): Update.
30902 (inline_read_section): Update.
30903 (inline_write_summary): Update.
30904 * ipa-inline.c (compute_uninlined_call_time): Update.
30905 (compute_inlined_call_time): Update.
30906 (recursive_inlining): Update.
30907 (inline_small_functions): Update.
30908 (dump_overall_stats): Update.
30909 * ipa-inline.h: Include sreal.h.
30910 (size_time_entry): Turn time to sreal.
30911 (inline_summary): Turn self_time nad time to sreal.
30913 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30915 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
30917 (sreal::stream_out, sreal::stream_in): New.
30918 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
30920 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30922 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
30925 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
30928 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
30930 (dimode_scalar_chain::compute_convert_gain): Ditto.
30931 (dimode_scalar_chain::make_vector_copies): Ditto.
30932 (dimode_scalar_chain::convert_reg): Ditto.
30933 (dimode_scalar_chain::convert_insn): Ditto.
30934 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
30935 (VI248_AVX512BW_1): New mode iterator.
30936 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
30937 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
30940 2017-04-25 Martin Sebor <msebor@redhat.com>
30942 PR tree-optimization/80497
30943 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
30944 constants are representable in HOST_WIDE_INT.
30945 (parse_directive): Ditto.
30947 2017-04-25 Martin Sebor <msebor@redhat.com>
30950 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
30951 (new_zero_array): Adjust signature.
30952 (dom_info::dom_init): Used unsigned rather that size_t.
30953 (dom_info::dom_info): Same.
30955 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30956 Jakub Jelinek <jakub@redhat.com>
30959 * config/arm/arm.c: Include gimple.h.
30960 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30961 returns negative, increment ncrn only if it returned positive.
30962 (arm_needs_doubleword_align): Return int instead of bool,
30963 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
30964 members, but if there is any such non-FIELD_DECL
30965 > PARM_BOUNDARY aligned decl, return -1 instead of false.
30966 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30967 returns negative, increment nregs only if it returned positive.
30968 (arm_setup_incoming_varargs): Likewise.
30969 (arm_function_arg_boundary): Emit -Wpsabi note if
30970 arm_needs_doubleword_align returns negative, return
30971 DOUBLEWORD_ALIGNMENT only if it returned positive.
30973 2017-04-25 Marek Polacek <polacek@redhat.com>
30976 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30977 first argument to type.
30979 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
30982 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
30983 type checks to test for compatibility instead of equality.
30985 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30986 Jakub Jelinek <jakub@redhat.com>
30989 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
30991 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
30992 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
30993 the alignment computation, but return their maximum in warn_alignment.
30994 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30995 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
30997 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
30998 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
31001 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31003 * config/arc/simdext.md (dmpyh): Fix typo.
31005 2017-04-25 Richard Biener <rguenther@suse.de>
31007 PR tree-optimization/80492
31008 * alias.c (compare_base_decls): Handle registers with asm
31009 specification conservatively.
31010 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
31011 compare_base_decls returning dont-know properly.
31013 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31015 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
31016 (legitimate_offset_address_p): New function.
31017 (arc_legitimate_address_p): Use above function.
31019 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31021 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
31023 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31025 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
31026 ACCH registers whenever they are available.
31028 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31030 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
31031 double regs fix when not used.
31033 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31035 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
31037 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
31038 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
31040 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31042 * config/arc/arc.c (arc_output_addsi): Check for h-register class
31043 when emitting short ADD instructions.
31045 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31047 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
31049 (cmpsi_cc_c_insn): Likewise.
31050 (cbranchsi4_scratch): Compute proper instruction length using
31051 compact_hreg_operand.
31052 * config/arc/predicates.md (compact_hreg_operand): New predicate.
31054 2017-04-25 Richard Biener <rguenther@suse.de>
31056 PR middle-end/80509
31057 * passes.c (pass_manager::pass_manager): Initialize
31058 m_name_to_pass_map.
31060 2017-04-25 Richard Biener <rguenther@suse.de>
31062 PR tree-optimization/79201
31063 * tree-ssa-sink.c (statement_sink_location): Handle calls.
31065 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31068 * config/s390/vector.md: Split MEM->GPR vector moves for
31069 non-s_operand addresses.
31071 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31074 * config/s390/predicates.md (reload_const_wide_int_operand): New
31076 * config/s390/s390.md ("movti"): Remove d/P alternative.
31077 ("movti_bigconst"): New pattern definition.
31079 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31082 * s390-protos.h (s390_expand_cs_hqi): Removed.
31083 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
31084 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
31085 modes as well as CCZ1mode and CCZmode.
31086 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
31087 signature of s390_emit_compare_and_swap.
31088 (s390_expand_cs_hqi): Likewise, make static.
31089 (s390_expand_cs_tdsi): Generate an explicit compare before trying
31090 compare-and-swap, in some cases.
31091 (s390_expand_cs): Wrapper function.
31092 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
31094 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
31095 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
31096 patterns for small and large integers. Forbid symref memory operands.
31097 Move expander to s390.c. Require cc register.
31098 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
31099 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
31100 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
31101 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
31102 symref memory operands. Remove CC mode and call s390_match_ccmode
31104 ("atomic_exchange<mode>"): Allow and implement all integer modes.
31106 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31108 * config/s390/s390.md (define_peephole2): New peephole to help
31109 combining the load-and-test pattern with volatile memory.
31111 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31113 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
31114 with CCZmode for TARGET_Z196.
31116 2017-04-25 Jakub Jelinek <jakub@redhat.com>
31118 PR rtl-optimization/80501
31119 * combine.c (make_compound_operation_int): Set subreg_code to SET
31120 even for AND with mask of the sign bit of mode.
31122 PR rtl-optimization/80500
31123 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
31124 sum's initial value.
31126 2017-04-25 Julian Brown <julian@codesourcery.com>
31127 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31129 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
31131 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
31133 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
31135 2017-04-25 Julian Brown <julian@codesourcery.com>
31136 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31138 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
31139 (thunderx2t99_sha): New Reservation.
31141 2017-04-25 Julian Brown <julian@codesourcery.com>
31142 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31144 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
31145 type for 1-element load.
31147 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
31149 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
31151 2017-04-24 Martin Jambor <mjambor@suse.cz>
31153 PR tree-optimization/80293
31154 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
31155 char arrays not totally scalarizable if it is false.
31156 (analyze_all_variable_accesses): Pass correct value in the new
31157 parameter. Add a statistics counter.
31159 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
31161 PR middle-end/79931
31162 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
31164 2017-04-24 Richard Biener <rguenther@suse.de>
31166 PR tree-optimization/80494
31167 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
31168 out for complex types.
31170 2017-04-24 Richard Biener <rguenther@suse.de>
31172 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
31173 * tree-ssa-sccvn.c (print_scc): Print SCC size.
31174 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
31175 (DFS): Adjust and never fail.
31176 (sccvn_dom_walker::fail): Remove.
31177 (sccvn_dom_walker::before_dom_children): Adjust.
31178 (run_scc_vn): Likewise and never fail.
31179 * tree-ssa-pre.c (pass_pre::execute): Adjust.
31180 (pass_fre::execute): Likewise.
31182 2017-04-24 Richard Biener <rguenther@suse.de>
31184 PR tree-optimization/79725
31185 * tree-ssa-sink.c (statement_sink_location): Return whether
31186 failure reason was zero uses. Move that check later.
31187 (sink_code_in_bb): Deal with zero uses by removing the stmt
31190 2017-04-24 Richard Biener <rguenther@suse.de>
31193 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
31194 pointer-based references.
31196 2017-04-24 Richard Biener <rguenther@suse.de>
31199 * pass_manager.h (pass_manager::operator new): Remove.
31200 (pass_manager::operator delete): Likewise.
31201 * passes.c (pass_manager::operator new): Remove.
31202 (pass_manager::operator delete): Likewise.
31203 (pass_manager::pass_manager): Zero individual pass members.
31205 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
31208 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
31209 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
31210 Check "XEXP (src, 1)" operand here.
31211 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
31212 Check "XEXP (src, 1)" operand here.
31213 (dimode_scalar_chain::make_vector_copies): Detect count register
31214 of a shift instruction. Zero extend count register from QImode
31215 to DImode to satisfy vector shift pattern count operand predicate.
31216 Substitute vector shift count operand with a DImode copy.
31217 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
31220 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
31222 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
31223 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
31224 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
31225 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
31226 (UNSPEC_NOREX_MEM): Remove definition.
31228 2017-04-21 Richard Biener <rguenther@suse.de>
31230 PR tree-optimization/79547
31231 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
31232 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
31233 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
31234 without any constraints.
31236 2017-04-21 Richard Biener <rguenther@suse.de>
31238 PR tree-optimization/78847
31239 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
31241 2017-04-21 Richard Biener <rguenther@suse.de>
31243 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
31244 (build_distinct_type_copy): Likewise.
31245 (build_variant_type_copy): Likewise.
31246 * tree.c (build_qualified_type): Pass down mem-stat info.
31247 (build_distinct_type_copy): Likewise.
31248 (build_variant_type_copy): Likewise.
31250 2017-04-21 Richard Biener <rguenther@suse.de>
31252 PR tree-optimization/80237
31253 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
31255 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
31256 for a simplified result.
31258 2016-04-21 Richard Biener <rguenther@suse.de>
31260 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
31261 sth as strict as a simple_iv but a chrec without symbols and an
31262 operand defined in the loop we are peeling (and not some subloop).
31263 (propagate_constants_for_unrolling): Propagate all constants.
31265 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
31268 * config/i386/i386.c (print_reg): Remove assert for disalowed
31269 regno values, call output_operand_lossage instead.
31271 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
31274 * config/i386/constraints.md (Yc): New register constraint.
31275 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
31276 Use Yc constraint for alternative 2 of operand 0. Remove
31277 preferred_for_speed attribute.
31279 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
31281 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
31282 lastprivate clauses in SIMT case.
31284 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
31286 * doc/invoke.texi (-Wextra-semi): Document new warning option.
31288 2017-04-20 Richard Biener <rguenther@suse.de>
31290 PR tree-optimization/57796
31291 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
31292 as N scalar stores.
31293 (vect_model_load_cost): Cost gathers as N scalar loads.
31295 2017-04-20 Richard Biener <rguenther@suse.de>
31297 * ggc-page.c (ggc_allocated_p): Rename to ...
31298 (safe_lookup_page_table_entry): ... this and return the lookup
31300 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
31302 2017-04-20 Richard Biener <rguenther@suse.de>
31304 PR tree-optimization/80453
31305 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
31306 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
31307 from the conditions.
31308 (vn_phi_eq): Pass them down.
31309 (vn_phi_lookup): Record them.
31310 (vn_phi_insert): Likewise.
31312 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
31314 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
31315 uninitialized variable warning to avoid buffer overrun.
31317 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
31320 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
31321 is suppressed for '{ 0 }' in C.
31323 2017-04-20 Jakub Jelinek <jakub@redhat.com>
31325 * BASE-VER: Set to 8.0.0.
31327 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
31329 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
31330 priority .init_array and .fini_array section with SECTION_NOTYPE
31333 2017-04-20 Jakub Jelinek <jakub@redhat.com>
31335 PR middle-end/80423
31336 * tree.h (build_array_type): Add typeless_storage default argument.
31337 * tree.c (type_cache_hasher::equal): Also compare
31338 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
31339 (build_array_type): Add typeless_storage argument, set
31340 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
31342 (build_nonshared_array_type): Adjust build_array_type_1 caller.
31343 (build_array_type): Likewise. Add typeless_storage argument.
31345 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
31346 Jakub Jelinek <jakub@redhat.com>
31348 PR tree-optimization/80426
31349 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
31350 operation on symbolic operands, also compute the overflow for the
31351 invariant part when the operation degenerates into a negation.
31353 2017-04-19 Jakub Jelinek <jakub@redhat.com>
31356 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
31357 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
31360 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
31362 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
31365 * config/avr/avr.c (tree.h): Include it.
31366 (cgraph.h): Include it.
31367 (avr_encode_section_info): Don't warn for uninitialized progmem
31368 variable if it's just an alias.
31370 2017-04-19 Richard Biener <rguenther@suse.de>
31373 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
31374 when needed by AutoPGO.
31376 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
31379 * doc/lto.texi: Remove an extra 'that'.
31381 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
31383 PR rtl-optimization/80429
31384 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
31385 are only used in debug insns.
31387 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
31388 Vladimir Makarov <vmakarov@redhat.com>
31390 * config/sparc/predicates.md (input_operand): Add comment. Return
31391 true for any memory operand when LRA is in progress.
31392 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
31394 2017-04-18 Jeff Law <law@redhat.com>
31397 * mips.md ({return,simple_return}_internal): Do not overwrite
31400 2017-04-18 Jakub Jelinek <jakub@redhat.com>
31402 PR tree-optimization/80443
31403 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
31404 instead of adding 1, subtract -1 and similarly instead of subtracting
31407 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
31409 PR rtl-optimization/80357
31410 * haifa-sched.c (tmp_bitmap): New variable.
31411 (model_recompute): Handle duplicate use records.
31412 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
31413 (free_global_sched_pressure_data): Free it.
31415 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
31418 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31419 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31420 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31421 instead of SYSTEM_HEADER_DIR.
31423 2017-04-18 Jeff Law <law@redhat.com>
31425 PR middle-end/80422
31426 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
31427 predecessors after walking up the insn chain.
31429 2017-04-18 Jakub Jelinek <jakub@redhat.com>
31432 * dwarf2out.c (modified_type_die): Try harder not to emit internal
31433 sizetype type into debug info.
31435 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
31438 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
31439 unneeded test for TARGET_UPPER_REGS_SF.
31440 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
31442 2017-04-18 Jakub Jelinek <jakub@redhat.com>
31445 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
31446 instead of gsi_after_labels.
31448 2017-04-18 Jeff Law <law@redhat.com>
31450 * regcprop.c (maybe_mode_change): Avoid creating copies of the
31454 2017-04-13 Jeff Law <law@redhat.com>
31455 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
31456 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
31458 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
31461 * config/avr/avr.c (intl.h): Include it.
31462 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
31464 2017-04-18 Martin Liska <mliska@suse.cz>
31466 PR gcov-profile/78783
31467 * gcov-tool.c (gcov_output_files): Validate that destination
31468 file is either removed by the tool or by a user.
31470 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
31471 Guy Benyei <guybe@mellanox.com>
31473 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
31474 block, and do not negate it, the stored id is already negative.
31476 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
31478 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
31480 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
31483 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
31484 masks of options that should be turned off if the VSX vector
31485 options are turned off.
31486 (OTHER_P8_VECTOR_MASKS): Likewise.
31487 (OTHER_VSX_VECTOR_MASKS): Likewise.
31488 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
31489 rs6000_disable_incompatible_switches to validate no type switches
31491 (rs6000_incompatible_switch): New function to disallow turning on
31492 other vector options if -mno-vsx, -mno-power8-vector, or
31493 -mno-power9-vector are specified.
31495 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
31497 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
31499 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
31501 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
31502 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
31503 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
31504 (ARG_POINTER_CFA_OFFSET): Likewise.
31506 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
31508 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
31509 conditions to take advantage of various optimizations.
31511 2017-04-13 Jeff Law <law@redhat.com>
31513 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
31514 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
31515 (zero_extendsidi2_dext): Likewise.
31517 2017-04-13 Jakub Jelinek <jakub@redhat.com>
31520 * fold-const.c (fold_ternary_loc): Revert
31521 use op0 instead of fold_convert_loc (loc, type, arg0) part of
31524 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
31526 PR rtl-optimization/80343
31527 * lra-remat.c (update_scratch_ops): Assign original hard reg to
31528 new scratch pseudo.
31530 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
31533 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
31534 to ubsan_encode_value.
31536 2017-04-13 Jeff Law <law@redhat.com>
31538 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
31539 appearing in DEBUG_INSNs.
31541 2017-04-13 Martin Liska <mliska@suse.cz>
31543 PR gcov-profile/80413
31544 * gcov-io.c (gcov_write_string): Copy to buffer just when
31545 allocated size is greater than zero.
31547 2017-04-13 Jakub Jelinek <jakub@redhat.com>
31550 * dwarf2out.c (decls_for_scope): Ignore declarations of
31551 current_function_decl in BLOCK_NONLOCALIZED_VARS.
31553 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
31556 * ipa-visibility.c (non_local_p): Fix typos.
31557 (localize_node): When localizing symbol in same comdat group,
31558 dissolve the group only when we know external symbols are going
31560 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
31562 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31564 PR tree-optimization/79390
31565 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
31566 order does not result in usable sequence, retry with reversed operand
31572 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
31573 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
31574 op0 instead of fold_convert_loc (loc, type, arg0).
31576 2017-04-12 Jeff Law <law@redhat.com>
31578 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
31579 has a delay slot in the generated code.
31581 * config/cris/cris.md (cris_preferred_reload_class): Return
31582 GENNONACR_REGS rather than GENERAL_REGS.
31584 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31587 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
31588 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
31589 signedness of the result type.
31591 2017-04-12 Richard Biener <rguenther@suse.de>
31592 Jeff Law <law@redhat.com>
31594 PR tree-optimization/80359
31595 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
31596 trim stores to TARGET_MEM_REFs.
31598 2017-04-12 Richard Biener <rguenther@suse.de>
31600 PR tree-optimization/79390
31601 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
31602 threading case even more.
31604 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
31607 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
31608 for quad_address_p for TImode, instead of just not indexed_address.
31610 2017-04-12 Richard Biener <rguenther@suse.de>
31611 Bernd Edlinger <bernd.edlinger@hotmail.de>
31613 PR middle-end/79671
31614 * alias.c (component_uses_parent_alias_set_from): Handle
31615 TYPE_TYPELESS_STORAGE.
31616 (get_alias_set): Likewise.
31617 * tree-core.h (tree_type_common): Add typeless_storage flag.
31618 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
31619 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
31620 for types containing members with TYPE_TYPELESS_STORAGE.
31621 (place_field): Likewise.
31622 (layout_type): Likewise for ARRAY_TYPE.
31623 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
31624 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
31625 TYPE_TYPELESS_STORAGE.
31626 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
31628 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31631 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
31632 first argument to type.
31634 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31638 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
31639 CONST0_RTX (mode) rather than const0_rtx where appropriate.
31640 (rs6000_expand_binop_builtin): Likewise.
31641 (rs6000_expand_ternop_builtin): Likewise; also add missing
31642 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
31643 vshasigma built-ins.
31644 * doc/extend.texi: Document that vec_xxpermdi's third argument
31645 must be a constant.
31647 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
31649 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
31650 Use shift_const cost parameter when calculating gain of STV shifts.
31652 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
31654 PR rtl-optimization/70478
31655 * lra-constraints.c (process_alt_operands): Check memory for
31656 disfavoring memory insn operand.
31658 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31660 PR middle-end/80100
31661 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
31662 left shift in unsigned HOST_WIDE_INT type.
31664 PR rtl-optimization/80385
31665 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
31666 (not (neg X)) into (plus X -1) for complex or non-integral modes.
31669 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
31670 if they have any depend clauses.
31672 2017-04-11 Martin Liska <mliska@suse.cz>
31675 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
31676 * ipa-split.c (split_function): Create a local comdat symbol
31677 if caller is in a comdat group.
31679 2017-04-11 Martin Liska <mliska@suse.cz>
31682 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
31685 2017-04-11 Martin Sebor <msebor@redhat.com>
31687 PR middle-end/80364
31688 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
31689 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
31691 (directive::set_width, directive::set_precision, format_character):
31693 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
31696 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
31699 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
31700 conflict, set target->arch_name instead of target->cpu_name.
31702 2017-04-11 Richard Biener <rguenther@suse.de>
31704 PR tree-optimization/80374
31705 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
31706 build_zero_cst, remove fold_convertible_p check again.
31708 2017-04-11 Martin Liska <mliska@suse.cz>
31711 * ubsan.c (instrument_object_size): Do not instrument register
31714 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31717 * config/i386/i386-builtin-types.def
31718 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
31719 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
31720 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
31721 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
31722 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
31723 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
31724 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
31725 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
31726 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
31727 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
31728 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
31729 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
31730 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
31731 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
31732 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
31733 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
31734 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
31735 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
31736 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
31737 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
31738 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
31739 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
31740 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
31741 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
31742 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
31743 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
31744 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
31745 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
31746 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
31747 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
31748 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
31749 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
31750 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
31751 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
31752 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
31753 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
31754 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
31755 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
31756 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
31757 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
31758 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
31759 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
31760 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
31761 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
31762 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
31763 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
31765 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
31766 flag to second_arg_count, handle 4 argument function type _COUNT
31767 aliases, handle second_arg_count on second argument rather than last.
31769 2017-04-10 Jeff Law <law@redhat.com>
31771 PR tree-optimization/80374
31772 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
31773 record anything if we can not convert integer_zero_node to the
31776 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
31779 * config/rs6000/rs6000.c (rs6000_option_override_internal):
31780 Enhance special handling given to the TARGET_P9_MINMAX option in
31781 relation to certain other options.
31783 2017-04-10 Bin Cheng <bin.cheng@arm.com>
31785 PR tree-optimization/80153
31786 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
31787 remove POINTER_PLUS_EXPR's base part directly, rather than through
31790 2017-04-10 Richard Biener <rguenther@suse.de>
31791 Bin Cheng <bin.cheng@arm.com>
31793 PR tree-optimization/80153
31794 * tree-affine.c (aff_combination_to_tree): Get base pointer from
31795 the first element of pointer type aff_tree. Build result expr in
31797 (add_elt_to_tree): Convert to type unconditionally. Remove other
31798 fold_convert calls.
31799 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
31800 (rewrite_use_nonlinear_expr): Check invariant using iv information.
31802 2017-04-10 Richard Biener <rguenther@suse.de>
31804 * tree-ssa-structalias.c (find_func_aliases): Properly handle
31807 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
31809 PR rtl-optimization/70478
31810 * lra-constraints.c (curr_small_class_check): New.
31811 (update_and_check_small_class_inputs): New.
31812 (process_alt_operands): Update curr_small_class_check. Disfavor
31813 alternative insn memory operands. Check available regs for small
31816 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31819 * config/mips/mips.opt (-mvirt): Update description.
31820 * doc/invoke.texi (-mvirt): Likewise.
31822 2017-04-10 Richard Biener <rguenther@suse.de>
31824 PR middle-end/80362
31825 * fold-const.c (fold_binary_loc): Look at unstripped ops when
31826 looking for NEGATE_EXPR in -A / -B to A / B folding.
31828 2017-04-10 Martin Liska <mliska@suse.cz>
31830 PR gcov-profile/80224
31831 * gcov.c (print_usage): Fix usage string.
31832 (get_gcov_intermediate_filename): Remove.
31833 (output_gcov_file): Use both for normal and intermediate format.
31834 (generate_results): Do not initialize special file for
31835 intermediate format.
31837 2017-04-10 Richard Biener <rguenther@suse.de>
31839 PR tree-optimization/80304
31840 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
31843 2017-04-10 Nathan Sidwell <nathan@acm.org>
31846 * config/rs6000/rs6000.c (rs6000_vector_type): New.
31847 (rs6000_init_builtins): Use it.
31849 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31851 * config/arm/arm.md (<mrc>): Add mode to SET source.
31852 (<mrrc>): Likewise.
31854 2017-04-10 Richard Biener <rguenther@suse.de>
31856 PR middle-end/80344
31857 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
31859 2017-04-10 Jakub Jelinek <jakub@redhat.com>
31862 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
31863 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
31864 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
31865 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
31866 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
31867 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
31868 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
31869 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
31870 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
31871 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
31872 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
31873 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
31874 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
31875 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
31876 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
31877 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
31878 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
31879 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
31880 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
31881 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
31882 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
31883 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
31884 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
31886 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
31888 PR rtl-optimization/70478
31889 * lra-constraints.c: Reverse the last patch.
31891 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
31893 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
31894 Add comment for WCHAR_T.
31896 2017-04-08 Martin Liska <mliska@suse.cz>
31899 2017-04-07 Martin Liska <mliska@suse.cz>
31902 * ipa-split.c (split_function): Add function part to a same comdat
31905 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31908 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
31910 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
31912 * rs6000/rs6000.c (vec_load_pendulum): Rename...
31913 (vec_pairing): ...to this.
31914 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
31915 (rs6000_sched_init): Adjust for name change.
31916 (struct rs6000_sched_context): Likewise.
31917 (rs6000_init_sched_context): Likewise.
31918 (rs6000_set_sched_context): Likewise.
31920 2017-04-07 Jakub Jelinek <jakub@redhat.com>
31926 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
31928 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
31929 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
31930 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
31932 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
31934 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
31936 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31938 PR rtl-optimization/70703
31939 * ira-color.c (update_conflict_hard_regno_costs): Use
31940 int64_t instead of HOST_WIDE_INT.
31942 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31944 PR rtl-optimization/70478
31945 * lra-constraints.c (process_alt_operands): Disfavor alternative
31946 insn memory operands.
31948 2017-04-07 Jeff Law <law@redhat.com>
31950 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
31951 CALL and NOTE_INSN_CALL_ARG_LOCATION.
31953 2017-04-07 Martin Liska <mliska@suse.cz>
31956 * config/aarch64/aarch64.c (aarch64_process_target_attr):
31957 Show error message instead of an ICE.
31959 2017-04-07 Martin Liska <mliska@suse.cz>
31962 * ipa-split.c (split_function): Add function part to a same comdat
31965 2017-04-07 Richard Biener <rguenther@suse.de>
31967 PR middle-end/80341
31968 * tree.c (get_unwidened): Also handle ! for_type case for
31970 * convert.c (do_narrow): Split out from ...
31971 (convert_to_integer_1): ... here. Do not pass final truncation
31972 type to get_unwidened for TRUNC_DIV_EXPR.
31974 2017-04-07 Richard Biener <rguenther@suse.de>
31976 * tree-affine.c (wide_int_ext_for_comb): Take type rather
31978 (aff_combination_const): Adjust.
31979 (aff_combination_scale): Likewise.
31980 (aff_combination_add_elt): Likewise.
31981 (aff_combination_add_cst): Likewise.
31982 (aff_combination_convert): Likewise.
31983 (add_elt_to_tree): Likewise. Remove unused argument.
31984 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
31986 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
31988 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
31990 * config/arm/arm.c (arm_default_short_enums): Use
31991 ARM_DEFAULT_SHORT_ENUMS.
31992 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
31994 2017-04-06 Jakub Jelinek <jakub@redhat.com>
31997 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
31998 members with redundant out-of-class redeclaration.
32000 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
32003 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
32004 * config/i386/i386.md (*zero_extendsidi2):
32005 Add (?*x,*x) and (?*v,*v) alternatives.
32007 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
32010 * config/i386/i386.c (ix86_expand_builtin)
32011 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
32012 mode from insn data. Convert operands to insn operand mode.
32013 Copy operands that don't satisfy insn predicate to a register.
32015 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
32017 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
32020 2017-04-06 Richard Biener <rguenther@suse.de>
32022 PR tree-optimization/80334
32023 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
32024 preserve alignment of accesses.
32026 2017-04-06 Richard Biener <rguenther@suse.de>
32028 PR tree-optimization/80262
32029 * tree-sra.c (build_ref_for_offset): Preserve address-space
32031 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
32032 Drop useless address-space information on MEM_REF offsets.
32034 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
32036 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
32038 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
32040 PR rtl-optimization/70703
32041 * ira-color.c (update_conflict_hard_regno_costs): Use
32042 HOST_WIDE_INT instead of long.
32044 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
32047 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
32048 not defined for x86_64 target. Add -mmmx target option when __SSE2__
32050 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
32051 for x86_64 target. Handle -m3dnowa option.
32053 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
32055 PR rtl-optimization/70703
32056 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
32057 (update_conflict_hard_regno_costs): Use long instead of unsigned
32058 arithmetic for cost calculation.
32060 2017-04-05 Jakub Jelinek <jakub@redhat.com>
32061 Bernd Edlinger <bernd.edlinger@hotmail.de>
32064 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
32067 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
32070 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
32071 ptr_mode with Pmode throughout.
32072 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
32073 into probe_stack_range and use DImode.
32075 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
32078 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
32079 call_eh_return is true.
32081 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32083 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
32084 Initialize last_match_fntype_index.
32086 2017-04-05 Jakub Jelinek <jakub@redhat.com>
32089 * tree-nvr.c: Include internal-fn.h.
32090 (pass_return_slot::execute): Ignore internal calls without
32093 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32094 Richard Biener <rguenther@suse.de>
32097 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
32098 captures used multiple times, except for the last use.
32099 * generic-match-head.c: Include gimplify.h.
32101 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32103 PR tree-optimization/79390
32104 * target.h (struct noce_if_info): Declare.
32105 * targhooks.h (default_noce_conversion_profitable_p): Declare.
32106 * target.def (noce_conversion_profitable_p): New target hook.
32107 * ifcvt.h (struct noce_if_info): New type, moved from ...
32108 * ifcvt.c (struct noce_if_info): ... here.
32109 (noce_conversion_profitable_p): Renamed to ...
32110 (default_noce_conversion_profitable_p): ... this. No longer
32112 (noce_try_store_flag_constants, noce_try_addcc,
32113 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
32114 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
32115 instead of noce_conversion_profitable_p.
32116 * config/i386/i386.c: Include ifcvt.h.
32117 (ix86_option_override_internal): Don't override
32118 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
32119 (ix86_noce_conversion_profitable_p): New function.
32120 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
32121 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
32122 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
32123 * doc/tm.texi: Regenerated.
32125 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32127 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
32130 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32133 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
32134 instructions for small multiply cores.
32136 2017-04-04 Jeff Law <law@redhat.com>
32138 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
32140 (mips_expand_vec_perm_const): Initialize elements in orig_perm
32141 that are not set by the loop over the elements.
32143 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32146 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
32147 int mode, convert_modes it to mode as unsigned, otherwise use
32148 lowpart_subreg to mode rather than SImode.
32149 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
32150 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
32151 Use DImode instead of SImode for the shift count operand.
32152 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
32155 2017-04-04 Richard Biener <rguenther@suse.de>
32157 PR middle-end/80281
32158 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
32159 arithmetic done for the negate or the plus. Simplify.
32160 (A - (-B) -> A + B): Likewise.
32161 * fold-const.c (split_tree): Make sure to not negate pointers.
32163 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
32165 PR rtl-optimization/60818
32166 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
32167 a compare of comparisons with the thing compared if this results
32168 in a different machine mode.
32170 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
32172 * alias.c (base_alias_check): Fix typo in comment.
32173 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
32174 * cgraphunit.c (symbol_table::compile): Likewise.
32175 * collect2.c (maybe_run_lto_and_relink): Likewise.
32176 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
32177 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
32178 * config/avr/avr.c (avr_map_op_t): Likewise.
32179 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
32180 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
32181 * config/epiphany/epiphany.md (movcc): Likewise.
32182 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
32183 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
32185 * config/mips/mips.c (mips_save_restore_reg): Likewise.
32186 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
32187 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
32188 * config/sh/sh.c (sh_rtx_costs): Likewise.
32189 * fold-const.c (fold_truth_andor): Likewise.
32190 * genautomata.c (collapse_flag): Likewise.
32191 * gengtype.h (struct type::u::s): Likewise.
32192 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
32193 * input.c (FORMAT_AMOUNT): Likewise.
32194 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
32195 (known_aggs_to_agg_replacement_list): Likewise.
32196 * ipa-inline-analysis.c: Likewise.
32197 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
32198 * ipa-polymorphic-call.c
32199 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
32200 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
32201 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
32203 * modulo-sched.c (apply_reg_moves): Likewise.
32204 * omp-expand.c (build_omp_regions_1): Likewise.
32205 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
32206 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
32207 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
32208 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
32209 * value-prof.c: Likewise.
32210 * var-tracking.c (val_reset): Likewise.
32212 2017-04-03 Richard Biener <rguenther@suse.de>
32214 PR tree-optimization/80275
32215 * fold-const.c (split_address_to_core_and_offset): Handle
32218 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
32220 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
32221 descriptors is at least equal to that of functions.
32223 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
32225 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
32227 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
32230 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
32231 (mov<IMOD4:mode>): New expander.
32232 (*mov<IMOD4:mode>_internal): New insn and split pattern.
32234 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
32236 PR rtl-optimization/79405
32237 * fwprop.c (propagations_left): New variable.
32238 (forward_propagate_into): Decrement it.
32239 (fwprop_init): Initialize it.
32240 (fw_prop): If the variable has reached zero, stop propagating.
32241 (fwprop_addr): Ditto.
32243 2017-03-31 Jakub Jelinek <jakub@redhat.com>
32246 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
32247 a FUNCTION_DECL, pass it as decl instead of origin to
32250 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
32252 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
32255 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
32258 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
32259 TARGET_VSX_SMALL_INTEGER.
32261 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32263 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
32264 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
32266 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
32268 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
32269 extraction from odd-numbered MSA register.
32271 2017-03-31 Jakub Jelinek <jakub@redhat.com>
32273 PR middle-end/80173
32274 * expmed.c (store_bit_field_1): Don't attempt to create
32275 a word subreg out of hard registers wider than word if they
32276 have HARD_REGNO_NREGS of 1 for their mode.
32278 PR middle-end/80163
32279 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
32280 conversions to integer types wider than word and pointer.
32283 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
32284 (rtx_equal_for_cselib_p): Pass 0 to it.
32285 * cselib.c (cselib_hasher::equal): Likewise.
32286 (rtx_equal_for_cselib_1): Add depth argument. If depth
32287 is 128, don't look up VALUE locs and punt. Increment
32288 depth in recursive calls when walking VALUE locs.
32290 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
32292 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
32293 (make_gcov_file_name): Use the canonical path name for generating
32295 (read_line): Fix handling of files with ascii null bytes.
32297 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
32299 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
32300 to initialise a vector register instead
32301 of using a const_int.
32303 2017-03-30 Jakub Jelinek <jakub@redhat.com>
32305 PR translation/80189
32306 * gimplify.c (omp_default_clause): Use %qs instead of %s in
32307 diagnostic messages.
32309 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
32312 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
32313 (dfp_diex_<mode>): Update mode of operand 1.
32314 * doc/extend.texi (dxex, dxexq): Document change to return type.
32315 (diex, diexq): Document change to argument type.
32317 2017-03-30 Martin Jambor <mjambor@suse.cz>
32320 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
32321 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
32322 it reflects the signature changes performed at the callee side.
32323 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
32324 to cgraph_build_function_type_skip_args.
32325 (build_function_decl_skip_args): Adjust call to the above function.
32327 2017-03-30 Jakub Jelinek <jakub@redhat.com>
32330 * config/i386/sse.md
32331 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
32332 register as dest whenever it is a MEM not rtx_equal_p to the
32333 corresponding dup operand, and when forcing into reg move the
32334 reg into the memory afterwards.
32335 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
32336 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
32337 for the force_reg mode.
32338 (avx512vl_vextractf128<mode>): Use register as dest either
32339 always when a MEM, or when it is a MEM not rtx_equal_p to the
32340 corresponding dup operand, or even not when it is a CONST_VECTOR
32341 depending on the mode and lo vs. hi.
32342 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
32344 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
32345 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
32346 Likewise. Require that operands[2] is even.
32347 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
32348 Remove extraneous parens. Require that operands[2] is a multiple
32350 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
32351 operands[0] is a MEM if <mask_applied>, the predicates/constraints
32352 disallow memory then.
32354 2017-03-30 Richard Biener <rguenther@suse.de>
32356 PR tree-optimization/77498
32357 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
32358 to non-constants over backedges.
32360 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
32362 PR rtl-optimization/80233
32363 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
32364 as last_combined_insn. Do not test for BARRIER_P separately.
32366 2017-03-29 Andreas Schwab <schwab@suse.de>
32369 * calls.c (prepare_call_address): Convert funexp to Pmode before
32370 copying to temp reg.
32372 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32374 PR tree-optimization/80158
32375 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
32376 Handle possible future case of more than one alternate
32378 (replace_rhs_if_not_dup): Likewise.
32379 (replace_one_candidate): Likewise.
32381 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
32383 PR rtl-optimization/80193
32384 * ira.c (ira): Do not check allocation for LRA.
32386 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
32388 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
32389 (nvptx_output_simt_exit): Declare.
32390 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
32391 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
32392 (init_softstack_frame): Move initialization of crtl->is_leaf to...
32393 (nvptx_declare_function_name): ...here. Emit declaration of local
32394 memory space buffer for omp_simt_enter insn.
32395 (nvptx_output_unisimt_switch): New.
32396 (nvptx_output_softstack_switch): New.
32397 (nvptx_output_simt_enter): New.
32398 (nvptx_output_simt_exit): New.
32399 * config/nvptx/nvptx.h (struct machine_function): New fields
32400 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
32401 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
32402 (UNSPECV_SIMT_EXIT): Ditto.
32403 (omp_simt_enter_insn): New insn.
32404 (omp_simt_enter): New expansion.
32405 (omp_simt_exit): New insn.
32406 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
32408 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
32409 (expand_GOMP_SIMT_ENTER_ALLOC): New.
32410 (expand_GOMP_SIMT_EXIT): New.
32411 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
32412 (GOMP_SIMT_ENTER_ALLOC): Ditto.
32413 (GOMP_SIMT_EXIT): Ditto.
32414 * target-insns.def (omp_simt_enter): New insn.
32415 (omp_simt_exit): Ditto.
32416 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
32418 (lower_rec_simd_input_clauses): Implement SIMT privatization.
32419 (lower_rec_input_clauses): Likewise.
32420 (lower_lastprivate_clauses): Handle SIMT privatization.
32422 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
32423 (ompdevlow_adjust_simt_enter): New.
32424 (find_simtpriv_var_op): New.
32425 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
32426 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
32428 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
32429 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
32430 (copy_decl_for_dup_finish): Ditto.
32432 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
32434 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
32437 * config/i386/i386.c (ix86_option_override_internal): Always
32438 allow -mpreferred-stack-boundary=3 for 64-bit targets.
32440 2017-03-28 Bin Cheng <bin.cheng@arm.com>
32442 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
32444 2017-03-28 Bin Cheng <bin.cheng@arm.com>
32446 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
32447 mark new edge's irreducible flag accordign to it.
32448 (vect_do_peeling): Check loop preheader edge's irreducible flag
32449 and pass it to function slpeel_add_loop_guard.
32451 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
32453 PR tree-optimization/80218
32454 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
32455 Update block frequencies and counts.
32457 2017-03-28 Richard Biener <rguenther@suse.de>
32459 PR tree-optimization/78644
32460 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
32461 of a simplification result we may not use it at all.
32463 2017-03-28 Richard Biener <rguenther@suse.de>
32466 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
32467 without arguments, generate default definition of a SSA name.
32469 2017-03-28 Richard Biener <rguenther@suse.de>
32471 PR middle-end/80222
32472 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
32473 TYPE_REF_CAN_ALIAS_ALL references.
32474 * fold-const.c (fold_indirect_ref_1): Likewise.
32476 2017-03-28 Martin Liska <mliska@suse.cz>
32479 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
32480 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
32482 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
32483 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
32485 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
32486 (EXTRA_SPECS): Define.
32487 (SUBTARGET_EXTRA_SPECS): Likewise.
32488 (SUBTARGET_CPP_SPEC): Likewise.
32489 * config/arc/elf.h (EXTRA_SPECS): Renamed to
32490 SUBTARGET_EXTRA_SPECS.
32491 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
32493 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
32495 * config/arc/simdext.md (vst64_insn): Update pattern.
32496 (vld32wh_insn): Likewise.
32497 (vld32wl_insn): Likewise.
32498 (vld64_insn): Likewise.
32499 (vld32_insn): Likewise.
32501 2017-03-28 Marek Polacek <polacek@redhat.com>
32504 * fold-const.c (fold_comparison): Use protected_set_expr_location
32505 instead of SET_EXPR_LOCATION.
32507 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
32509 * tree.c (add_expr): Avoid name lookup warning.
32511 2017-03-27 Jeff Law <law@redhat.com>
32513 PR tree-optimization/80216
32514 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
32515 function name. Limit recursion depth.
32516 (record_temporary_equivalences): Corresponding changes.
32518 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
32520 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
32523 2017-03-27 Jakub Jelinek <jakub@redhat.com>
32526 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
32528 * cfgcleanup.c (reg_note_cfa_p): New array.
32529 (insns_have_identical_cfa_notes): New function.
32530 (old_insns_match_p): Don't cross-jump in between /f
32531 and non-/f instructions. If both i1 and i2 are frame related,
32532 verify all CFA notes, their order and content.
32534 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
32537 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
32538 HImode and SImode with zero extend to DImode to one insn.
32539 (bswap<mode>2_extenddi): Likewise.
32540 (bswapsi2_extenddi): Likewise.
32541 (bswaphi2_extendsi): Likewise.
32542 (bswaphi2): Combine bswap HImode and SImode into one insn.
32543 Separate memory insns from swapping register.
32544 (bswapsi2): Likewise.
32545 (bswap<mode>2): Likewise.
32546 (bswaphi2_internal): Delete, no longer used.
32547 (bswapsi2_internal): Likewise.
32548 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
32549 store, and gpr<-gpr swap insns.
32550 (bswap<mode>2_store): Likewise.
32551 (bswaphi2_reg): Register only splitter, combine with the splitter.
32552 (bswaphi2 splitter): Likewise.
32553 (bswapsi2_reg): Likewise.
32554 (bswapsi2 splitter): Likewise.
32555 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
32556 the insns into load, store, and register/register insns.
32557 (bswapdi2_ldbrx): Likewise.
32558 (bswapdi2_load): Likewise.
32559 (bswapdi2_store): Likewise.
32560 (bswapdi2_reg): Likewise.
32562 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
32564 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
32565 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
32567 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
32570 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
32572 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32573 special handling for target option conflicts between dform
32574 options (-mpower9-dform, -mpower9-dform-vector,
32575 -mpower9-dform-scalar) and -mno-direct-move.
32577 2017-03-27 Richard Biener <rguenther@suse.de>
32579 PR tree-optimization/80181
32580 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
32582 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32584 * config/arc/predicates.md (move_double_src_operand): Replace the
32585 call to move_double_src_operand with a call to address_operand.
32587 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32589 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
32590 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
32591 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
32593 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32595 * config/arc/predicates.md (long_immediate_loadstore_operand):
32596 Consider scaled addresses cases.
32598 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32600 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
32601 restored when in interrupt.
32602 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
32603 doesn't have delay slot.
32605 2017-03-27 Richard Biener <rguenther@suse.de>
32608 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
32609 inlined thunk clones.
32611 2017-03-27 Jakub Jelinek <jakub@redhat.com>
32614 * asan.c (instrument_derefs): Copy over last operand from
32615 original COMPONENT_REF to the new COMPONENT_REF with
32616 DECL_BIT_FIELD_REPRESENTATIVE.
32617 * ubsan.c (instrument_object_size): Likewise.
32619 2017-03-27 Richard Biener <rguenther@suse.de>
32621 PR tree-optimization/80170
32622 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
32623 sure DR/SCEV didnt fold in constants we do not see when looking
32624 at the reference base alignment.
32626 2017-03-27 Richard Biener <rguenther@suse.de>
32628 PR middle-end/80171
32629 * gimple-fold.c (fold_ctor_reference): Properly guard against
32630 NULL return value from canonicalize_constructor_val.
32632 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
32635 * config/i386/i386.c (ix86_expand_builtin)
32636 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
32637 flags reg setting and flags reg using instructions.
32638 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
32639 clobbering instructions to zero extend op2.
32641 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
32643 * doc/install.texi (Configuration) <--with-aix-soname>:
32644 Update link to AIX ld.
32646 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
32648 PR rtl-optimization/80160
32649 PR rtl-optimization/80159
32650 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
32651 reg_alternate_class into account.
32653 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
32656 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
32657 to consider in curr_insn_transform.
32659 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32661 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
32662 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
32663 and emit_mode_inner.
32665 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32667 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
32668 argument to the overloaded builtin variants. Use the new flag to
32669 deprecate certain builtin variants.
32670 * config/s390/s390-builtin-types.def: Add new builtin types.
32671 * config/s390/s390-builtins.h: Support new flags field for
32672 overloaded builtins.
32673 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
32674 (s390_macro_to_expand): Enable vector float data type.
32675 (s390_cpu_cpp_builtins_internal): Indicate support of the new
32676 builtins by incrementing the __VEC__ version number.
32677 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
32679 (s390_resolve_overloaded_builtin): Emit error messages depending
32680 on the builtin flags.
32681 * config/s390/s390.c (s390_expand_builtin): Support additional
32682 flags argument. Change error message to match the messages
32683 emitted in s390-c.c.
32684 * config/s390/s390.md: New UNSPEC_* constants.
32685 (op_type): Add new instruction types.
32686 * config/s390/vecintrin.h: Add new builtins and test data class
32688 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
32689 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
32690 (VEC_INEXACT, VEC_NOINEXACT): New constants.
32691 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
32692 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
32693 ("vec_mergel<mode>"): V_HW -> VEC_HW.
32695 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
32696 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
32697 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
32698 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
32700 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
32701 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
32702 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
32703 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
32705 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
32706 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
32707 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
32708 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
32709 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
32710 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
32711 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
32713 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
32714 ("vec_scatter_element<V_HW_4:mode>_DI")
32715 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
32716 ("vec_fpint<mode>", "vflls")
32717 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
32718 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
32719 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
32720 ("*vec_cmphe<mode>_cc"): ... these.
32722 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
32723 mode constant instead of magic value.
32725 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32727 * config/s390/s390.c (s390_expand_vec_compare): Support other
32728 vector floating point modes than just V2DF.
32729 (s390_expand_vcond): Likewise.
32730 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
32731 (s390_cannot_change_mode_class): Prevent mode changes between TF
32732 and V1TF in vector registers.
32733 * config/s390/s390.md (DF, SF): New mode attributes.
32734 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
32735 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
32736 SFmode support for VRs.
32737 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
32739 (VFT, VF_HW): New mode iterators.
32740 (vw, sdx): New mode attributes.
32741 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
32742 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
32743 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
32744 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
32745 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
32746 also the new vector floating point modes. Renaming to ...
32748 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
32749 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
32750 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
32751 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
32752 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
32753 ("vec_unordered<mode>"): ... these.
32755 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
32756 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
32757 ("*vec_extendv2df"): New insn definitions.
32759 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32761 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
32762 ("mulditi3_2", "*muldi3_sign"): New patterns.
32763 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
32764 rename the pattern definition.
32766 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32768 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
32770 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
32772 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32774 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
32775 instruction if possible.
32776 * config/s390/vector.md (vec_halfnumelts): New mode
32778 ("*vec_vllezlf<mode>"): New pattern.
32780 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32782 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
32783 ("popcountv4si2", "popcountv2di2"): Rename to ...
32784 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
32785 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
32787 ("popcount<mode>2_vxe"): New pattern.
32789 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32791 * common/config/s390/s390-common.c (processor_flags_table): Add
32793 * config.gcc: Add arch12.
32794 * config/s390/driver-native.c (s390_host_detect_local_cpu):
32795 Default to arch12 for unknown CPU model numbers.
32796 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
32797 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
32798 PROCESSOR_max sanity check.
32799 * config/s390/s390-opts.h (enum processor_type): Add
32801 * config/s390/s390.c (processor_table): Add arch12.
32802 (s390_expand_builtin): Add check for B_VXE flag.
32803 (s390_issue_rate): Add PROCESSOR_ARCH12.
32804 (s390_get_sched_attrmask): Likewise.
32805 (s390_get_unit_mask): Likewise.
32806 (s390_sched_score): Enable z13 scheduling for arch12.
32807 (s390_sched_reorder): Likewise.
32808 (s390_sched_variable_issue): Likewise.
32809 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
32811 (s390_tune_attr): Use z13 scheduling also for arch12.
32812 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
32813 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
32814 (TARGET_VXE_P): New macros.
32815 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
32816 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
32817 * config/s390/s390.opt: Add arch12 as processor_type.
32819 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32821 * config/s390/s390.md
32822 ("fixuns_truncdddi2", "fixuns_trunctddi2")
32823 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
32824 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
32826 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
32827 Rename expanders to ...
32829 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
32830 ("fixuns_truncdddi2_emu"): ... these.
32832 ("fixuns_trunc<mode>si2_emu"): New expander.
32834 ("*fixuns_truncdfdi2_z13"): Rename to ...
32835 ("*fixuns_truncdfdi2_vx"): ... this.
32837 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32839 * config/s390/2964.md: Remove the single element vector compare
32840 instructions which are no longer used.
32841 * config/s390/s390.c (s390_select_ccmode): Remove handling of
32843 (s390_canonicalize_comparison): Remove handling of DFmode
32845 (s390_expand_vec_compare_scalar): Remove function.
32846 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
32847 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
32849 ("*cmp<mode>_ccs"): Add wfcdb instruction.
32851 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32853 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
32855 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
32856 will anyway by matched by mov<mode>_64dfp.
32858 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32860 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
32861 vlef/vstef. Add missing operand to vleif.
32863 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32865 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
32866 pair for all vector types with 64 bit elements.
32867 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
32868 * config/s390/vector.md (V_HW_64): ... here.
32869 (V_128_NOSINGLE): New mode iterator.
32870 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
32871 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
32872 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
32873 ("*vec_load_pairv2di"): Change to ...
32874 ("*vec_load_pair<mode>"): ... this one.
32876 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32878 * config/s390/constraints.md: Add comments.
32879 (jKK): Reject element sizes > 8 bytes.
32880 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
32882 * config/s390/s390.md: Add the s_operand checks formerly in
32883 s390_split_ok_p to various splitters where they are still
32885 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
32886 for 128 bit vectors. Plus two splitters.
32888 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32890 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
32893 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32896 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
32897 error if the boundary argument is not constant.
32899 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32901 PR rtl-optimization/80112
32902 * loop-doloop.c (doloop_condition_get): Don't check condition
32903 if cmp isn't SET with IF_THEN_ELSE src.
32905 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32907 PR tree-optimization/80158
32908 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
32909 replacing a candidate statement, also replace it for the
32910 candidate's alternate interpretation.
32911 (replace_rhs_if_not_dup): Likewise.
32912 (replace_one_candidate): Likewise.
32914 2017-03-24 Richard Biener <rguenther@suse.de>
32916 PR tree-optimization/80167
32917 * graphite-isl-ast-to-gimple.c
32918 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
32920 (translate_isl_ast_to_gimple::get_rename): Likewise.
32922 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32924 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
32925 handling of certain combinations of target options, including the
32926 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
32927 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
32929 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32932 * config/arm/arm.md (*load_multiple): Add reload_completed to
32933 matching condition.
32935 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32936 Richard Biener <rguenth@suse.de>
32938 PR tree-optimization/79908
32939 PR tree-optimization/80136
32940 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32941 been cast away, gimplify_and_add suffices.
32943 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
32945 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
32947 2017-03-23 Richard Biener <rguenther@suse.de>
32949 PR tree-optimization/80032
32950 * gimplify.c (gimple_push_cleanup): Forced unconditional
32951 cleanups still have to go to the conditional_cleanups
32954 2017-03-22 Jakub Jelinek <jakub@redhat.com>
32956 PR tree-optimization/80072
32957 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
32959 (next_operand_entry_id): Change type to unsigned int.
32960 (sort_by_operand_rank): Make sure to return the right return value
32961 even if unsigned fields are bigger than INT_MAX.
32962 (struct oecount): Change cnt and id type to unsigned int.
32963 (oecount_hasher::equal): Formatting fix.
32964 (oecount_cmp): Make sure to return the right return value
32965 even if unsigned fields are bigger than INT_MAX.
32966 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
32969 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
32970 TREE_READONLY on result if writing it more than once.
32973 * doc/invoke.texi (-fsanitize=thread): Document that with
32974 -fnon-call-exceptions atomics are not able to throw
32978 * tsan.c: Include tree-eh.h.
32979 (instrument_builtin_call): Call maybe_clean_eh_stmt or
32980 maybe_clean_or_replace_eh_stmt where needed.
32981 (instrument_memory_accesses): Add cfg_changed argument.
32982 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
32983 if it returned true.
32984 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
32986 PR rtl-optimization/63191
32987 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
32988 wrapper function, moved the whole old content into ...
32989 (ix86_delegitimize_address_1): ... this. New inline function.
32990 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
32991 true as last argument instead of ix86_delegitimize_address.
32993 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32995 * config/aarch64/aarch64.c (generic_branch_cost): Copy
32996 cortexa57_branch_cost.
32998 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
33000 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
33002 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33005 * doc/md.texi (Constraints): Document wA constraint.
33006 * config/rs6000/constraints.md (wA): New.
33007 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
33008 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
33009 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
33010 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
33012 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
33015 * gimplify.c (is_oacc_declared): New function.
33016 (oacc_default_clause): Use it to set default flags for acc declared
33017 variables inside parallel regions.
33018 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
33019 declared variables.
33020 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
33021 declare attribute to any decl as necessary.
33023 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
33026 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
33027 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
33028 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
33029 (arm_arch_lpae): This.
33030 * config/arm/arm.c (arm_arch7ve): Rename into ...
33031 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
33032 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
33035 2017-03-22 Martin Liska <mliska@suse.cz>
33038 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
33039 error message instead of an ICE.
33041 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33043 * doc/extend.texi (6.11 Additional Floating Types): Revise.
33045 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
33047 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
33049 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33052 2017-03-21 Martin Sebor <msebor@redhat.com>
33054 * doc/extend.texi: Use "cannot" instead of "can't."
33055 * doc/hostconfig.texi: Same.
33056 * doc/install.texi: Same.
33057 * doc/invoke.texi: Same.
33058 * doc/loop.texi: Same.
33059 * doc/md.texi: Same.
33060 * doc/objc.texi: Same.
33061 * doc/rtl.texi: Same.
33062 * doc/tm.texi: Same.
33063 * doc/tm.texi.in: Same.
33064 * doc/trouble.texi: Same.
33066 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
33069 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
33070 (collect_checksum_attributes): Set it.
33071 (die_checksum_ordered): Use it.
33073 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33075 PR tree-optimization/79908
33076 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
33077 change: For a VA_ARG whose LHS has been cast away, use
33078 force_gimple_operand to construct the side effects.
33080 2017-03-21 David Malcolm <dmalcolm@redhat.com>
33082 PR translation/80001
33083 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
33084 more amenable to translation.
33085 (oacc_loop_auto_partitions): Likewise.
33087 2017-03-21 Marek Polacek <polacek@redhat.com>
33088 Martin Sebor <msebor@redhat.com>
33090 PR tree-optimization/80109
33091 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
33092 on INTEGRAL_TYPE_P.
33094 2017-03-21 Jakub Jelinek <jakub@redhat.com>
33095 Segher Boessenkool <segher@kernel.crashing.org>
33098 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
33099 check reg_used_between_p between insn and one of succ or succ2
33100 depending on if succ is artificial insn not inserted into insn
33103 2017-03-21 Martin Liska <mliska@suse.cz>
33105 PR gcov-profile/80081
33106 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
33107 * doc/gcc.texi: Include gcov-dump stuff.
33108 * doc/gcov-dump.texi: New file.
33110 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
33112 PR rtl-optimization/79150
33113 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
33114 conditional jump, if the jump is the last insn of the loop.
33116 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33117 Richard Biener <rguenth@suse.de>
33119 PR tree-optimization/79908
33120 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
33121 been cast away, use force_gimple_operand to construct the side
33124 2017-03-21 Martin Liska <mliska@suse.cz>
33126 PR libfortran/79956
33127 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
33130 2017-03-21 Brad Spengler <spender@grsecurity.net>
33133 * plugin.c (htab_hash_plugin): New function.
33134 (add_new_plugin): Use it and adjust.
33135 (parse_plugin_arg_opt): Adjust.
33136 (init_one_plugin): Likewise.
33138 2017-03-21 Richard Biener <rguenther@suse.de>
33140 PR tree-optimization/80032
33141 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
33142 if set force the cleanup to happen unconditionally.
33143 (gimplify_target_expr): Push inserted clobbers with force_uncond
33144 to avoid them being removed by control-dependent DCE.
33146 2017-03-21 Richard Biener <rguenther@suse.de>
33148 PR tree-optimization/80122
33149 * tree-inline.c (copy_bb): Do not expans va-arg packs or
33150 va_arg_pack_len when the inlined call stmt requires pack
33152 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
33154 2017-03-21 Jakub Jelinek <jakub@redhat.com>
33157 * tsan.c (instrument_builtin_call): If the memory model argument
33158 is not a constant, assume it is valid.
33161 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
33164 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
33166 PR rtl-optimization/79910
33167 * combine.c (can_combine_p): Do not allow combining an I0 or I1
33168 if its dest is used by an insn before I2 (other than the combined
33169 insns themselves, which are properly handled already).
33171 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
33174 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
33176 * combine.c (record_used_regs): New static function.
33177 (try_combine): Handle situations where there is an additional
33178 instruction between I2 and I3 which needs to have a LOG_LINK
33182 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
33184 * combine.c (try_combine): Delete redundant i1 test. Call
33185 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
33187 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33190 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
33191 alternatives 13/14.
33193 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33195 PR tree-optimization/80054
33196 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
33197 the optimization if a PHI or any of its arguments is not dominated
33198 by the candidate's basis. Use gphi* rather than gimple* as
33200 (replace_profitable_candidates): Clean up a gimple* variable that
33201 should be a gphi* variable.
33203 2017-03-20 Martin Sebor <msebor@redhat.com>
33206 * doc/extend.texi (attribute constructor): Document present limitation.
33208 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
33211 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
33212 __POWER9_VECTOR__ #ifdef control, change template definition to
33213 use Power9-specific built-in function.
33214 (vec_any_eq): Likewise.
33215 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
33216 to control outcomes from this test.
33217 (vector_ae_<mode>p): For VEC_F modes, likewise.
33219 2017-03-20 Ian Lance Taylor <iant@google.com>
33221 * config/i386/i386.c (ix86_function_regparm): Save an extra
33222 register for -fsplit-stack with DECL_STATIC_CHAIN.
33224 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
33227 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
33228 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
33230 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
33232 * config/riscv/riscv.c (riscv_print_operand): Use "fence
33234 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
33237 2017-03-20 Marek Polacek <polacek@redhat.com>
33240 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
33242 2017-03-20 Richard Biener <rguenther@suse.de>
33244 PR tree-optimization/80113
33245 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
33246 allocate extra SSA name for PHI def.
33247 (add_close_phis_to_outer_loops): Likewise.
33248 (add_close_phis_to_merge_points): Likewise.
33249 (copy_loop_close_phi_args): Likewise.
33250 (copy_cond_phi_nodes): Likewise.
33252 2017-03-20 Martin Liska <mliska@suse.cz>
33254 PR middle-end/79753
33255 * tree-chkp.c (chkp_build_returned_bound): Do not build
33256 returned bounds for a LHS that's not a BOUNDED_P type.
33258 2017-03-20 Martin Liska <mliska@suse.cz>
33262 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
33263 COMPLEX_CST and VECTOR_CST.
33265 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33268 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
33269 target operand. A new splitter adds the clobber statement in case
33270 the target operand is dead anyway.
33272 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
33274 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
33275 to age-old versions of binutils and glibc.
33277 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
33279 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
33281 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
33283 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
33285 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
33287 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
33288 requirement for binutils 2.13.
33290 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
33292 * combine.c (try_combine): Delete redundant i1 test. Call
33293 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
33295 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
33297 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
33298 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
33300 <riscv64-*-elf>: Re-arrange section
33301 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
33302 <riscv32-*-linux>: Likewise.
33303 <riscv64-*-elf>: Likewise
33304 <riscv64-*-linux>: Likewise.
33306 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
33309 * aarch64.opt(verbose-cost-dump): Fix typo.
33311 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
33314 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
33315 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
33317 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
33319 * reload.c (find_reloads): When reloading a nonoffsettable address,
33320 use RELOAD_OTHER for it and its address reloads.
33322 PR rtl-optimization/79910
33323 * combine.c (record_used_regs): New static function.
33324 (try_combine): Handle situations where there is an additional
33325 instruction between I2 and I3 which needs to have a LOG_LINK
33328 2017-03-17 Jeff Law <law@redhat.com>
33330 PR tree-optimization/71437
33331 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
33332 conditional in the hash table first.
33333 (vrp_dom_walker::before_dom_children): Extract condition from
33334 ASSERT_EXPR. Record condition, its inverion and any implied
33335 conditions as well.
33337 2017-03-17 Marek Polacek <polacek@redhat.com>
33338 Markus Trippelsdorf <markus@trippelsdorf.de>
33340 PR tree-optimization/80079
33341 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
33344 2017-03-17 Richard Biener <rguenther@suse.de>
33346 PR middle-end/80075
33347 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
33348 Properly verify the LHS before the RHS possibly claims to be
33350 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
33353 2017-03-17 Martin Jambor <mjambor@suse.cz>
33355 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
33356 (List of -O2 options): Likewise.
33357 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
33360 2017-03-17 Tom de Vries <tom@codesourcery.com>
33362 * gcov-dump.c (print_usage): Print bug_report_url.
33364 2017-03-17 Richard Biener <rguenther@suse.de>
33366 PR middle-end/80050
33367 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
33368 (parser::peek): Likewise.
33370 2017-03-17 Richard Biener <rguenther@suse.de>
33372 PR tree-optimization/80048
33373 * sese.c (free_sese_info): Properly release rename_map and
33374 copied_bb_map elements.
33376 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
33378 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
33379 Add linked-list forward and backlinks. Insert on
33380 construction, remove on destruction.
33381 (class pass_store_merging): Add m_stores_head field.
33382 (pass_store_merging::terminate_and_process_all_chains):
33383 Iterate over m_stores_head list.
33384 (pass_store_merging::terminate_all_aliasing_chains):
33386 (pass_store_merging::execute): Check for debug stmts first.
33387 Push new chains onto the m_stores_head stack.
33389 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
33392 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
33393 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
33394 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
33396 2017-03-16 Jeff Law <law@redhat.com>
33398 PR tree-optimization/71437
33399 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
33400 member function. Implementation moved into after_dom_children
33401 member function and into the threader's thread_outgoing_edges
33403 (dom_opt_dom_walker::after_dom_children): Simplify by moving
33404 some code into new thread_outgoing_edges.
33405 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
33406 definition. Simplify marker handling (do it here). Assume we always
33407 have the available expression and the const/copies tables.
33408 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
33410 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
33411 * tree-vrp.c (equiv_stack): No longer file scoped.
33412 (vrp_dom_walker): New class.
33413 (vrp_dom_walker::before_dom_children): New member function.
33414 (vrp_dom_walker::after_dom_children): Likewise.
33415 (identify_jump_threads): Setup domwalker. Use it rather than
33416 walking edges in a random order by hand. Simplify setup/finalization.
33417 (finalize_jump_threads): Remove.
33418 (vrp_finalize): Do not call identify_jump_threads here.
33419 (execute_vrp): Do it here instead and call thread_through_all_blocks
33422 PR tree-optimization/71437
33423 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
33425 (simplify_stmt_for_jump_threading): Add basic_block argument. All
33427 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
33428 (dom_opt_dom_walker::thread_across_edge): Remove
33429 handle_dominating_asserts argument. All callers changed.
33430 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
33431 changes. Remove calls to lhs_of_dominating_assert. Other
33432 uses of handle_dominating_asserts turn into unconditional code
33433 (simplify_control_stmt_condition_1): Likewise.
33434 (simplify_control_stmt_condition): Likewise.
33435 (thread_through_normal_block, thread_across_edge): Likewise.
33436 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
33437 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
33438 object if it is not an SSA_NAME.
33439 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
33440 before calling into the VRP specific simplifiers.
33441 (identify_jump_threads): Remove handle_dominating_asserts
33444 2017-03-16 Jakub Jelinek <jakub@redhat.com>
33447 * tree-diagnostic.c (default_tree_printer): No longer static.
33448 * tree-diagnostic.h (default_tree_printer): New prototype.
33450 2017-03-16 Tamar Christina <tamar.christina@arm.com>
33452 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
33453 Change ins into fmov.
33455 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33457 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
33458 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
33459 Use h_con constraint for operand 1.
33460 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
33461 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
33463 2017-03-15 Jeff Law <law@redhat.com>
33465 PR tree-optimization/71437
33466 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
33467 (record_temporary_equivalences): Use it.
33469 PR tree-optimization/71437
33470 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
33471 tree-ssa-scopedtables.
33472 (lookup_avail_expr, build_and_record_new_cond): Likewise.
33473 (record_conditions, record_cond, vuse_eq): Likewise.
33474 (record_edge_info): Adjust to API tweak of record_conditions.
33475 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
33476 (record_temporary_equivalences, optimize_stmt): Likewise.
33477 (eliminate_redundant_computations): Likewise.
33478 (record_equivalences_from_stmt): Likewise.
33479 * tree-ssa-scopedtables.c: Include options.h and params.h.
33480 (vuse_eq): New function, moved from tree-ssa-dom.c
33481 (build_and_record_new_cond): Likewise.
33482 (record_conditions): Likewise. Accept vector of conditions rather
33483 than edge_equivalence structure for first argument.
33484 for the first argument.
33485 (avail_exprs_stack::lookup_avail_expr): New member function, moved
33486 from tree-ssa-dom.c.
33487 (avail_exprs_stack::record_cond): Likewise.
33488 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
33489 from tree-ssa-dom.c.
33490 (avail_exprs_stack): Add new member functions lookup_avail_expr
33492 (record_conditions): Declare.
33494 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
33497 * lra-constraints.c (process_alt_operands): Increase reject for
33498 reloading an input/output operand.
33500 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
33503 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
33504 insns to convert from signed/unsigned char/short to IEEE 128-bit
33506 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
33508 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
33511 * config/i386/i386.c (ix86_vector_duplicate_value): Create
33512 subreg of inner mode for values already in registers.
33514 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
33516 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
33517 iteration reg is used after the loop.
33519 2017-03-14 Martin Sebor <msebor@redhat.com>
33521 PR tree-optimization/79800
33522 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
33523 precision in negative-positive range.
33524 (format_floating): Call non-const overload with adjusted precision.
33526 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33529 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
33532 2017-03-14 Martin Sebor <msebor@redhat.com>
33534 PR middle-end/80020
33535 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
33536 * builtins.def (aligned_alloc): Use it.
33539 * Makefile.in (GTFILES): Add calls.c.
33540 * calls.c: Include "gt-calls.h".
33542 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
33544 PR rtl-optimization/79728
33545 * regs.h (struct target_regs): New field
33546 x_contains_allocatable_regs_of_mode.
33547 (contains_allocatable_regs_of_mode): New macro.
33548 * reginfo.c (init_reg_sets_1): Initialize it, and change
33549 contains_reg_of_mode so it includes global regs as well.
33550 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
33551 rather than contains_regs_of_mode.
33553 2017-03-14 Martin Liska <mliska@suse.cz>
33555 * doc/invoke.texi: Document options that can't be combined with
33556 -fcheck-pointer-bounds.
33558 2017-03-14 Martin Liska <mliska@suse.cz>
33560 PR middle-end/79831
33561 * doc/invoke.texi (-Wchkp): Document the option.
33563 2017-03-14 Martin Liska <mliska@suse.cz>
33565 * Makefile.in: Install gcov-dump.
33567 2017-03-14 Martin Liska <mliska@suse.cz>
33569 * multiple_target.c (expand_target_clones): Bail out for
33570 an invalid attribute.
33572 2017-03-14 Richard Biener <rguenther@suse.de>
33574 * alias.c (struct alias_set_entry): Pack properly.
33575 * cfgloop.h (struct loop): Likewise.
33576 * cse.c (struct set): Likewise.
33577 * ipa-utils.c (struct searchc_env): Likewise.
33578 * loop-invariant.c (struct invariant): Likewise.
33579 * lra-remat.c (struct cand): Likewise.
33580 * recog.c (struct change_t): Likewise.
33581 * rtl.h (struct address_info): Likewise.
33582 * symbol-summary.h (function_summary): Likewise.
33583 * tree-loop-distribution.c (struct partition): Likewise.
33584 * tree-object-size.c (struct object_size_info): Likewise.
33585 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
33586 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
33587 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
33588 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
33589 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
33590 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
33591 (struct _stmt_vec_info): Likewise.
33593 2017-03-14 Martin Liska <mliska@suse.cz>
33596 * multiple_target.c (create_dispatcher_calls): Check that
33597 a target can create a function dispatcher.
33599 2017-03-14 Martin Liska <mliska@suse.cz>
33602 * multiple_target.c (expand_target_clones): Drop local.local
33603 flag for default implementation.
33605 2017-03-14 Richard Biener <rguenther@suse.de>
33607 PR tree-optimization/80030
33608 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
33610 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
33612 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
33613 gcc_fallthrough() instead of __attribute__((fallthrough));
33615 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
33617 * doc/gcc.texi: Remove "up" link to (DIR).
33618 * doc/gccint.texi: Ditto.
33620 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
33622 * doc/install.texi (Specific) <avr>: Remove reference to
33625 2017-03-13 Jeff Law <law@redhat.com>
33627 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
33628 attribute rather than comments.
33630 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
33631 match_scratch operand is highest.
33633 2017-03-13 Martin Liska <mliska@suse.cz>
33635 PR middle-end/78339
33636 * ipa-pure-const.c (warn_function_noreturn): If the declarations
33637 is a CHKP clone, use original declaration.
33639 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33641 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
33642 (arc_conditional_register_usage): Use a different allocation order
33643 when optimizing for size.
33644 * common/config/arc/arc-common.c (arc_option_optimization_table):
33645 Section anchors default on when optimizing for size.
33647 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33649 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
33651 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33653 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
33654 * config/arc/arc.md (cpu_facility): Add cd variant.
33655 (*movqi_insn): Add code density variant.
33656 (*movhi_insn): Likewise.
33657 (*movqi_insn): Likewise.
33658 (*addsi3_mixed): Likewise.
33659 (subsi3_insn): Likewise.
33661 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33663 * config/arc/arc.md (movsi_cond_exec): Update constraint.
33665 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33667 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
33668 expressions with MINUS and UNARY ops.
33670 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33673 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
33675 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
33676 between vec_select and vector argument.
33677 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
33678 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
33679 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
33680 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
33681 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
33682 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
33684 2017-03-13 Richard Biener <rguenther@suse.de>
33687 * params.def (vect-max-peeling-for-alignment): Fix typo.
33689 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33691 * doc/install.texi (Specific) <mips-*-*>: Remove description of
33692 issue that only occurred with binutils below 2.18.
33694 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33696 * doc/install.texi (Specific) <cris-axis-elf>: No longer
33697 refer to binutils 2.11/2.12 minimum.
33699 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33701 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
33702 ftp.kernel.org and simplify binutils requirement.
33704 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
33706 * doc/invoke.texi (Warning Options): Fix spelling of link-time
33708 (Optimize Options): Ditto. Also remove redundancy.
33710 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33712 PR translation/79848
33713 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
33715 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
33716 to G_ to avoid double translation.
33718 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33720 PR translation/79923
33721 * auto-profile.c (get_combined_location): Convert leading
33722 character of diagnostics to lower case and remove trailing period.
33723 (read_profile): Likewise for various diagnostics.
33724 * config/arm/arm.c (arm_option_override): Remove trailing period
33725 from various diagnostics.
33726 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
33727 (msp430_expand_delay_cycles): Likewise.
33729 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33732 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
33733 full command-line argument, rather than just "str".
33734 (aarch64_validate_march): Likewise.
33735 (aarch64_validate_mtune): Likewise.
33737 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
33739 PR rtl-optimization/78911
33740 * lra-assigns.c (must_not_spill_p): New function.
33741 (spill_for): Use it.
33743 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33745 PR tree-optimization/79981
33746 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
33747 ATOMIC_COMPARE_EXCHANGE ifn result.
33748 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
33749 IFN_ATOMIC_COMPARE_EXCHANGE.
33751 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33754 * opts.c (parse_sanitizer_options): Add missing question mark to
33755 "did you mean" message.
33757 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33759 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
33761 (VMULEUH_UNS): Likewise.
33762 (VMULOUB_UNS): Likewise.
33763 (VMULOUH_UNS): Likewise.
33764 * config/rs6000/rs6000.c (builtin_function_type): Remove
33765 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
33767 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33770 * read-rtl-function.c (function_reader::read_rtx_operand): Update
33771 x with result of extra_parsing_for_operand_code_0.
33772 (function_reader::extra_parsing_for_operand_code_0): Convert
33773 return type from void to rtx, returning x. When reading
33774 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
33775 larger size containing struct block_symbol.
33777 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
33779 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
33780 -mfloat128-hardware without -m64.
33782 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
33785 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
33786 entries to the case statement that marks unsigned arguments to
33787 overloaded functions.
33789 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33791 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
33792 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
33794 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
33797 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
33798 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
33800 2017-03-10 Martin Liska <mliska@suse.cz>
33804 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
33805 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
33808 2017-03-10 Olivier Hainque <hainque@adacore.com>
33810 * tree-switch-conversion (array_value_type): Start by resetting
33811 candidate type to it's main variant.
33813 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33815 PR rtl-optimization/79909
33816 * combine.c (try_combine): Use simplify_replace_rtx on individual
33817 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
33818 of the whole CALL_INSN_FUNCTION_USAGE.
33820 PR tree-optimization/79972
33821 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
33822 get_range_info on SSA_NAMEs. Formatting fixes.
33824 2017-03-10 Richard Biener <rguenther@suse.de>
33825 Jakub Jelinek <jakub@redhat.com>
33827 PR tree-optimization/77975
33828 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
33829 edge to be constant.
33830 (get_val_for): For constant x return it. Formatting fix.
33831 (loop_niter_by_eval): Avoid pointless looping if the next iteration
33832 would use the same bases as the current one.
33834 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33836 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
33837 instead of vec_select for V1TImode.
33838 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
33840 (VSX_LE_128): Add V1TI to this mode iterator.
33841 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
33842 (*vsx_le_perm_store_<mode>): Likewise.
33843 (pre-reload splitter for VSX stores): Likewise.
33844 (post-reload splitter for VSX stores): Likewise.
33845 (*vsx_xxpermdi2_le_<mode>): Likewise.
33846 (*vsx_lxvd2x2_le_<mode>): Likewise.
33847 (*vsx_stxvd2x2_le_<mode>): Likewise.
33849 2017-03-09 Michael Eager <eager@eagercon.com>
33851 Correct failures with --enable-checking=yes,rtl.
33853 * config/microblaze/microblaze.c (microblaze_expand_shift):
33854 Replace GET_CODE test with CONST_INT_P and INTVAL test with
33855 test for const0_rtx.
33856 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
33857 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
33859 2017-03-09 Richard Biener <rguenther@suse.de>
33861 PR tree-optimization/79977
33862 * graphite-scop-detection.c (scop_detection::merge_sese):
33863 Handle the case of extra exits to blocks dominating the entry.
33865 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
33867 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
33870 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
33872 PR rtl-optimization/79949
33873 * lra-constraints.c (process_alt_operands): Check memory when
33874 trying to predict a cycle. Print about the overall increase.
33876 2017-03-09 Richard Biener <rguenther@suse.de>
33878 PR middle-end/79971
33879 * gimple-expr.c (useless_type_conversion_p): Preserve
33880 TYPE_SATURATING for fixed-point types.
33882 2017-03-09 Richard Biener <rguenther@suse.de>
33885 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
33886 alignment of BLKmode params.
33888 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33891 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
33892 (VALL_NO_V2Q): Likewise.
33894 * config/aarch64/aarch64-simd.md
33895 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
33897 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
33898 VALL_NO_V2Q mode iterator.
33899 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
33901 2017-03-09 Martin Liska <mliska@suse.cz>
33903 PR tree-optimization/79631
33904 * tree-chkp-opt.c (chkp_is_constant_addr): Call
33905 tree_int_cst_sign_bit just for INTEGER constants.
33907 2017-03-09 Martin Liska <mliska@suse.cz>
33911 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
33914 2017-03-09 Marek Polacek <polacek@redhat.com>
33917 * tree.c (inchash::add_expr): Handle TREE_VEC.
33919 2017-03-09 Martin Liska <mliska@suse.cz>
33922 (chkp_narrow_size_and_offset): New function.
33923 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
33924 (void chkp_parse_bit_field_ref): New function.
33925 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
33926 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
33928 2017-03-09 Martin Liska <mliska@suse.cz>
33931 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
33932 (chkp_find_bounds_1): Remove gcc_unreachable.
33934 2017-03-09 Jakub Jelinek <jakub@redhat.com>
33937 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
33938 BUILT_IN_SYNC*, determine the access type from the size suffix and
33939 always build a MEM_REF with that type. Handle forgotten
33940 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
33943 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
33944 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
33945 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
33946 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
33947 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
33948 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
33949 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
33950 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
33951 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
33952 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
33953 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
33954 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
33955 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
33956 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
33957 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
33958 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
33959 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
33960 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
33961 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
33962 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
33963 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
33964 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
33965 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
33966 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
33967 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
33968 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
33969 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
33970 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
33971 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
33972 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
33973 definitions outside of __OPTIMIZE__ guarded section.
33976 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
33977 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
33978 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
33979 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
33982 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33984 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
33985 ("vfenez<mode>"): Add missing constraints.
33987 2017-03-08 Martin Sebor <msebor@redhat.com>
33990 * config/nds32/nds32.c (nds32_option_override):
33991 Fix misspelled diagnostic.
33993 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33996 * gimplify.c (gimplify_omp_for): Replace index var in outer
33997 taskloop statement with an artificial variable and add
33998 OMP_CLAUSE_PRIVATE clause for it.
34000 2017-03-08 Richard Biener <rguenther@suse.de>
34002 PR tree-optimization/79955
34003 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
34004 for accesses that are completely outside of the variable.
34006 2017-03-08 Andrew Haley <aph@redhat.com>
34008 PR tree-optimization/79943
34009 * tree-ssa-loop-split.c (compute_new_first_bound): When
34010 calculating the new upper bound, (END-BEG) should be added, not
34013 2017-03-08 Jakub Jelinek <jakub@redhat.com>
34015 * config/avr/avr.md (setmemhi): Make sure match_dup
34016 operand number comes before match_scratch.
34018 2017-03-08 Richard Biener <rguenther@suse.de>
34020 PR tree-optimization/79920
34021 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
34022 with ncopies == 1 to ...
34023 (vect_transform_slp_perm_load): ... here. Properly compute
34024 all element loads by iterating VF times over the group. Do
34025 not handle ncopies (computed in a broken way) in
34026 vect_create_mask_and_perm.
34028 2017-03-08 Jakub Jelinek <jakub@redhat.com>
34031 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
34032 is a uniform vector, use uniform_vector_p return value instead of
34033 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
34035 2017-03-07 Marek Polacek <polacek@redhat.com>
34037 PR middle-end/79809
34038 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
34039 (alloca_call_type): Likewise.
34041 2017-03-07 Martin Liska <mliska@suse.cz>
34043 * gcov.c (process_args): Put comment to correct location.
34045 2017-03-07 Martin Liska <mliska@suse.cz>
34047 PR middle-end/68270
34048 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
34049 Use array_at_struct_end_p instead of DECL_CHAIN (field).
34050 (chkp_narrow_bounds_for_field): Likewise.
34051 (chkp_parse_array_and_component_ref): Pass one more argument to
34054 2017-03-07 Richard Biener <rguenther@suse.de>
34056 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
34059 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
34061 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
34062 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
34064 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34067 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
34068 to end of description.
34069 (PARAM_MAX_STORES_TO_MERGE): Likewise.
34071 2017-03-07 Jakub Jelinek <jakub@redhat.com>
34073 PR rtl-optimization/79901
34074 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
34076 (*avx512f_<code><mode>3<mask_name>): ... this.
34077 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
34078 iterator instead of VI8_AVX2_AVX512BW.
34080 PR rtl-optimization/79901
34081 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
34082 min/max expander, expand it using expand_vec_cond_expr.
34085 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
34088 2017-03-06 Jakub Jelinek <jakub@redhat.com>
34091 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
34092 to void * for PCH reasons.
34093 * dwarf2out.c (output_loc_operands, output_die): Cast
34094 v.val_vec.array to unsigned char *.
34096 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
34099 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
34102 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
34104 PR rtl-optimization/79571
34105 * lra-constraints.c (process_alt_operands): Calculate static
34106 reject and subtract it from overall when only addresses will be
34109 2017-03-06 Julia Koval <julia.koval@intel.com>
34112 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
34113 incoming stack boundary to 128 for 64-bit targets.
34115 2017-03-06 Richard Biener <rguenther@suse.de>
34117 PR tree-optimization/79894
34118 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
34119 to NULL after folding it.
34121 2017-03-06 Richard Biener <rguenther@suse.de>
34123 PR tree-optimization/79824
34124 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
34125 check disabling peeling for gaps.
34127 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
34129 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
34130 attributes): Document gettimeofday.
34132 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
34134 * config/s390/s390.c (s390_option_override_internal): Set
34135 PARAM_MIN_VECT_LOOP_BOUND
34137 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
34139 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
34140 * config/s390/s390.md: Likewise.
34142 2017-03-06 Jakub Jelinek <jakub@redhat.com>
34145 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
34146 (<avx2_avx512>_perm<mode>): Rename to ...
34147 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
34149 (<avx512>_perm<mode>_mask): Rename to ...
34150 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
34152 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
34153 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
34154 instead of VI8F_256_512.
34155 (avx512f_perm<mode>): New define_expand.
34156 (avx512f_perm<mode>_mask): Likewise.
34157 (avx512f_perm<mode>_1<mask_name>): New define_insn.
34158 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
34160 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34162 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
34163 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
34165 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
34167 2017-03-06 Martin Liska <mliska@suse.cz>
34170 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
34171 when having a SSA NAME w/o VAR_DECL assigned to it.
34173 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34175 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
34176 msa_dpsub_<su>_d): Fix MODE for vec_select.
34178 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34180 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
34182 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
34184 2017-03-06 Richard Biener <rguenther@suse.de>
34186 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
34187 * plugin.c (register_plugin_info): Likewise.
34188 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
34190 2017-03-05 Jakub Jelinek <jakub@redhat.com>
34192 * config/i386/sse.md (sse_storehps, sse_storelps,
34193 avx_<castmode><avxsizesuffix>_<castmode>,
34194 avx512f_<castmode><avxsizesuffix>_<castmode>,
34195 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
34196 in condition that at least one operand is not a MEM.
34198 2017-03-03 Jakub Jelinek <jakub@redhat.com>
34200 PR middle-end/79805
34201 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
34202 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
34204 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
34205 gimple_call_nothrow_p flag based on whether original builtin can throw.
34206 If it can, emit following stmts on the fallthrough edge.
34207 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
34208 don't create new bb if inserting just debug stmts on the edge, try to
34209 insert them on the fallthru bb or just reset debug stmts.
34211 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
34214 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
34215 restore recog_data (including the operand rtxes inside it) around
34216 the call to get_insn_template.
34218 2017-03-03 Martin Sebor <msebor@redhat.com>
34220 PR tree-optimization/79699
34221 * context.c (context::~context): Free MPFR caches to avoid
34222 a memory leak on program exit.
34224 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34226 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
34227 Use wide_int::ulow () instead of .elt (0).
34229 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
34231 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
34232 (*pushxf): Limit oF constraint to 32bit targets and add oC
34233 constraint for 64bit targets.
34234 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
34235 (*pushdf): Change rmF constraint to rmC.
34237 2017-03-03 Martin Liska <mliska@suse.cz>
34239 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
34240 Remove unused variable.
34242 2017-03-03 Jakub Jelinek <jakub@redhat.com>
34245 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
34246 is a memory operand, increase num_memory.
34247 (ix86_expand_args_builtin): Likewise.
34249 2017-03-03 Jan Hubicka <jh@suse.cz>
34252 * ipa-devirt.c (maybe_record_node): Properly handle
34253 __cxa_pure_virtual visibility.
34255 2017-03-03 Martin Liska <mliska@suse.cz>
34257 PR tree-optimization/79803
34258 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
34260 (pass_loop_prefetch::execute): Disabled optimization if an
34261 assumption about L1 cache size is not met.
34263 2017-03-03 Martin Liska <mliska@suse.cz>
34265 PR rtl-optimization/79574
34266 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
34267 (hash_scan_set): Likewise.
34268 (dump_hash_table): Likewise.
34269 (hoist_code): Likewise.
34271 2017-03-03 Richard Biener <rguenther@suse.de>
34273 * fixed-value.c (fixed_from_string): Restore use of elt (1)
34274 in place of uhigh ().
34275 (fixed_convert_from_real): Likewise.
34277 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
34280 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
34282 2017-03-03 Richard Biener <rguenther@suse.de>
34284 PR middle-end/79818
34285 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
34286 TYPE_OVERFLOW_UNDEFINED check.
34288 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34290 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
34292 (vector_ae_<mode>_p): Likewise.
34293 (vector_nez_<mode>_p): Likewise.
34294 (vector_ne_v2di_p): Likewise.
34295 (vector_ae_v2di_p): Likewise.
34296 (vector_ne_<mode>_p): Likewise.
34297 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
34299 (vsx_tsqrt<mode>2_fe): Likewise.
34301 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
34304 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
34306 2017-03-02 Jakub Jelinek <jakub@redhat.com>
34308 PR rtl-optimization/79780
34309 * cprop.c (one_cprop_pass): When second and further conditional trap
34310 in a single basic block is turned into an unconditional trap, turn it
34311 into a deleted note to avoid RTL verification failures.
34313 2017-03-02 Richard Biener <rguenther@suse.de>
34315 * fold-const.c (const_binop): Use ulow () instead of elt (0).
34317 2017-03-02 Richard Biener <rguenther@suse.de>
34319 PR tree-optimization/79345
34321 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
34322 param and abort the walk, returning -1 if it is hit.
34323 (walk_aliased_vdefs): Take a limit param and pass it on.
34324 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
34325 defaulting to 0 and return a signed int.
34326 * tree-ssa-uninit.c (struct check_defs_data): New struct.
34327 (check_defs): New helper.
34328 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
34329 about uninitialized memory.
34330 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
34331 bogus uninitialized warning.
34332 (fixed_convert_from_real): Likewise.
34334 2017-03-02 Bin Cheng <bin.cheng@arm.com>
34336 PR tree-optimization/66768
34337 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
34338 iv_use if base object can't be determined.
34340 2017-03-02 Jakub Jelinek <jakub@redhat.com>
34342 PR tree-optimization/79345
34343 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
34344 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
34345 (get_pattern_stats): Initialize it.
34346 * genemit.c (gen_expand): Verify match_scratch numbers come after
34347 match_operand/match_dup numbers.
34348 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
34349 match_scratch numbers.
34350 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
34352 * config/s390/s390.md (trunctdsd2): Likewise.
34354 2017-03-02 Richard Biener <rguenther@suse.de>
34356 * wide-int.h (wide_int_storage::operator=): Implement in terms
34359 2017-03-02 Richard Biener <rguenther@suse.de>
34361 PR tree-optimization/79777
34362 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
34363 the to insert expression to sth existing.
34365 2017-03-01 Martin Sebor <msebor@redhat.com>
34367 PR middle-end/79692
34368 * gimple-ssa-sprintf.c
34369 (directive::known_width_and_precision): New function.
34370 (format_integer): Use it.
34371 (get_mpfr_format_length): Consider the full range of precision
34372 when computing %g output with the # flag. Set the likely byte
34373 count to 3 rather than 1 when precision is indeterminate.
34374 (format_floating): Correct the lower bound of precision.
34376 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34378 * doc/invoke.texi: Document default code model for 64-bit Linux.
34380 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34383 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
34384 udiv rather than div since input pattern is unsigned.
34386 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
34388 * config/i386/i386.c (print_reg): Warn for values of
34389 unsupported size in integer register.
34391 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
34394 * config/rs6000/predicates.md (current_file_function_operand): Do
34395 not allow self calls to be local if the function is replaceable.
34397 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
34400 * config/rs6000/altivec.h (vec_ctz and others): Change the
34401 preprocessor macro that controls conditional compilation from
34402 _ARCH_PWR9 to __POWER9_VECTOR__.
34403 (vec_all_ne): Change parameterization of __altivec_scalar_pred
34404 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
34405 control (instead of _ARCH_PWR9 control) so that template
34406 definition uses power9-specific function.
34407 (vec_any_eq): Likewise.
34408 (vec_all_ne): Change macro definition to use a power9-specific
34409 expansion under #ifdef __POWER9_VECTOR__ control (instead of
34410 _ARCH_PWR9 control).
34411 (vec_any_eq) Likewise.
34412 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
34413 expansion for CMPNEF to remove support for xvcmpnesp instruction.
34414 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
34415 support for xvcmpnedp instruction.
34416 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
34417 macro expansion so that Power9 implementation of vec_all_ne does
34418 not use the AltiVec predicate framework.
34419 (VCMPNEH_P): Likewise.
34420 (VCMPNEW_P): Likewise.
34421 (VCMPNED_P): Likewise.
34422 (VCMPNEFP_P): Likewise.
34423 (VCMPNEDP_P): Likewise.
34424 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
34425 implementation of vec_any_eq to not use AltiVec predicate
34427 (VCMPAEH_P): Likewise.
34428 (VCMPAEW_P): Likewise.
34429 (VCMPAED_P): Likewise.
34430 (VCMPAEFP_P): Likewise.
34431 (VCMPAEDP_P): Likewise.
34432 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
34433 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
34434 not use the AltiVec predicate framework.
34435 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
34436 of vec_any_eq to not use AltiVec predicate framework.
34437 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34438 support for predefined __POWER9_VECTOR__ macro to indicate that
34439 Power9 instruction selection is enabled.
34440 (altivec_overloaded_builtins): Remove extraneous
34441 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
34442 function argument types RS6000_BTI_bool_V16QI and
34443 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
34444 entry for overloaded function argument types RS6000_BTI_bool_V4SI
34445 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
34446 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
34447 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
34448 Power9 for implementations of vec_cmpne. Change the signature for
34449 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
34450 (representing vec_all_ne) to remove the previously described first
34451 argument of type RS6000_BTI_INTSI, as this was an artifact of
34452 reliance on the AltiVec predicate framework, which is no longer
34453 used in the implementation of these functions. Add
34454 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
34455 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
34456 since, unlike the AltiVec predicate framework implementation, we
34457 do not share function descriptors between vec_alle and vec_anyeq.
34458 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
34459 set of modes that receive special treatment even when
34460 TARGET_P9_VECTOR is true. The special treatment emits code that
34461 does not depend on Power9 instructions.
34462 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
34463 define_expand to not rely on AltiVec predicate framework.
34464 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
34466 (vector_ne_v2di_p): Change this define_expand to not rely on
34467 AltiVec predicate framework.
34468 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
34470 (vector_ne_<mode>_p): Change this define_expand to not rely on
34471 AltiVec predicate framework.
34472 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
34474 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
34475 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
34476 define_insn pattern.
34477 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
34478 define_insn pattern because the xvcmpne<VSs>. instruction is not
34480 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
34481 instruction is not supported.
34483 2017-03-01 Jakub Jelinek <jakub@redhat.com>
34485 * config/nvptx/nvptx.c: Include intl.h.
34487 2017-03-01 Martin Jambor <mjambor@suse.cz>
34490 * ipa-prop.h (ipa_bits): Removed field known.
34491 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
34492 to pointers. Adjusted their comments to warn about their sharing.
34493 (ipcp_transformation_summary): Change bits to a vector of pointers.
34494 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
34495 (ipa_get_ipa_bits_for_value): Declare.
34496 * tree-vrp.h (value_range): Mark as GTY((for_user)).
34497 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
34498 (ipa_bits_hash_table): Likewise.
34499 (ipa_vr_ggc_hash_traits): Likewise.
34500 (ipa_vr_hash_table): Likewise.
34501 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
34502 being pointers and vr_known being removed.
34503 (ipa_set_jf_unknown): Likewise.
34504 (ipa_get_ipa_bits_for_value): New function.
34505 (ipa_set_jfunc_bits): Likewise.
34506 (ipa_get_value_range): New overloaded functions.
34507 (ipa_set_jfunc_vr): Likewise.
34508 (ipa_compute_jump_functions_for_edge): Use the above functions to
34509 construct bits and vr parts of jump functions.
34510 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
34511 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
34513 (ipcp_grow_transformations_if_necessary): Also allocate
34514 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
34516 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
34517 them. Fix too long lines.
34518 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
34519 vr_known being removed.
34520 (ipa_read_jump_function): Use new setter functions to construct bits
34521 and vr parts of jump functions or set them to NULL.
34522 (write_ipcp_transformation_info): Adjust for bits being pointers.
34523 (read_ipcp_transformation_info): Likewise.
34524 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
34526 Include gt-ipa-prop.h.
34527 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
34529 (ipcp_store_bits_results): Likewise.
34530 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
34531 Do not write to existing jump functions but use a temporary instead.
34533 2017-03-01 Jakub Jelinek <jakub@redhat.com>
34536 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
34537 attempt to use its first operand as BIT_FIELD_REF base.
34539 2017-03-01 Richard Biener <rguenther@suse.de>
34541 PR middle-end/79721
34542 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
34543 interpolating formula in wrapping arithmetic.
34544 (chrec_apply): Convert chrec_evaluate return value to wanted type.
34546 2017-03-01 Jakub Jelinek <jakub@redhat.com>
34548 PR tree-optimization/79734
34549 * tree-vect-generic.c (expand_vector_condition): Optimize
34550 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
34551 Handle VEC_COND_EXPR where comparison has different inner width from
34552 type's inner width.
34554 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
34556 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
34557 markup, and similar issues. Remove @opindex entries for things
34558 that aren't options. Add missing -mmpy-option entries.
34560 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34562 PR tree-optimization/79737
34563 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
34564 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
34565 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
34566 instead of byte_size. Formatting fix.
34567 (shift_bytes_in_array_right): Formatting fix.
34569 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
34572 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
34573 condition on optimize for the leaf function test.
34575 2017-02-28 Martin Liska <mliska@suse.cz>
34578 * read-rtl-function.c (function_reader::handle_unknown_directive):
34579 Bail out when one uses -flto.
34581 2017-02-28 Martin Liska <mliska@suse.cz>
34583 * common.opt: Replace space with tabular for options of <number>
34585 * config/i386/i386.opt: Show <number> value for
34586 -mlarge-data-threshold.
34587 * opts.c (print_filtered_help): Do not display number in hexadecimal
34590 2017-02-28 Martin Liska <mliska@suse.cz>
34592 * common.opt: Fix --help=option -Q for options which are of
34595 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
34597 * config/i386/i386.c (print_reg): Error out for values
34598 of 8-bit size in invalid integer register.
34600 2017-02-28 Martin Sebor <msebor@redhat.com>
34602 PR tree-optimization/79691
34603 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
34605 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34608 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
34609 gcc_unreachable with output_operand_lossage.
34611 2017-02-28 Richard Biener <rguenther@suse.de>
34613 PR tree-optimization/79740
34614 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
34616 (visit_nary_op): Insert the nary into the hashtable if we
34617 pattern-matched sth.
34618 * tree-ssa-pre.c (eliminate_insert): Robustify.
34620 2017-02-28 Richard Biener <rguenther@suse.de>
34622 PR middle-end/79731
34623 * fold-const.c (decode_field_reference): Reject out-of-bound
34626 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34628 * config/i386/i386.c: Include intl.h.
34629 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
34630 instead of just cond ? "..." : "...".
34631 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
34632 * coverage.c (read_counts_file): Likewise.
34633 * omp-offload.c: Include intl.h.
34634 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
34635 of just cond ? "..." : "...".
34636 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
34637 of just cond ? "..." : "...".
34639 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
34642 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
34644 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
34645 'tune for' CPU name.
34646 * config/arm/arm-cpu-data.h: Regenerated.
34648 2017-02-28 Richard Biener <rguenther@suse.de>
34650 PR tree-optimization/79732
34651 * tree-inline.c (expand_call_inline): Do not shadow var.
34653 2017-02-28 Richard Biener <rguenther@suse.de>
34655 PR tree-optimization/79723
34656 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
34657 address-space properly.
34659 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
34661 * doc/optinfo.texi (Optimization groups): Fix option used for
34663 * doc/invoke.texi (-fopt-info): Document "omp".
34664 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
34665 (OPTGROUP_ALL): Add OPTGROUP_OMP.
34666 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
34667 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
34668 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
34670 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
34672 * dumpfile.c (optgroup_options): Instead of "openmp", associate
34673 OPTGROUP_OMP with "omp".
34675 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
34678 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
34679 for arithmetic shift of unsigned V2DI.
34681 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
34683 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
34684 arc/linux.h headers.
34685 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
34686 (LINK_SPEC): Likewise.
34687 (ARC_TLS_EXTRA_START_SPEC): Likewise.
34688 (EXTRA_SPECS): Likewise.
34689 (STARTFILE_SPEC): Likewise.
34690 (ENDFILE_SPEC): Likewise.
34691 (LIB_SPEC): Likewise.
34692 (TARGET_SDATA_DEFAULT): Likewise.
34693 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
34694 (MULTILIB_DEFAULTS): Likewise.
34695 (DWARF2_UNWIND_INFO): Likewise.
34696 * config/arc/big.h: New file.
34697 * config/arc/elf.h: Likewise.
34698 * config/arc/linux.h: Likewise.
34699 * config/arc/t-uClibc: Remove.
34701 2017-02-27 Bin Cheng <bin.cheng@arm.com>
34703 PR tree-optimization/77536
34704 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
34705 (tree_transform_and_unroll_loop): Use above function to compute the
34706 estimated niter of unrolled loop and use it when scaling profile.
34707 Also use count info rather than frequency if it's non-zero.
34708 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
34709 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
34710 (vect_transform_loop): Call above function.
34712 2017-02-27 Richard Biener <rguenther@suse.de>
34714 PR tree-optimization/45397
34715 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
34716 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
34717 (visit_nary_op): Add pattern matching for CSEing sign-changed
34718 or truncated operations with wider ones.
34720 2017-02-27 Richard Biener <rguenther@suse.de>
34722 PR tree-optimization/79690
34723 * tree-vect-stmts.c (vectorizable_store): Use vector type
34724 built from the DR with address-space.
34726 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
34728 * doc/invoke.texi (Optimize Options): Refine the description
34729 of asan-use-after-return.
34731 2017-02-25 Alan Modra <amodra@gmail.com>
34733 PR rtl-optimization/79584
34734 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
34735 base, not ad->base_term, the reg within base. Remove assertion
34736 that ad->base == ad->base_term. Replace gen_int_mode using
34737 bogus mode with const0_rtx.
34739 2017-02-25 Jakub Jelinek <jakub@redhat.com>
34741 PR middle-end/79396
34742 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
34743 FMA_EXPR like tcc_binary or tcc_unary.
34745 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
34748 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
34750 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
34751 (output_loc_operands): Handle DW_OP_call_ref and
34752 DW_OP_GNU_variable_value.
34753 (struct variable_value_struct): New type.
34754 (struct variable_value_hasher): Likewise.
34755 (variable_value_hash): New variable.
34756 (string_types): Remove.
34757 (copy_loc_descr): New function.
34758 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
34759 (prepend_loc_descr_to_each): New function.
34760 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
34761 instead of add_loc_descr_to_each if the first argument is single
34762 location list and the second has multiple.
34763 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
34764 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
34765 when looking for variable value which doesn't have other location info.
34766 (loc_list_from_tree): Formatting fix.
34767 (gen_array_type_die): Simplify DW_AT_string_length handling.
34768 (adjust_string_types): Remove.
34769 (gen_subprogram_die): Don't call adjust_string_types nor test/set
34770 string_types. Call resolve_variable_values.
34771 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
34772 (resolve_addr_in_expr): Likewise. Add A argument.
34773 (copy_deref_exprloc): Remove deref argument. Adjust for the
34774 original expression being DW_OP_GNU_variable_value with optionally
34775 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
34776 optionally after it.
34777 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
34778 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
34779 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
34780 (variable_value_hasher::hash, variable_value_hasher::equal): New
34782 (resolve_variable_value_in_expr, resolve_variable_value,
34783 resolve_variable_values, note_variable_value_in_expr,
34784 note_variable_value): New functions.
34785 (dwarf2out_early_finish): Call note_variable_value on all toplevel
34788 2017-02-24 Jakub Jelinek <jakub@redhat.com>
34791 * opts.h (handle_generated_option): Add GENERATED_P argument.
34792 * opts-common.c (handle_option): Adjust function comment.
34793 (handle_generated_option): Add GENERATED_P argument, pass it to
34795 (control_warning_option): Pass false to handle_generated_option
34797 * opts.c (maybe_default_option): Pass true to handle_generated_option
34799 * optc-gen.awk: Likewise.
34801 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34803 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
34804 a REG, look at the REG it is a SUBREG of.
34805 (splitter for cmpeqsi_t): Ditto.
34807 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34809 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
34810 the special USEs with the pattern of the insn, not the insn itself.
34812 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
34815 * doc/invoke.texi: Document -mload-store-pairs.
34817 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34818 Sandra Loosemore <sandra@codesourcery.com>
34820 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
34821 argument isn't a CONST_INT.
34822 (nios2_alternate_compare_const): Assert op is a CONST_INT.
34823 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
34824 (nios2_validate_compare): Bypass alternate compare logic if *op2
34825 is not a CONST_INT.
34826 (ldstwm_operation_p): Return false if first_base is not a REG or
34827 if first_offset is not a CONST_INT.
34829 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34831 * config/cris/cris.md: Use correct operand in a define_peephole2.
34833 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34835 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
34837 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34839 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
34840 this_insn if it is an INSN or JUMP_INSN.
34841 (force_offsettable): Look at base, not at addr.
34842 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
34843 on things that aren't necessarily CONST_INTs.
34845 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
34847 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
34848 -mfpmath=sse is the default also for x86-32 targets with SSE2
34849 instruction set when @option{-ffast-math} is enabled
34851 2017-02-24 Jeff Law <law@redhat.com>
34853 PR rtl-optimizatoin/79286
34854 * ira.c (update_equiv_regs): Drop may_trap_p exception to
34857 2017-02-24 Richard Biener <rguenther@suse.de>
34859 PR tree-optimization/79389
34860 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
34863 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
34865 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
34866 function comment to reflect reality.
34867 (loop_exits_before_overflow): Fix typo in function description.
34869 2017-02-24 Richard Biener <rguenther@suse.de>
34871 PR tree-optimization/79389
34872 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
34873 properly that a threading opportunity exists. Detect conditional
34874 copy/constant propagation opportunities.
34876 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
34878 * config/visium/visium.md (type): Add trap.
34879 (b): New mode attribute.
34880 (*btst): Rename into...
34881 (*btst<mode>): ...this and adjust.
34882 (*cbranchsi4_btst_insn): Rename into...
34883 (*cbranch<mode>4_btst_insn): ...this and adjust.
34884 (trap): New define_insn.
34886 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34888 PR tree-optimization/79389
34889 * ifcvt.c (struct noce_if_info): Add rev_cond field.
34890 (noce_reversed_cond_code): New function.
34891 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
34892 reversed_comparison_code. Formatting fix.
34893 (noce_try_store_flag): Test rev_cond != NULL in addition to
34894 reversed_comparison_code.
34895 (noce_try_store_flag_constants): Likewise.
34896 (noce_try_store_flag_mask): Likewise.
34897 (noce_try_addcc): Use rev_cond if non-NULL instead of
34898 reversed_comparison_code.
34899 (noce_try_cmove_arith): Likewise. Formatting fixes.
34900 (noce_try_minmax, noce_try_abs): Clear rev_cond.
34901 (noce_find_if_block): Initialize rev_cond.
34902 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
34903 instead of false as last argument never attempt to reverse it
34906 2017-02-23 Bin Cheng <bin.cheng@arm.com>
34908 PR tree-optimization/79663
34909 * tree-predcom.c (combine_chains): Process refs in reverse order
34910 only for ZERO length chains, and add explaining comment.
34912 2017-02-23 Jeff Law <law@redhat.com>
34914 PR tree-optimization/79578
34915 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
34916 in call to operand_equal_p.
34918 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
34921 * config/i386/cpuid.h: Fix another undefined behavior.
34923 2017-02-23 Richard Biener <rguenther@suse.de>
34925 PR tree-optimization/79683
34926 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
34927 vector types for data-refs.
34929 2017-02-23 Martin Liska <mliska@suse.cz>
34931 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
34933 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34935 PR middle-end/79665
34936 * internal-fn.c (get_range_pos_neg): Moved to ...
34937 * tree.c (get_range_pos_neg): ... here. No longer static.
34938 * tree.h (get_range_pos_neg): New prototype.
34939 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
34940 are known to be in between 0 and signed maximum inclusive, try to
34941 expand both unsigned and signed divmod and use the cheaper one from
34944 2017-02-22 Jeff Law <law@redhat.com>
34946 PR tree-optimization/79578
34947 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
34948 to compare base operands.
34950 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
34953 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
34954 gpc_reg_operand instead of fpr_reg_operand.
34956 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
34958 * config/mips/mips.c (mips_return_in_memory): Force FP
34959 vector types to be returned in memory for o32 ABI.
34961 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34963 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
34964 instead of DW_TAG_member for static data member declarations and don't
34965 set no_linkage_name for static inline data members.
34966 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
34969 2017-02-22 Martin Liska <mliska@suse.cz>
34971 * doc/invoke.texi: Replace inequality signs with square brackets
34974 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
34977 * lra-constraints.c (simplify_operand_subreg): Handle
34978 WORD_REGISTER_OPERATIONS targets.
34980 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34983 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
34984 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
34985 elimination by swapping fld*.
34987 2017-02-22 Richard Biener <rguenther@suse.de>
34989 PR tree-optimization/79673
34990 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
34991 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
34992 irrelevant address-space qualifiers and avoiding a
34993 ADDR_SPACE_CONVERT_EXPR from fold_convert.
34995 2017-02-22 Richard Biener <rguenther@suse.de>
34997 PR tree-optimization/79666
34998 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
34999 to not symbolically negate if that may introduce undefined
35002 2017-02-22 Martin Liska <mliska@suse.cz>
35005 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
35006 * data-streamer-out.c (streamer_write_gcov_count_stream):
35008 * value-prof.c (stream_out_histogram_value): Make assert more
35009 precise based on type of counter.
35011 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
35014 * config/i386/i386.md (standard_x87sse_constant_load splitter):
35015 Use nonimmediate_operand instead of memory_operand for operand 1.
35016 (float-extend standard_x87sse_constant_load splitter): Ditto.
35018 2017-02-21 Jeff Law <law@redhat.com>
35020 PR tree-optimization/79621
35021 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
35022 blocks with edges to themselves.
35024 2017-02-21 Jakub Jelinek <jakub@redhat.com>
35027 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
35028 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
35029 Use gimple_call_builtin_p.
35032 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
35033 on temporarily removed DEBUG_INSNs.
35035 PR tree-optimization/79649
35036 * tree-loop-distribution.c (classify_partition): Give up on
35037 non-generic address space loads/stores.
35039 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
35041 * doc/loop.texi (Loop manipulation): Remove nonexistent
35042 tree_ssa_loop_version from the documentation.
35043 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
35045 2017-02-21 Jakub Jelinek <jakub@redhat.com>
35048 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
35049 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
35050 * config/rs6000/rs6000.c: Include except.h.
35051 (rs6000_expand_split_stack_prologue): Call
35052 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
35054 2017-02-21 Martin Jambor <mjambor@suse.cz>
35057 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
35058 have been analyzed.
35060 2017-02-21 Martin Jambor <mjambor@suse.cz>
35062 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
35063 for backward compatibility only.
35064 * doc/invoke.texi (Option Summary): Remove all references to
35065 -fipa-cp-alignment.
35067 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
35071 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35073 * lra-constraints.c (curr_insn_transform): Handle
35074 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
35076 2017-02-21 Martin Liska <mliska@suse.cz>
35078 * config/i386/i386.opt: Replace -masm-dialect with -masm.
35080 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
35082 PR translation/79638
35083 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
35085 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
35088 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
35089 (arm_function_ok_for_sibcall): Return false for an indirect call by
35090 descriptor if all the argument registers are used.
35091 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
35092 alignment of the function.
35094 2017-02-21 Jakub Jelinek <jakub@redhat.com>
35096 PR tree-optimization/61441
35097 * simplify-rtx.c (simplify_const_unary_operation): For
35098 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
35099 the sNaN unmodified.
35101 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
35103 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
35104 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
35105 instead of SYSTEM_HEADER_DIR.
35107 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
35108 Martin Liška <mliska@suse.cz>
35110 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
35111 Fix typos and grammar, use active voice, and clarify.
35113 2017-02-20 Marek Polacek <polacek@redhat.com>
35115 PR middle-end/79537
35116 * gimplify.c (gimplify_expr): Handle unused *&&L;.
35119 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
35121 2017-02-20 Jakub Jelinek <jakub@redhat.com>
35124 * config/i386/i386.c (ix86_expand_builtin): Handle
35125 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
35126 ix86_builtins_isa[fcode].isa as a requirement of those
35127 flags and any other flag in the bitmask.
35128 (ix86_init_mmx_sse_builtins): Use 0 instead of
35129 ~OPTION_MASK_ISA_64BIT as mask.
35130 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
35131 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
35132 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
35133 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
35135 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35138 * lra-constraints.c (split_reg): Check requested split mode
35139 is supported by the register.
35141 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35143 * lra-constraints.c (simplify_operand_subreg): Remove early
35146 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35149 * lra-constraints.c (curr_insn_transform): Tighten condition
35150 for converting SUBREG reloads from OP_OUT to OP_INOUT.
35152 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35155 * lra-constraints.c (curr_insn_transform): Handle
35156 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
35158 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
35161 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
35163 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
35165 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
35168 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
35171 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
35173 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
35176 2017-02-18 Jakub Jelinek <jakub@redhat.com>
35178 * final.c (last_columnnum, override_columnnum): New variables.
35179 (final_start_function): Set last_columnnum, pass it to begin_prologue
35180 hook and pass 0 to dwarf2out_begin_prologue.
35181 (final_scan_insn): Update override_columnnum. Pass last_columnnum
35182 to source_line debug hook.
35183 (notice_source_line): Compute last_columnnum and for debug_column_info
35184 return true on column changes.
35185 * debug.h (struct gcc_debug_hooks): Add column argument to
35186 source_line and begin_prologue hooks.
35187 (debug_nothing_int_charstar_int_bool): Remove prototype.
35188 (debug_nothing_int_int_charstar,
35189 debug_nothing_int_int_charstar_int_bool): New prototypes.
35190 (dwarf2out_begin_prologue): Add column argument.
35191 * debug.c (do_nothing_debug_hooks): Adjust source_line and
35192 begin_prologue hooks.
35193 (debug_nothing_int_charstar_int_bool): Remove.
35194 (debug_nothing_int_int_charstar,
35195 debug_nothing_int_int_charstar_int_bool): New functions.
35196 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
35197 through to dwarf2out_source_line.
35198 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
35199 (dwarf2out_source_line): Add column argument, emit it if requested.
35200 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
35202 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
35203 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
35204 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
35205 through to dwarf2out_begin_prologue.
35206 (vmsdbgout_source_line): Add column argument, pass it through to
35207 dwarf2out_source_line.
35208 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
35209 dbxout_source_line caller.
35210 (dbxout_source_line): Add column argument.
35212 * common.opt (gno-column-info, gcolumn-info): New options.
35213 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
35214 (check_die): Also test for multiple DW_AT_decl_column attributes.
35215 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
35216 DW_AT_decl_column if requested.
35217 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
35219 (gen_variable_die): Likewise.
35220 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
35221 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
35224 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
35225 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
35226 (ix86_handle_option): Handle OPT_m3dnowa.
35227 * doc/invoke.texi (-m3dnowa): Document.
35228 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
35229 -m3dnowa instead of -m3dnow -march=athlon.
35232 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
35233 instead of gcc_assert for K, r and R code checks. Formatting fixes.
35235 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35238 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
35239 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
35240 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
35241 generator for vsx_xxpermdi_<mode>_be.
35242 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
35243 force big-endian semantics.
35244 (vsx_xxpermdi_<mode>_be): New define_expand with same
35245 implementation as previous version of vsx_xxpermdi_<mode>.
35247 2017-02-17 Jakub Jelinek <jakub@redhat.com>
35249 PR tree-optimization/79327
35250 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
35251 variable, its initialization and use.
35253 2017-02-17 Julia Koval <julia.koval@intel.com>
35255 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
35256 (OPTION_MASK_ISA_PKU_UNSET): New.
35257 (ix86_handle_option): Handle -mrdpid.
35258 * config/i386/cpuid.h (bit_RDPID): New.
35259 * config/i386/driver-i386.c (host_detect_local_cpu):
35260 Detect RDPID feature.
35261 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
35262 * config/i386/i386-c.c (ix86_target_macros_internal):
35264 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
35265 (ix86_valid_target_attribute_inner_p): Add "rdpid".
35266 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
35267 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
35268 * config/i386/i386.md (define_insn "rdpid"): New.
35269 * config/i386/i386.opt Add -mrdpid.
35270 * config/i386/immintrin.h (_rdpid_u32): New.
35272 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
35274 PR rtl-optimization/79541
35275 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
35276 instead of transforming it into USE.
35278 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
35280 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
35281 If HONOR_SNANS (SFmode) force the input to a register.
35282 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
35283 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
35284 an frsp or similar insn.
35286 2017-02-17 Martin Liska <mliska@suse.cz>
35288 PR rtl-optimization/79577
35289 * params.def (selsched-max-sched-times): Increase minimum to 1.
35291 2017-02-17 Martin Liska <mliska@suse.cz>
35293 PR rtl-optimization/79574
35294 * gcse.c (want_to_gcse_p): Prevent integer overflow.
35296 2017-02-17 Martin Liska <mliska@suse.cz>
35298 PR tree-optimization/79529
35299 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
35300 ssa_defined_default_def_p to handle cases which are implicitly
35302 * tree-ssa.c (ssa_defined_default_def_p): New function.
35303 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
35304 which are implicitly defined.
35305 * tree-ssa.h (ssa_defined_default_def_p): Declare.
35307 2017-02-17 Richard Biener <rguenther@suse.de>
35309 PR middle-end/79576
35310 * params.def (max-ssa-name-query-depth): Limit to 10.
35312 2017-02-17 Richard Biener <rguenther@suse.de>
35314 PR tree-optimization/79552
35315 * tree-ssa-structalias.c (visit_loadstore): Properly verify
35318 2017-02-17 Richard Biener <rguenther@suse.de>
35321 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
35323 2017-02-17 Marek Polacek <polacek@redhat.com>
35325 PR middle-end/79536
35326 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
35327 (fold_negate_expr): New wrapper.
35329 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
35331 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
35332 Correct terminology and de-emphasize pre-standard behavior.
35334 2017-02-16 Alan Modra <amodra@gmail.com>
35336 PR rtl-optimization/79286
35337 * ira.c (def_dominates_uses): New function.
35338 (update_equiv_regs): Don't create an equivalence for insns that
35339 may trap where the register def does not dominate the use.
35341 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
35343 PR rtl-optimization/78127
35344 * lra.c (lra): Call lra_eliminate before finish the loop after
35347 2017-02-16 Richard Biener <rguenther@suse.de>
35349 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
35351 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
35352 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
35353 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
35354 (isl_val_int_from_wi): New function.
35355 (extract_affine_gmp): Rename to ...
35356 (extract_affine_wi): ... this, take a widest_int.
35357 (extract_affine_int): Just wrap extract_affine_wi.
35358 (add_param_constraints): Use isl_val_int_from_wi.
35359 (add_loop_constraints): Likewise, and extract_affine_wi.
35361 2017-02-15 Jeff Law <law@redhat.com>
35363 PR middle-end/79521
35364 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
35365 ira_init_register_move_cost_if_necessary.
35367 2017-02-15 Martin Sebor <msebor@redhat.com>
35369 PR middle-end/32003
35370 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
35371 removed in a prior commit.
35373 2017-02-15 Bin Cheng <bin.cheng@arm.com>
35375 PR tree-optimization/79347
35376 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
35377 counters during peeling.
35379 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
35381 * Makefile.in (site.exp): Remove "set ISLVER".
35383 2017-02-15 Jakub Jelinek <jakub@redhat.com>
35386 * real.c (real_from_integer): Call real_convert even for decimal.
35388 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
35391 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
35393 2017-02-14 Andrew Pinski <apinski@cavium.com>
35395 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
35396 cores and change the partno/implementer to be correct.
35397 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
35398 the 'B" as the implementer.
35399 * config/aarch64/aarch64-tune.md: Regenerate.
35401 2017-02-14 Carl Love <cel@us.ibm.com>
35403 * config/rs6000/rs6000.c: Add case statement entry to make the
35404 xvcvuxdsp built-in argument unsigned.
35405 * config/rs6000/vsx.md: Fix the source and return operand types so they
35406 match the instruction definitions from the ISA document. Fix typo
35407 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
35410 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
35413 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
35414 member early_clobber_alts.
35415 * lra-lives.c (reg_early_clobber_p): New.
35416 (process_bb_lives): Use it.
35417 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
35418 (debug_operand_data): Initialize early_clobber_alts.
35419 (setup_operand_alternative): Set up early_clobber_alts.
35420 (collect_non_operand_hard_regs): Ditto. Pass early clobber
35421 alternatives to new_insn_reg.
35422 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
35424 (lra_update_insn_regno_info): Pass the new arg.
35426 2017-02-14 Jakub Jelinek <jakub@redhat.com>
35428 PR middle-end/79505
35429 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
35430 (new_oacc_loop_raw): Don't clear already cleared fields.
35433 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
35434 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
35435 _mm512_prefetch_i64gather_ps): New inline functions and macros.
35437 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
35440 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
35442 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
35445 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
35446 the extra instruction to the right place to store 128-bit constant
35449 2017-02-14 Martin Sebor <msebor@redhat.com>
35451 PR middle-end/79448
35452 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
35453 warning for strings of unknown length.
35455 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
35457 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
35459 2017-02-14 Jeff Law <law@redhat.com>
35462 * ira-costs.c (scan_one_insn): Initialize register move costs
35463 for pseudos seen in USE/CLOBBER insns.
35465 PR tree-optimization/79095
35466 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
35467 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
35468 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
35469 if the operands are known to be not equal, then the resulting range
35471 (intersect_ranges): If the new range is ~[0,0] and the old range is
35472 wide, then prefer ~[0,0].
35473 * tree-vrp.c (overflow_comparison_p_1): New function.
35474 (overflow_comparison_p): New function.
35475 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
35476 if NAME is used in an overflow test.
35477 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
35478 overflow check that can be expressed as an equality test, then adjust
35479 ops to be that equality test.
35481 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35483 * config/s390/s390-builtin-types.def: Remove flags argument.
35484 * config/s390/s390.c (s390_init_builtins): Likewise.
35486 2017-02-14 Martin Liska <mliska@suse.cz>
35488 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
35489 vector. Fix trailing white spaces.
35491 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
35493 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
35496 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35498 PR rtl-optimization/68664
35499 * config/arm/arm.c (arm_sched_can_speculate_insn):
35500 New function. Declare prototype.
35501 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
35503 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35505 PR rtl-optimization/68664
35506 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
35508 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
35510 2017-02-14 Amit Pawar <amit.pawar@amd.com>
35512 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
35513 max skip bytes for function, loop and jump.
35515 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
35517 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
35518 ABS_EXPR for gimple dump.
35520 2017-02-14 Jakub Jelinek <jakub@redhat.com>
35523 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
35525 PR tree-optimization/79408
35526 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
35527 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
35528 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
35529 also if rhs1 is INTEGER_CST.
35531 2017-02-14 Richard Biener <rguenther@suse.de>
35533 PR middle-end/79432
35534 * tree-into-ssa.c (insert_phi_nodes): When the function can
35535 have abnormal edges rewrite SSA names with broken use-def
35536 dominance out of SSA and register them for PHI insertion.
35538 2017-02-13 Martin Sebor <msebor@redhat.com>
35540 PR middle-end/79496
35541 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
35542 clearing info.nowrite flag when snprintf size argument is a range.
35544 2017-02-13 Jakub Jelinek <jakub@redhat.com>
35546 * cprop.c (cprop_jump): Add missing space in string literal.
35547 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
35548 (get_constraint_for_component_ref): Likewise.
35549 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
35550 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
35551 * lra-constraints.c (process_alt_operands): Likewise.
35552 * ipa-inline.c (inline_small_functions): Likewise.
35553 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
35554 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
35555 * trans-mem.c (diagnose_tm_1_op): Likewise.
35556 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
35557 (grid_parallel_clauses_gridifiable): Likewise.
35559 * config/nvptx/mkoffload.c (process): Add space in between
35562 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
35563 "MOD4_SSE_REGS" and "ALL_REGS".
35565 * spellcheck.c (test_data): Add , in between "foo" and "food".
35567 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35570 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
35571 boundary crossing check and subsequent code generation agree.
35573 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35575 * config/aarch64/aarch64.c (has_memory_op): Delete.
35576 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
35579 2017-02-13 Jakub Jelinek <jakub@redhat.com>
35581 PR rtl-optimization/79388
35582 PR rtl-optimization/79450
35583 * combine.c (distribute_notes): When removing TEM_INSN for which
35584 corresponding dest has last value recorded, invalidate that last
35587 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35589 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
35590 of explicit '@'. Add missing assembly comment marker on branch costs
35593 2017-02-13 Nathan Sidwell <nathan@acm.org>
35595 * gengtype-lex.l (<in_struct>): Add '/'.
35597 2017-02-13 Martin Liska <mliska@suse.cz>
35600 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
35602 2017-02-13 Richard Biener <rguenther@suse.de>
35604 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
35606 * configure: Re-generate.
35607 * config.in: Likewise.
35608 * graphite-dependences.c: Simplify as if
35609 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
35610 * graphite-isl-ast-to-gimple.c: Likewise.
35611 * graphite-optimize-isl.c: Likewise.
35612 * graphite-poly.c: Likewise.
35613 * graphite-sese-to-poly.c: Likewise.
35614 * graphite.h: Likewise.
35615 * toplev.c: Include isl/version.h and use isl_version () for
35616 printing the ISL version.
35617 * doc/install.texi: Update ISL requirement.
35619 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
35621 * doc/standards.texi (Standards): Update reference to
35624 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
35626 * doc/extend.texi (Named Address Spaces): sourceware.org now
35628 * doc/install.texi (Binaries): Ditto.
35631 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
35633 * doc/cpp.texi: Replace "stringify"/"stringification" with C
35634 standard terminology "stringize"/"stringizing" throughout.
35635 * doc/cppinternals.texi: Likewise.
35637 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
35639 * doc/extend.texi: Fix some spelling mistakes and typos.
35640 * doc/invoke.texi: Likewise.
35642 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
35645 * params.def (inline-min-speedup) Change from 10 to 8.
35647 2017-02-11 Jakub Jelinek <jakub@redhat.com>
35649 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
35652 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
35655 * ipa-inline-analysis.c (get_minimal_bb): New function.
35656 (record_modified): Use it.
35657 (remap_edge_change_prob): Handle also ancestor functions.
35659 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
35661 * doc/contrib.texi (Contributors): Remove broken link into
35662 the Mauve CVS repository.
35664 2017-02-11 Jakub Jelinek <jakub@redhat.com>
35666 PR middle-end/79454
35667 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
35668 result computation whenever lhs doesn't have vector mode, not
35669 just when it has BLKmode.
35671 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35673 * doc/makefile.texi (profiledbootstrap): Refer to the
35674 installation instructions only in textual form.
35676 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35679 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
35681 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35683 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
35684 (Specific): Update mingw-w64 reference.
35686 (Specific): Remove broken link to Renesas RX processor.
35688 2017-02-10 Richard Biener <rguenther@suse.de>
35690 * toplev.c (process_options): Do not mention obsolete graphite
35691 options when printing sorry message about missing graphite support.
35692 Mention -floop-nest-optimize.
35694 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
35696 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
35697 (vtst_p16): Likewise.
35698 (vtstq_p8): Likewise.
35699 (vtstq_p16): Likewise.
35701 (vtstq_p64): Likewise.
35702 * config/arm/arm_neon.h (vgetq_lane_p64): New.
35703 (vset_lane_p64): New.
35704 (vsetq_lane_p64): New.
35706 2017-02-10 Jakub Jelinek <jakub@redhat.com>
35708 PR tree-optimization/79411
35709 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
35710 stmt operands are SSA_NAMEs used in abnormal phis.
35711 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
35714 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35717 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
35720 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35722 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
35724 2017-02-09 Jakub Jelinek <jakub@redhat.com>
35726 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
35730 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
35731 not arbitrary TREE_CONSTANT.
35734 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
35735 "omp declare target link" attribute unless is_global_var.
35736 * omp-offload.c (find_link_var_op): Likewise.
35738 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
35739 Chung-Lin Tang <cltang@codesourcery.com>
35741 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
35743 (gimplify_adjust_omp_clauses): Don't delete TILE.
35744 (gimplify_omp_for): Deal with TILE.
35745 * internal-fn.c (expand_GOACC_TILE): New function.
35746 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
35748 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
35749 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
35751 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
35752 avoid DIV for outermost collapse var.
35753 (expand_oacc_for): Insert tile element loop as needed. Adjust.
35754 Remove out of date comments, fix whitespace.
35755 * omp-general.c (omp_extract_for_data): Deal with tiling.
35756 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
35757 adjust OLF_DIM_BASE value.
35758 (struct omp_for_data): Add tiling field.
35759 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
35760 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
35761 for auto loops. Remove default auto determining, moved to
35762 oacc_loop_fixed_partitions.
35763 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
35764 stmts, add e_mask field.
35765 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
35766 (oacc_thread_numbers): Use oacc_dim_call.
35767 (oacc_xform_tile): New.
35768 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
35769 (finish_oacc_loop): Adjust for ifns vector.
35770 (oacc_loop_discover_walk): Append loop abstraction sites to list,
35771 add case for GOACC_TILE fns.
35772 (oacc_loop_xform_loop): Delete.
35773 (oacc_loop_process): Iterate over call list directly, and add
35774 handling for GOACC_TILE fns.
35775 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
35777 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
35778 vector partitioning to outer loops. Assign 2 partitions to loops
35779 when available. Add TILE handling.
35780 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
35781 (execite_oacc_device_lower): Process GOACC_TILE fns,
35782 ignore unknown specs.
35783 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
35784 * tree.c (omp_clause_num_ops): Adjust TILE ops.
35785 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
35787 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
35789 * configure.ac (ACX_BUGURL): Update.
35790 * configure: Regenerate.
35792 2017-02-09 Richard Biener <rguenther@suse.de>
35794 PR tree-optimization/69823
35795 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
35796 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
35798 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35800 * config/arc/arc-c.def: Add __NPS400__ definition.
35801 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
35802 (TARGET_NPS400): Define.
35804 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35806 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
35808 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
35809 pointer, arch_info.
35810 (arc_cpu_types): Fill the arch_info field with a pointer into the
35811 arc_arch_types table.
35812 (arc_selected_cpu): Declare.
35813 * config/arc/arc.c (arc_selected_cpu): Make global.
35814 (arc_selected_arch): Delete.
35815 (arc_base_cpu): Delete.
35816 (arc_override_options): Remove references to deleted variables,
35817 update access to arch information.
35818 (ARC_OPT): Update access to arch information.
35819 (ARC_OPTX): Likewise.
35820 * config/arc/arc.h (arc_base_cpu): Remove declaration.
35821 (TARGET_ARC600): Update access to arch information.
35822 (TARGET_ARC601): Likewise.
35823 (TARGET_ARC700): Likewise.
35824 (TARGET_EM): Likewise.
35825 (TARGET_HS): Likewise.
35826 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
35829 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
35832 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
35833 condition/operands for integer GE/LE/GEU/LEU operations.
35835 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
35837 PR translation/79397
35838 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
35841 2017-02-08 Martin Jambor <mjambor@suse.cz>
35844 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
35845 whether allocation happened.
35846 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
35847 nothing was allocated.
35849 2017-02-08 Jakub Jelinek <jakub@redhat.com>
35851 PR tree-optimization/79408
35852 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
35853 constant, but SSA_NAME with a known integer range, use the minimum
35854 of that range instead of op1 to determine if modulo can be replaced
35855 with its first operand.
35857 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35859 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
35861 2017-02-08 Richard Biener <rguenther@suse.de>
35863 PR tree-optimization/71824
35864 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35865 Check all loops contained in the merged region.
35867 2017-02-07 Andrew Pinski <apinski@cavium.com>
35869 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
35871 2017-02-07 Andrew Pinski <apinski@cavium.com>
35873 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
35874 (thunderxt88): Likewise.
35875 (thunderxt81): Disable LSE and change v8.1 to v8.
35876 (thunderxt83): Likewise.
35878 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35879 Richard Biener <rguenther@suse.de>
35881 PR middle-end/79399
35882 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
35883 type from int to size_t.
35884 * ira-costs.c (struct_costs_size): Change type from int to size_t.
35886 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35888 PR rtl-optimization/79386
35889 * cprop.c (bypass_conditional_jumps): Initialize
35890 bypass_last_basic_block already before splitting bbs after
35891 unconditional traps...
35892 (bypass_conditional_jumps): ... rather than here.
35895 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
35896 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
35897 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
35898 fix -masm=intel patterns.
35900 2017-02-07 Richard Biener <rguenther@suse.de>
35902 PR tree-optimization/79256
35903 PR middle-end/79278
35904 * builtins.c (get_object_alignment_2): Use min_align_of_type
35905 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
35906 and ADJUST_FIELD_ALIGN.
35908 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
35910 * doc/tm.texi: Regenerate.
35911 * stor-layout.c (layout_decl): Adjust.
35912 (update_alignment_for_field): Likewise.
35913 (place_field): Likewise.
35914 (min_align_of_type): Likewise.
35915 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
35916 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
35917 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
35918 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
35919 * config/frv/frv.c (frv_adjust_field_align): Likewise.
35920 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
35921 * config/i386/i386.c (x86_field_alignment): Likewise.
35922 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
35923 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
35924 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
35925 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
35926 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
35927 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
35931 2017-01-30 Richard Biener <rguenther@suse.de>
35933 PR tree-optimization/79256
35934 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35935 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35938 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
35940 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
35941 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
35942 builtins to SImode and emit a zero-extend, if necessary.
35944 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35946 * docs/invoke.texi (RISC-V Options): Alphabetize.
35948 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35950 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
35953 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35955 * config/riscv/riscv.c: New file.
35956 * common/config/riscv/riscv-common.c: Likewise.
35957 * config.gcc: Likewise.
35958 * config/riscv/constraints.md: Likewise.
35959 * config/riscv/elf.h: Likewise.
35960 * config/riscv/generic.md: Likewise.
35961 * config/riscv/linux.h: Likewise.
35962 * config/riscv/multilib-generator: Likewise.
35963 * config/riscv/peephole.md: Likewise.
35964 * config/riscv/pic.md: Likewise.
35965 * config/riscv/predicates.md: Likewise.
35966 * config/riscv/riscv-builtins.c: Likewise.
35967 * config/riscv/riscv-c.c: Likewise.
35968 * config/riscv/riscv-ftypes.def: Likewise.
35969 * config/riscv/riscv-modes.def: Likewise.
35970 * config/riscv/riscv-opts.h: Likewise.
35971 * config/riscv/riscv-protos.h: Likewise.
35972 * config/riscv/riscv.h: Likewise.
35973 * config/riscv/riscv.md: Likewise.
35974 * config/riscv/riscv.opt: Likewise.
35975 * config/riscv/sync.md: Likewise.
35976 * config/riscv/t-elf-multilib: Likewise.
35977 * config/riscv/t-linux: Likewise.
35978 * config/riscv/t-linux-multilib: Likewise.
35979 * config/riscv/t-riscv: Likewise.
35980 * configure.ac: Likewise.
35981 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
35982 Waterman as RISC-V maintainers.
35983 * doc/install.texi: Add RISC-V entries.
35984 * doc/invoke.texi: Add RISC-V options section.
35985 * doc/md.texi: Add RISC-V constraints section.
35986 * configure: Regenerated.
35988 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
35991 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
35992 false values to be constant vectors with all 0 or all 1 bits set.
35993 (vcondu<mode><mode>): Likewise.
35994 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
35996 (fpmask_comparison_operator): Update comment.
35997 (vecint_comparison_operator): New predicate.
35998 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
35999 vector conditionals when the true and false values are constant
36000 vectors with all 0 bits or all 1 bits set.
36002 2017-02-06 Martin Sebor <msebor@redhat.com>
36004 PR tree-optimization/79376
36005 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
36007 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
36009 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
36010 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
36011 to simplify split condition.
36013 2017-02-06 Jakub Jelinek <jakub@redhat.com>
36015 * omp-expand.c (oxpand_omp_atomic_fetch_op,
36016 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
36019 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
36021 PR rtl-optimization/68664
36022 * target.def (can_speculate_insn): New hook.
36023 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
36024 * doc/tm.texi: Regenerate.
36025 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
36026 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
36027 (rs6000_sched_can_speculate_insn): New function.
36029 2017-02-06 Jakub Jelinek <jakub@redhat.com>
36031 PR tree-optimization/79284
36032 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
36033 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
36034 vectorizable_mask_load_store, vectorizable_operation,
36035 vect_is_simple_cond, get_same_sized_vectype): Use it instead
36036 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
36037 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
36038 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
36039 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
36040 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
36041 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
36042 is_gimple_assign (stmt). Replace another such test with
36043 is_gimple_assign (stmt).
36045 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
36048 * config/avr/avr.c (rtl-iter.h): Include it.
36049 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
36050 (avr_legitimate_combined_insn): ...and implementation.
36052 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
36054 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
36055 * config/s390/s390.c (s390_const_operand_ok)
36056 (s390_canonicalize_comparison, s390_extract_part)
36057 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
36058 (s390_contiguous_bitmask_p, s390_rtx_costs)
36059 (legitimize_pic_address): Likewise.
36060 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
36061 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
36062 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
36063 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
36064 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
36066 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
36068 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
36069 REGNO($0) == REGNO($1).
36071 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36073 * config/s390/linux.h(SIZE_TYPE): Add comment.
36075 2017-02-06 Julian Brown <julian@codesourcery.com>
36076 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
36077 Virendra Pathak <virendra.pathak@broadcom.com>
36079 * config/aarch64/aarch64-cores.def: Change the scheduler
36081 * config/aarch64/aarch64.md: Include thunderx2t99.md.
36082 * config/aarch64/thunderx2t99.md: New file.
36084 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36086 * doc/standards.texi (Go Language): Update link to language
36089 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
36091 * tree-eh.c (lower_resx): Sanitize profile.
36092 (cleanup_empty_eh_move_lp): Likewise.
36094 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
36097 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
36099 * cfgloopmanip.h (loop_version): Update prototype.
36100 * modulo-sched.c (sms_schedule): Update call of loop_version.
36101 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
36102 * tree-parloops.c (gen_parallel_loop): Likewise.
36103 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
36104 * tree-ssa-loop-split.c (split_loop): Likewise.
36105 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
36106 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
36108 2017-02-05 Martin Liska <mliska@suse.cz>
36111 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
36113 (print_operand_address): Initialize a struct to zero.
36115 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36117 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
36118 garbage collector only in textual form.
36120 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36122 * doc/extend.texi (x86 specific memory model extensions for
36123 transactional memory): Simplify a phrase.
36125 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
36128 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
36129 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
36130 (atomic_storedi_1): Likewise.
36132 2017-02-04 Jakub Jelinek <jakub@redhat.com>
36134 PR tree-optimization/79338
36135 * tree-parloops.c (gather_scalar_reductions): Don't call
36136 vect_analyze_loop_form for loop->inner before destroying loop's
36139 2017-02-03 Martin Sebor <msebor@redhat.com>
36141 PR tree-optimization/79327
36142 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
36143 when precision has resulted in leading zeros.
36144 (format_integer): Adjust the likely counter to assume an unknown
36145 argument that may be zero is non-zero.
36147 2017-02-03 Jason Merrill <jason@redhat.com>
36150 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
36151 avoid copying non-taken branch.
36153 2017-02-03 Jakub Jelinek <jakub@redhat.com>
36155 PR tree-optimization/79340
36156 * tree-vect-loop.c (vectorizable_reduction): Release
36157 vec_defs elements after safe_splicing them into other vectors.
36160 PR tree-optimization/79327
36161 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
36162 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
36164 (format_integer): Use wide_int_to_tree instead of build_int_cst
36165 + to_?hwi. If argmin is NULL, just set argmin and argmax to
36166 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
36167 of shortest and longest sequence.
36169 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
36171 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
36172 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
36174 2017-02-03 Walter Lee <walt@tilera.com>
36177 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
36178 after initial stackframe link reg save.
36179 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
36181 2017-02-03 Jakub Jelinek <jakub@redhat.com>
36184 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
36185 wu for stxssp alternative.
36187 2017-02-03 Martin Sebor <msebor@redhat.com>
36189 PR tree-optimization/79352
36190 * gimple-fold.c (get_range_strlen): Add argument.
36191 (get_range_strlen): Change return type to bool.
36192 (get_maxval_strlen): Pass in a dummy argument.
36193 * gimple-fold.h (get_range_strlen): Change return type to bool.
36194 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
36195 * tree.h (array_at_struct_end_p): Add argument.
36196 * tree.c (array_at_struct_end_p): Handle it.
36198 2017-02-03 Martin Liska <mliska@suse.cz>
36201 * multiple_target.c (create_dispatcher_calls): Redirect edge
36202 from a caller of a dispatcher.
36203 (expand_target_clones): Make the clones local.
36204 (ipa_target_clone): Do both target clones and resolvers.
36205 (ipa_dispatcher_calls): Remove the pass.
36206 (pass_dispatcher_calls::gate): Likewise.
36207 (make_pass_dispatcher_calls): Likewise.
36208 * passes.def (pass_target_clone): Put as very first IPA early
36211 2017-02-03 Martin Liska <mliska@suse.cz>
36213 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
36214 in case of a function with ifunc attribute.
36216 2017-02-03 Martin Liska <mliska@suse.cz>
36218 * cgraph.c (cgraph_node::dump): Dump function version info.
36219 * symtab.c (symtab_node::dump_base): Add missing new line.
36221 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
36223 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
36224 (ifcombine_ifandif): Use it.
36226 2017-02-03 Martin Liska <mliska@suse.cz>
36228 * doc/invoke.texi: Document default value for
36229 use-after-scope-direct-emission-threshold.
36231 2017-02-03 Martin Liska <mliska@suse.cz>
36233 PR tree-optimization/79339
36234 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
36235 (format_floating): Likewise.
36237 2017-02-03 Martin Liska <mliska@suse.cz>
36240 * ipa-prop.c (ipa_node_params_t::insert): Remove current
36242 (ipa_node_params_t::remove): Likewise.
36243 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
36244 initialization from removed ipa_node_params_t::insert.
36245 (ipa_node_params::~ipa_node_params): Move from removed
36246 ipa_node_params_t::release.
36247 * symbol-summary.h (symbol_summary::m_released): New member.
36248 Do not release a summary twice. Do not allow to call finalizer
36249 for types of a summary that live in GGC memory.
36251 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
36253 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
36256 2017-02-02 Martin Sebor <msebor@redhat.com>
36258 PR middle-end/79275
36259 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
36260 (format_string): Tighten up the range of output for non-constant
36261 strings and correct the expected range for wide non-constant strings.
36263 2017-02-02 Martin Sebor <msebor@redhat.com>
36265 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
36267 PR middle-end/32003
36268 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
36270 (-fdump-tree-@var): Add to index and document how to come up
36271 with pass-specific option and dump file names.
36272 (-fdump-passes): Clarify where to look for output.
36274 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
36276 PR middle-end/77445
36277 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
36278 statistics of the analyzed path; allow threading for speed when
36279 any of BBs along the path are optimized for speed.
36281 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
36283 PR middle-end/78468
36284 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
36285 settings of the virtual registers.
36288 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
36290 * explow.c (get_dynamic_stack_size): Take known alignment of stack
36291 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
36294 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36296 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
36297 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
36299 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36301 * config/s390/s390.md: Add missing comments with the expanded
36303 * config/s390/vector.md: Likewise.
36304 * config/s390/vx-builtins.md: Likewise.
36306 2017-02-02 Jakub Jelinek <jakub@redhat.com>
36309 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
36310 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
36311 conditions on a single line.
36313 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36315 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
36316 __S390_VX__ to __VX__.
36318 2017-02-01 Andrew Pinski <apinski@cavium.com>
36320 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
36321 stmt_info to record_stmt_cost.
36322 (vect_get_known_peeling_cost): Pass stmt_info if known to
36324 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
36325 cpu_vector_cost field into
36326 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
36327 field into vec_int_stmt_cost and vec_fp_stmt_cost.
36328 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
36329 splitting of scalar_stmt_cost and vec_stmt_cost.
36330 (thunderx_vector_cost): Likewise.
36331 (cortexa57_vector_cost): LIkewise.
36332 (exynosm1_vector_cost): Likewise.
36333 (xgene1_vector_cost): Likewise.
36334 (thunderx2t99_vector_cost): Improve after the splitting of the two
36336 (aarch64_builtin_vectorization_cost): Update for the splitting of
36337 scalar_stmt_cost and vec_stmt_cost.
36339 2017-02-01 Torvald Riegel <triegel@redhat.com>
36340 Richard Henderson <rth@redhat.com>
36342 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
36343 conditional on existance of a fast atomic load.
36344 * optabs-query.c (can_atomic_load_p): New function.
36345 * optabs-query.h (can_atomic_load_p): Declare it.
36346 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
36347 no fast atomic load is available for the particular size of access.
36348 (expand_atomic_compare_and_swap): Likewise.
36349 (expand_atomic_load): Likewise.
36350 (expand_atomic_store): Likewise.
36351 (expand_atomic_fetch_op): Likewise.
36352 * testsuite/lib/target-supports.exp
36353 (check_effective_target_sync_int_128): Remove x86 because it provides
36354 no fast atomic load.
36355 (check_effective_target_sync_int_128_runtime): Likewise.
36357 2017-02-01 Richard Biener <rguenther@suse.de>
36359 * graphite.c: Include tree-vectorizer.h for find_loop_location.
36360 (graphite_transform_loops): Provide opt-info for optimized nests.
36361 * tree-parloop.c (parallelize_loops): Provide opt-info for
36362 parallelized loops.
36364 2017-02-01 Richard Biener <rguenther@suse.de>
36366 PR middle-end/79315
36367 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
36368 was not set before.
36370 2017-02-01 Richard Biener <rguenther@suse.de>
36372 PR tree-optimization/71824
36373 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
36374 Verify the loops are valid in the merged SESE region.
36375 (scop_detection::can_represent_loop_1): Check analyzing the
36376 evolution of the number of iterations in the region succeeds.
36378 2017-01-31 Ian Lance Taylor <iant@golang.org>
36380 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
36381 REG_ARGS_SIZE note to 32-bit push insns and call insn.
36383 2017-01-31 David Malcolm <dmalcolm@redhat.com>
36385 PR preprocessor/79210
36386 * input.c (get_substring_ranges_for_loc): Replace line_width
36387 assertion with error-handling.
36389 2017-01-31 Richard Biener <rguenther@suse.de>
36391 PR tree-optimization/77318
36392 * graphite-sese-to-poly.c (extract_affine): Fix assert.
36393 (create_pw_aff_from_tree): Take loop parameter.
36394 (add_condition_to_pbb): Pass loop of the condition to
36395 create_pw_aff_from_tree.
36397 2017-01-31 Jakub Jelinek <jakub@redhat.com>
36399 * config/s390/s390.c (s390_asan_shadow_offset): New function.
36400 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
36402 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
36406 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
36408 (convert_int_to_float128): Likewise.
36409 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
36410 (convert_int_to_float128): Likewise.
36411 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
36412 (UNSPEC_IEEE128_CONVERT): Likewise.
36413 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
36414 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
36415 Use local variables for IBM extended format.
36416 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
36417 (fix_trunc<mode>si2_fprs): Likewise.
36418 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
36419 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
36420 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
36421 to know that we can now have integers of all sizes in vector
36423 (fix<uns>_<mode>di2_hw): Likewise.
36424 (float<uns>_<mode>si2_hw): Likewise.
36425 (fix_<mode>si2_hw): Likewise.
36426 (fixuns_<mode>si2_hw): Likewise.
36427 (float<uns>_<mode>di2_hw): Likewise.
36428 (float_<mode>di2_hw): Likewise.
36429 (float_<mode>si2_hw): Likewise.
36430 (floatuns_<mode>di2_hw): Likewise.
36431 (floatuns_<mode>si2_hw): Likewise.
36432 (xscvqp<su>wz_<mode>): Delete, no longer used.
36433 (xscvqp<su>dz_<mode>): Likewise.
36434 (xscv<su>dqp_<mode>): Likewise.
36435 (ieee128_mfvsrd_64bit): Likewise.
36436 (ieee128_mfvsrd_32bit): Likewise.
36437 (ieee128_mfvsrwz): Likewise.
36438 (ieee128_mtvsrw): Likewise.
36439 (ieee128_mtvsrd_64bit): Likewise.
36440 (ieee128_mtvsrd_32bit): Likewise.
36442 2017-01-31 Martin Liska <mliska@suse.cz>
36445 * ipa-prop.c (ipa_free_all_node_params): Call release method
36446 instead of ~sumbol_summary to not to trigger double times
36449 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
36451 PR tree-optimization/71691
36452 * bitmap.h (class auto_bitmap): New.
36453 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
36454 is_maybe_undefined instead of ssa_undefined_value_p.
36456 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36458 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
36459 __S390_ARCH_LEVEL__ to __ARCH__.
36461 2017-01-31 Jakub Jelinek <jakub@redhat.com>
36463 PR tree-optimization/79267
36464 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
36465 if should_remove_lhs_p is true.
36467 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
36470 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
36471 (add_alignment_attribute): New.
36472 (base_type_die): Add alignment attribute.
36473 (subrange_type_die): Likewise.
36474 (modified_type_die): Likewise.
36475 (gen_array_type_die): Likewise.
36476 (gen_descr_array_type_die: Likewise.
36477 (gen_enumeration_type_die): Likewise.
36478 (gen_subprogram_die): Likewise.
36479 (gen_variable_die): Likewise.
36480 (gen_field_die): Likewise.
36481 (gen_ptr_to_mbr_type_die): Likewise.
36482 (gen_struct_or_union_type_die): Likewise.
36483 (gen_subroutine_type_die): Likewise.
36484 (gen_typedef_die): Likewise.
36485 (base_type_cmp): Compare alignment attribute.
36487 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36490 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
36491 (setb_unsigned) New pattern for setb with CCUNS.
36492 * config/rs6000/rs6000.c (expand_block_compare): Use a different
36493 subfc./subfe sequence to avoid overflow problems. Generate a
36494 shorter sequence with cmpld/setb for power9.
36495 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
36496 for generating subfc. instruction.
36497 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
36498 now uses this instruction.
36500 2017-01-30 Ian Lance Taylor <iant@google.com>
36503 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
36504 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
36506 2017-01-30 Martin Sebor <msebor@redhat.com>
36508 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
36509 Move constant to the right of a relational operator.
36510 (get_mpfr_format_length, format_character, format_string): Ditto.
36511 (should_warn_p, maybe_warn): Same.
36513 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
36515 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
36518 * asan.c (get_translation_unit_decl): Remove function.
36519 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
36521 2017-01-30 Martin Liska <mliska@suse.cz>
36523 PR gcov-profile/79259
36524 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
36525 -fprofile-generate.
36527 2017-01-30 Martin Liska <mliska@suse.cz>
36530 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
36531 Initialize variables with NULL value.
36533 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
36536 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
36538 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
36540 2017-01-30 Richard Biener <rguenther@suse.de>
36542 PR tree-optimization/79276
36543 * tree-vrp.c (process_assert_insertions): Properly adjust common
36544 when removing a duplicate.
36546 2017-01-30 Richard Biener <rguenther@suse.de>
36548 PR tree-optimization/79256
36549 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
36550 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
36552 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
36554 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
36557 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
36558 ("*r<noxa>sbg_<mode>_sll_bitmask")
36559 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
36560 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
36561 Use contiguous_bitmask_nowrap_operand.
36563 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36566 * config/rs6000/altivec.h (vec_xl): Revise #define.
36567 (vec_xst): Likewise.
36569 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
36571 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
36573 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
36575 PR rtl-optimization/79194
36576 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
36577 traps before call to bypass_conditional_jumps.
36579 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
36581 PR tree-optimization/71374
36582 * lra-constraints.c (check_conflict_input_operands): New.
36583 (match_reload): Use it.
36585 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
36588 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
36589 account to calculate conflict_set.
36591 2017-01-27 Bin Cheng <bin.cheng@arm.com>
36593 PR rtl-optimization/78559
36594 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
36595 other_insn in combine.
36597 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
36599 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
36600 uint16_type_node for BT_UINT16.
36602 2017-01-27 David Malcolm <dmalcolm@redhat.com>
36604 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
36605 "RTL Tests" to menu.
36606 (GIMPLE Tests): New node.
36607 (RTL Tests): New node.
36609 2017-01-27 Richard Biener <rguenther@suse.de>
36611 PR tree-optimization/79245
36612 * tree-loop-distribution.c (distribute_loop): Apply cost
36613 modeling also to detected patterns.
36615 2017-01-27 Richard Biener <rguenther@suse.de>
36617 PR tree-optimization/71433
36618 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
36619 (compare_assert_loc): New function.
36620 (process_assert_insertions): Sort and optimize assert locations
36621 to remove duplicates and push down identical assertions on
36622 edges to their destination block.
36624 2017-01-27 Richard Biener <rguenther@suse.de>
36626 PR tree-optimization/79244
36627 * tree-vrp.c (remove_range_assertions): Forcefully propagate
36628 out SSA names even if abnormal.
36630 2017-01-27 Jakub Jelinek <jakub@redhat.com>
36632 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
36633 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
36634 instead of MPFR_RNDN.
36636 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
36639 * arm.c (arm_option_override): Don't call build_target_option_node
36640 until after doing all option overrides.
36641 (arm_valid_target_attribute_tree): Likewise.
36643 2017-01-27 Martin Liska <mliska@suse.cz>
36645 * doc/invoke.texi (-fprofile-arcs): Document profiling support
36646 for {cd}tors and C++ {cd}tors.
36648 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
36650 * config/s390/s390.md ("*setmem_long_and")
36651 ("*setmem_long_and_31z"): Use zero_extend instead of and.
36653 2017-01-26 Martin Sebor <msebor@redhat.com>
36655 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
36658 2017-01-26 Martin Sebor <msebor@redhat.com>
36660 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
36661 HAVE_DFmode before using XFmode or DFmode.
36662 (parse_directive): Avoid using the z length modifier to avoid
36663 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
36665 PR middle-end/78703
36666 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
36667 to accept adjustment as an array.
36668 (get_int_range): New function.
36669 (struct directive): Make width and prec arrays.
36670 (directive::set_width, directive::set_precision): Call get_int_range.
36671 (format_integer, format_floating): Handle width and precision ranges.
36672 (format_string, parse_directive): Same.
36674 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36677 * dwarf2out.c (generate_skeleton_bottom_up): For children with
36678 comdat_type_p set, just clone them, but keep the children in the
36682 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
36683 which have direct callers with -fvar-tracking-assignments enabled
36685 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
36686 inside of type units.
36688 2017-01-26 Martin Sebor <msebor@redhat.com>
36690 PR middle-end/78703
36691 * gimple-ssa-sprintf.c (struct result_range): Add likely and
36693 (struct format_result): Replace number_chars, number_chars_min,
36694 and number_chars_max with a single member of struct result_range.
36696 (format_result::operator+=): Adjust.
36697 (struct fmtresult): Remove bounded. Handle likely and unlikely
36699 (fmtresult::adjust_for_width_or_precision): New function.
36700 (fmtresult:type_max_digits): New function.
36701 (bytes_remaining): Handle likely and unlikely counters.
36702 (min_bytes_remaining): Remove.
36703 (format_percent): Simplify.
36704 (format_integer, format_floating): Set likely and unlikely counters.
36705 (get_string_length, format_character, format_string): Same.
36706 (format_plain, should_warn_p): New function.
36707 (maybe_warn): Call should_warn_p. Update diagnostic messages
36708 and handle those for all directives, including plain strings.
36709 (format_directive): Handle likely and unlikely counters.
36710 Remove unnecessary quoting from diagnostics. Add an informational
36712 (add_bytes): Remove.
36713 (pass_sprintf_length::compute_format_length): Simplify.
36714 (try_substitute_return_value): Handle likely and unlikely counters.
36716 2017-01-26 Carl Love <cel@us.ibm.com>
36718 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
36719 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
36721 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
36724 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
36725 endianess for subregs into account.
36726 * lra-constraints.c (lra_constraints): Do risky transformations
36727 always on the first iteration.
36728 * lra-lives.c (check_pseudos_live_through_calls): Add arg
36729 last_call_used_reg_set.
36730 (process_bb_lives): Define and use last_call_used_reg_set.
36731 * lra.c (lra): Always continue after lra_constraints on the first
36734 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
36736 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
36738 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
36740 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36742 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
36743 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
36744 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
36745 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
36746 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
36747 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
36748 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
36749 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
36750 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
36752 2017-01-26 Marek Polacek <polacek@redhat.com>
36755 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
36756 for the third operand.
36758 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36760 PR middle-end/79236
36761 * omp-low.c (struct omp_context): Add simt_stmt field.
36762 (scan_omp_for): Return omp_context *.
36763 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
36764 context to the _simt_ SIMD stmt.
36765 (lower_omp_for): For combined SIMD with sibling _simt_
36766 SIMD, make sure to use the same decls in _looptemp_
36767 clauses as in the sibling.
36769 2017-01-26 David Sherwood <david.sherwood@arm.com>
36771 PR middle-end/79212
36772 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
36775 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36778 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
36779 emit fld b; fld a; if possible.
36781 * brig-builtins.def: Update copyright years.
36782 * config/arm/arm_acle_builtins.def: Update copyright years.
36784 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
36787 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
36788 constraint instead of o for the stxsd instruction.
36790 2017-01-25 Carl Love <cel@us.ibm.com>
36792 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
36793 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
36795 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
36797 * doc/invoke.texi (C++ Dialect Options): Fix typo.
36799 2017-01-25 Richard Biener <rguenther@suse.de>
36801 PR tree-optimization/69264
36802 * target.def (vector_alignment_reachable): Improve documentation.
36803 * doc/tm.texi: Regenerate.
36804 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
36806 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
36807 earlier changes with respect to TYPE_USER_ALIGN.
36808 (vector_alignment_reachable_p): Likewise. Improve dumping.
36810 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36813 * config/arm/arm.md (xordi3): Force constant operand into a register
36816 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36818 * doc/invoke.texi (-fstore-merging): Correct default optimization
36819 levels at which it is enabled.
36820 (-O): Move -fstore-merging from list to...
36823 2017-01-25 Richard Biener <rguenther@suse.de>
36826 * omp-expand.c: Include debug.h.
36827 (expand_omp_taskreg): Make sure to generate early debug before
36828 outlining anything from a function.
36829 (expand_omp_target): Likewise.
36830 (grid_expand_target_grid_body): Likewise.
36832 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
36835 * asan.c (get_translation_unit_decl): New function.
36836 (asan_add_global): Extract modules file name from globals
36837 TRANSLATION_UNIT_DECL name.
36839 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
36842 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
36843 for long calls with APCS frame and VFP.
36845 2017-01-24 David Malcolm <dmalcolm@redhat.com>
36847 * cfg.c (original_copy_tables_initialized_p): New function.
36848 * cfg.h (original_copy_tables_initialized_p): New decl.
36849 * cfgrtl.c (relink_block_chain): Guard the call to
36850 free_original_copy_tables with a call to
36851 original_copy_tables_initialized_p.
36852 * cgraph.h (symtab_node::native_rtl_p): New decl.
36853 * cgraphunit.c (symtab_node::native_rtl_p): New function.
36854 (symtab_node::needed_p): Don't assert for early assembly output
36855 for __RTL functions.
36856 (cgraph_node::finalize_function): Set "force_output" for __RTL
36858 (cgraph_node::analyze): Bail out early for __RTL functions.
36859 (analyze_functions): Update assertion to support __RTL functions.
36860 (cgraph_node::expand): Bail out early for __RTL functions.
36861 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
36863 * function.h (struct function): Update comment for field
36865 * gimple-expr.c: Include "tree-pass.h".
36866 (gimple_has_body_p): Return false for __RTL functions.
36867 * Makefile.in (OBJS): Add run-rtl-passes.o.
36868 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
36870 (gcc::pass_manager::get_clean_slate): New accessor.
36871 * passes.c: Include "insn-addr.h".
36872 (should_skip_pass_p): Add logging. Update logic for running
36873 "expand" to be compatible with both __GIMPLE and __RTL. Guard
36874 property-provider override so it is only done for gimple passes.
36876 (skip_pass): New function.
36877 (execute_one_pass): Call skip_pass when skipping passes.
36878 * read-md.c (md_reader::read_char): Support filtering
36879 the input to a subset of line numbers.
36880 (md_reader::md_reader): Initialize fields
36881 m_first_line and m_last_line.
36882 (md_reader::read_file_fragment): New function.
36883 * read-md.h (md_reader::read_file_fragment): New decl.
36884 (md_reader::m_first_line): New field.
36885 (md_reader::m_last_line): New field.
36886 * read-rtl-function.c (function_reader::create_function): Only
36887 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
36888 curr_properties. Set DECL_INITIAL to a dummy block.
36889 (read_rtl_function_body_from_file_range): New function.
36890 * read-rtl-function.h (read_rtl_function_body_from_file_range):
36892 * run-rtl-passes.c: New file.
36893 * run-rtl-passes.h: New file.
36895 2017-01-24 Jeff Law <law@redhat.com>
36897 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
36900 2017-01-24 Bin Cheng <bin.cheng@arm.com>
36902 PR tree-optimization/79159
36903 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
36904 (record_nonwrapping_iv): Improve boundary using above function if no
36905 value range information.
36907 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
36908 Martin Jambor <mjambor@suse.cz>
36910 * brig-builtins.def: New file.
36911 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
36912 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
36913 (DEF_HSAIL_SAT_BUILTIN): Likewise.
36914 (DEF_HSAIL_INTR_BUILTIN): Likewise.
36915 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
36916 * builtin-types.def (BT_INT8): New.
36917 (BT_INT16): Likewise.
36918 (BT_UINT8): Likewise.
36919 (BT_UINT16): Likewise.
36920 (BT_FN_ULONG): Likewise.
36921 (BT_FN_UINT_INT): Likewise.
36922 (BT_FN_UINT_ULONG): Likewise.
36923 (BT_FN_UINT_LONG): Likewise.
36924 (BT_FN_UINT_PTR): Likewise.
36925 (BT_FN_ULONG_PTR): Likewise.
36926 (BT_FN_INT8_FLOAT): Likewise.
36927 (BT_FN_INT16_FLOAT): Likewise.
36928 (BT_FN_UINT32_FLOAT): Likewise.
36929 (BT_FN_UINT16_FLOAT): Likewise.
36930 (BT_FN_UINT8_FLOAT): Likewise.
36931 (BT_FN_UINT64_FLOAT): Likewise.
36932 (BT_FN_UINT16_UINT32): Likewise.
36933 (BT_FN_UINT32_UINT16): Likewise.
36934 (BT_FN_UINT16_UINT16_UINT16): Likewise.
36935 (BT_FN_INT_PTR_INT): Likewise.
36936 (BT_FN_UINT_PTR_UINT): Likewise.
36937 (BT_FN_LONG_PTR_LONG): Likewise.
36938 (BT_FN_ULONG_PTR_ULONG): Likewise.
36939 (BT_FN_VOID_UINT64_UINT64): Likewise.
36940 (BT_FN_UINT8_UINT8_UINT8): Likewise.
36941 (BT_FN_INT8_INT8_INT8): Likewise.
36942 (BT_FN_INT16_INT16_INT16): Likewise.
36943 (BT_FN_INT_INT_INT): Likewise.
36944 (BT_FN_UINT_FLOAT_UINT): Likewise.
36945 (BT_FN_FLOAT_UINT_UINT): Likewise.
36946 (BT_FN_ULONG_UINT_UINT): Likewise.
36947 (BT_FN_ULONG_UINT_PTR): Likewise.
36948 (BT_FN_ULONG_ULONG_ULONG): Likewise.
36949 (BT_FN_UINT_UINT_UINT): Likewise.
36950 (BT_FN_VOID_UINT_PTR): Likewise.
36951 (BT_FN_UINT_UINT_PTR: Likewise.
36952 (BT_FN_UINT32_UINT64_PTR): Likewise.
36953 (BT_FN_INT_INT_UINT_UINT): Likewise.
36954 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
36955 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
36956 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
36957 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
36958 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
36959 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
36960 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
36961 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
36962 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
36963 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
36964 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
36965 * doc/frontends.texi: List BRIG FE.
36966 * doc/install.texi (Testing): Add BRIG tesring requirements.
36967 * doc/invoke.texi (Overall Options): Mention BRIG.
36968 * doc/standards.texi (Standards): Doucment BRIG HSA version.
36970 2017-01-24 Richard Biener <rguenther@suse.de>
36972 PR translation/79208
36973 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
36975 2017-01-24 Martin Jambor <mjambor@suse.cz>
36978 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
36979 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
36980 and known_contexts.
36982 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
36984 PR middle-end/79123
36985 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
36986 casts from signed to unsigned really don't have a range.
36988 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
36990 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
36991 GMP_RNDx for compatiblity.
36993 2017-01-24 Martin Liska <mliska@suse.cz>
36996 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
36997 that would prevent us to call alloca with -1 as argument.
36999 2017-01-24 Jakub Jelinek <jakub@redhat.com>
37001 * dwarf2out.c (output_compilation_unit_header, output_file_names):
37002 Avoid -Wformat-security warning.
37004 2017-01-23 Andrew Pinski <apinski@cavium.com>
37006 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
37009 2017-01-23 Martin Sebor <msebor@redhat.com>
37011 PR middle-end/78703
37012 * gimple-ssa-sprintf.c (warn_level): New global.
37013 (format_integer): Use it here and throughout the rest of the file.
37014 Use the same switch to compute sign as base.
37015 (maybe_warn): New function.
37016 (format_directive): Factor out warnings into maybe_warn.
37017 Add debugging output. Use warn_level.
37018 (add_bytes): Use warn_level.
37019 (pass_sprintf_length::compute_format_length): Add debugging output.
37020 (try_substitute_return_value): Same.
37021 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
37023 PR middle-end/78703
37024 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
37025 (struct fmtresult, format_integer, format_floating): Adjust.
37026 (fmtresult::fmtresult): Set max correctly in two argument ctor.
37027 (get_string_length, format_string,format_directive): Same.
37028 (pass_sprintf_length::compute_format_length): Same.
37029 (try_substitute_return_value): Simplify slightly.
37031 PR middle-end/78703
37032 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
37033 (fmtresult::operator+=): Outlined.
37034 (struct fmtresult): Add ctors.
37035 (struct conversion_spec): Rename...
37036 (struct directive): ...to this. Add and remove data members.
37037 (directive::set_width, directive::set_precision): New functions.
37038 (format_percent): Use fmtresult ctor.
37039 (get_width_and_precision): Remove.
37040 (format_integer): Make naming changes. Avoid computing width and
37042 (format_floating): Same. Adjust indentation.
37043 (format_character, format_none): New functions.
37044 (format_string): Moved character handling to format_character.
37045 (format_directive): Remove arguments, change return type.
37046 (parse_directive): New function.
37047 (pass_sprintf_length::compute_format_length): Move directive
37048 parsing to parse_directive.
37050 2017-01-23 Jakub Jelinek <jakub@redhat.com>
37052 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
37053 (assign_assembler_name_if_needed): ... this.
37054 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
37055 (assign_assembler_name_if_needed): ... this.
37056 (free_lang_data_in_cgraph): Adjust callers.
37057 * cgraphunit.c (cgraph_node::analyze): Likewise.
37058 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
37061 2017-01-23 Richard Biener <rguenther@suse.de>
37063 PR tree-optimization/79088
37064 PR tree-optimization/79188
37065 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
37066 resetting loop bounds after last path deletion. Reset loop
37067 bounds of the target loop, make code match the comments.
37068 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
37069 Make sure loops need no fixups.
37071 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
37073 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
37074 exponent support with double type for first argument.
37075 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
37076 type returned by __builtin_vec_extract_sig,
37077 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
37078 functions from "vector int" to "vector unsigned int" or from
37079 "vector long long int" to "vector unsigned long long int".
37080 Changed type returned by __builtin_vec_extract_exp,
37081 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
37082 functions from "vector int" to "vector unsigned int" or from
37083 "vector long long int" to "vector unsigned long long int".
37084 Changed return type of __builtin_vec_test_data_class,
37085 __builtin_vec_test_data_class_sp, and
37086 __builtin_vec_test_data_class_dp from "vector int" to
37087 "vector bool int" or from "vector long long int" to "vector bool
37088 long long int" and changed second argument type from "unsigned
37089 int" to "int". Added new overloaded function forms "vector float
37090 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
37091 "vector float __builtin_vec_insert_exp_sp (vector float, vector
37092 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
37093 double, vector unsigned long long int)" and "vector double
37094 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
37095 long int)". Changed return type of
37096 __builtin_scalar_test_data_class and
37097 __builtin_scalar_test_data_class_sp and
37098 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
37099 int" and changed second argument from "unsigned int" to "int".
37100 Changed type returned by __builtin_scalar_test_neg,
37101 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
37102 from "int" to "bool int". Added new overloaded function form
37103 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
37104 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
37105 exponent double-precision with floating point first argument.
37106 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
37107 documentation of scalar_test_data_class, scalar_test_neg,
37108 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
37109 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
37110 vec_test_data_class built-in functions to reflect refinements in
37111 their type signatures.
37113 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
37115 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
37117 (aarch64_elf_asm_destructor): Likewise.
37119 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
37121 PR rtl-optimization/78634
37122 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
37123 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
37124 * ifcvt.c (noce_try_cmove): Add missing cost check.
37126 PR rtl-optimization/71724
37127 * combine.c (if_then_else_cond): Look for situations where it is
37128 beneficial to undo the work of one of the recursive calls.
37130 2017-01-23 Bin Cheng <bin.cheng@arm.com>
37132 PR tree-optimization/70754
37133 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
37134 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
37135 combined stmt before it if not NULL.
37136 (combine_chains): Process refs reversely and compute dominance point
37139 2017-01-23 Martin Liska <mliska@suse.cz>
37141 PR tree-optimization/79196
37142 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
37143 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
37145 (strlen_optimize_stmt): Call the renamed function.
37147 2017-01-23 Michael Matz <matz@suse.de>
37149 PR tree-optimization/78384
37150 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
37152 2017-01-23 Richard Biener <rguenther@suse.de>
37154 PR tree-optimization/79186
37155 * tree-vrp.c (register_new_assert_for): Make sure we've seen
37156 both incoming edges before moving an assert.
37158 2017-01-23 Martin Jambor <mjambor@suse.cz>
37160 * ipa-prop.c (load_from_param_1): Removed.
37161 (load_from_unmodified_param): Bits from load_from_param_1 put back
37163 (load_from_param): Removed.
37164 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
37165 with stmt. Reverted back to use of load_from_unmodified_param.
37167 2017-01-23 Martin Jambor <mjambor@suse.cz>
37170 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
37171 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
37172 field a pointer to garbage collected vector, mark lattices and
37173 ipcp_orig_node with GTY((skip)).
37174 (ipa_get_param_count): Adjust to descriptors being a pointer.
37175 (ipa_get_param): Likewise.
37176 (ipa_get_type): Likewise.
37177 (ipa_get_param_move_cost): Likewise.
37178 (ipa_set_param_used): Likewise.
37179 (ipa_get_controlled_uses): Likewise.
37180 (ipa_set_controlled_uses): Likewise.
37181 (ipa_is_param_used): Likewise.
37182 (ipa_node_params_t): Move into garbage collector. New methods insert
37184 (ipa_node_params_sum): Annotate wth GTY(()).
37185 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
37187 (ipa_load_from_parm_agg): Adjust declaration.
37188 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
37189 * ipa-profile.c (ipa_profile): Likewise.
37190 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
37191 (ipa_populate_param_decls): Make descriptors parameter garbage
37193 (ipa_dump_param): Adjust to descriptors being a pointer.
37194 (ipa_alloc_node_params): Likewise.
37195 (ipa_initialize_node_params): Likewise.
37196 (load_from_param_1): Make descriptors parameter garbage collected.
37197 (load_from_unmodified_param): Likewise.
37198 (load_from_param): Likewise.
37199 (ipa_load_from_parm_agg): Likewise.
37200 (ipa_node_params::~ipa_node_params): Removed.
37201 (ipa_free_all_node_params): Remove call to delete operator.
37202 (ipa_node_params_t::insert): New.
37203 (ipa_node_params_t::remove): Likewise.
37204 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
37205 copy known_csts and known_contexts vectors.
37206 (ipa_read_node_info): Adjust to descriptors being a pointer.
37207 (ipcp_modif_dom_walker): Make m_descriptors field garbage
37209 (ipcp_transform_function): Make descriptors variable garbage
37212 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
37214 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
37215 * config/i386/avx512dqintrin.h: Ditto.
37216 * config/i386/avx512fintrin.h: Ditto.
37217 * config/i386/i386.c: Handle new builtins.
37218 * config/i386/i386-builtin.def: Add new builtins.
37219 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
37220 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
37222 2017-01-23 Jakub Jelinek <jakub@redhat.com>
37223 Martin Liska <mliska@suse.cz>
37225 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
37226 * asan.c (asan_expand_poison_ifn): Support stores and use
37227 appropriate ASAN report function.
37228 * internal-fn.c (expand_ASAN_POISON_USE): New function.
37229 * internal-fn.def (ASAN_POISON_USE): Declare.
37230 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
37231 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
37232 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
37233 ASAN_POISON calls w/o LHS.
37234 * tree-ssa.c (execute_update_addresses_taken): Create clobber
37235 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
37236 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
37237 * gimplify.c (asan_poison_variables): Add attribute
37238 use_after_scope_memory to variables that really needs to live
37240 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
37241 having the attribute.
37243 2017-01-23 Martin Liska <mliska@suse.cz>
37245 * asan.c (create_asan_shadow_var): New function.
37246 (asan_expand_poison_ifn): Likewise.
37247 * asan.h (asan_expand_poison_ifn): New declaration.
37248 * internal-fn.c (expand_ASAN_POISON): Likewise.
37249 * internal-fn.def (ASAN_POISON): New builtin.
37250 * sanopt.c (pass_sanopt::execute): Expand
37251 asan_expand_poison_ifn.
37252 * tree-inline.c (copy_decl_for_dup_finish): Make function
37254 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
37255 * tree-ssa.c (is_asan_mark_p): New function.
37256 (execute_update_addresses_taken): Rewrite local variables
37257 (identified just by use-after-scope as addressable) into SSA.
37259 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
37261 * doc/install.texi (Specific): opensource.apple.com uses https
37262 now. Remove trailing slash.
37264 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
37266 * README.Portability: Remove note on an Irix compatibility issue.
37268 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
37270 * gcov.c (INCLUDE_ALGORITHM): Define.
37271 (INCLUDE_VECTOR): Define.
37272 No longer include <vector> and <algorithm> directly.
37274 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
37276 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
37278 * doc/invoke.texi (Code Gen Options): Ditto.
37280 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
37283 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
37285 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
37287 rtl-optimization/79125
37288 * cprop.c (local_cprop_pass): Handle cases where we make an
37289 unconditional trap.
37291 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
37295 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
37296 read from, for big endian.
37298 2017-01-20 Jiong Wang <jiong.wang@arm.com>
37300 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
37301 register pauth builtins for LP64 only.
37303 2017-01-20 Marek Polacek <polacek@redhat.com>
37306 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
37309 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
37311 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
37313 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
37314 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
37315 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
37317 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37320 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
37321 in big-endian mode when they are not a single duplicated value.
37323 2017-01-20 Richard Biener <rguenther@suse.de>
37325 * BASE-VER: Bump to 7.0.1.
37327 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
37329 * omp-low.c (omplow_simd_context): New struct. Use it...
37330 (lower_rec_simd_input_clauses): ...here and...
37331 (lower_rec_input_clauses): ...here to hold common data. Adjust all
37332 references to idx, lane, max_vf, is_simt.
37334 2017-01-20 Graham Markall <graham.markall@embecosm.com>
37336 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
37339 2017-01-20 Martin Jambor <mjambor@suse.cz>
37341 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
37342 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
37344 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
37345 (GTFILES): Rename hsa.c to hsa-common.c.
37346 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
37347 * hsa-dump.c: Likewise.
37348 * hsa-gen.c: Likewise.
37349 * hsa-regalloc.c: Likewise.
37350 * ipa-hsa.c: Likewise.
37351 * omp-expand.c: Likewise.
37352 * omp-low.c: Likewise.
37353 * toplev.c: Likewise.
37355 2017-01-20 Marek Polacek <polacek@redhat.com>
37358 * doc/invoke.texi: Document -Wduplicated-branches.
37359 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
37360 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
37361 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
37362 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
37363 return 0 only when not OEP_LEXICOGRAPHIC.
37364 (fold_build_cleanup_point_expr): Use the expression
37365 location when building CLEANUP_POINT_EXPR.
37366 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
37367 * tree.c (add_expr): Handle error_mark_node.
37369 2017-01-20 Martin Liska <mliska@suse.cz>
37372 * tree-profile.c (init_ic_make_global_vars): Do not call
37374 (gimple_init_gcov_profiler): Likewise.
37376 2017-01-20 Martin Liska <mliska@suse.cz>
37379 * cgraph.h (maybe_create_reference): Remove argument and
37381 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
37383 * ipa-cp.c (create_specialized_node): Likewise.
37384 * symtab.c (symtab_node::maybe_create_reference): Handle
37385 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
37387 2017-01-20 Martin Liska <mliska@suse.cz>
37389 * read-rtl-function.c (function_reader::create_function): Use
37390 build_decl instread of build_decl_stat.
37392 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
37394 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
37395 * config/i386/avx512dqintrin.h: Ditto.
37396 * config/i386/avx512fintrin.h: Ditto.
37397 * config/i386/i386-builtin-types.def: Add new types.
37398 * config/i386/i386.c: Handle new types.
37399 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
37400 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
37401 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
37402 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
37403 (__builtin_ia32_kshiftridi): New.
37404 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
37406 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
37410 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
37411 define to rs6000_init_stack_protect_guard.
37412 (rs6000_init_stack_protect_guard): New function.
37414 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
37415 Yunqiang Su <yunqiang.su@imgtec.com>
37417 * config.gcc (supported_defaults): Add madd4.
37418 (with_madd4): Add validation.
37419 (all_defaults): Add madd4.
37420 * config/mips/mips.opt (mmadd4): New option.
37421 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
37423 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
37425 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
37426 (ISA_HAS_FUSED_MADD4): Likewise.
37427 * doc/invoke.texi (-mmadd4): Document the new option.
37428 * doc/install.texi (--with-madd4): Document the new option.
37430 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37432 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
37433 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
37434 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
37435 (aarch64_init_pauth_hint_builtins): New.
37436 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
37437 (aarch64_expand_builtin): Expand new builtins.
37439 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37441 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
37442 * combine-stack-adj.c (no_unhandled_cfa): Handle
37443 REG_CFA_TOGGLE_RA_MANGLE.
37444 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
37445 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
37446 info for return address signing.
37447 (aarch64_expand_epilogue): Likewise.
37449 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37451 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
37452 * config/aarch64/aarch64-protos.h
37453 (aarch64_return_address_signing_enabled): New declaration.
37454 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
37456 (aarch64_expand_prologue): Sign return address before it's pushed onto
37458 (aarch64_expand_epilogue): Authenticate return address fetched from
37460 (aarch64_override_options): Sanity check for ILP32 and ISA level.
37461 (aarch64_attributes): New function attributes for "sign-return-address".
37462 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
37463 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
37464 ("*do_return"): Generate combined instructions according to key index.
37465 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
37466 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
37468 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
37469 * config/aarch64/aarch64.opt (msign-return-address=): New.
37470 * doc/extend.texi (AArch64 Function Attributes): Documents
37471 "sign-return-address=".
37472 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
37474 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
37476 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
37477 overall option summary.
37479 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37481 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
37482 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
37483 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
37484 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
37486 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
37488 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
37489 -mpower9-minmax by default for -mcpu=power9.
37490 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
37491 128-bit floating point.
37493 2017-01-20 Alan Modra <amodra@gmail.com>
37495 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
37496 optimizing for size.
37498 2017-01-20 Alan Modra <amodra@gmail.com>
37501 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
37502 for strcmp and strncmp from corresponding builtin decl.
37504 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
37506 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
37507 instead of i386/rtems-64.h.
37508 * config/i386/rtems-64.h: Remove.
37510 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
37514 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
37516 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
37518 2017-01-19 Tamar Christina <tamar.christina@arm.com>
37520 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
37521 Change int to HOST_WIDE_INT.
37522 * config/aarch64/aarch64-protos.h
37523 (aarch64_simd_gen_const_vector_dup): Likewise.
37524 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
37526 2017-01-19 David Malcolm <dmalcolm@redhat.com>
37528 * langhooks-def.h (lhd_type_for_size): New decl.
37529 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
37530 * langhooks.c (lhd_type_for_size): New function, taken from
37533 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
37535 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
37536 define_bypass for CR latency.
37537 (power9-cracked-alu): Update bypass latency and remove power9-branch.
37538 (power9-alu2): Add define_bypass for CR latency.
37540 (power9-mul): Update insn latency.
37541 (power9-mul-compare): Update insn latency, bypass latency and remove
37544 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37546 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
37548 * config/aarch64/aarch64.md
37549 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
37550 aarch64_nopcrelative_literal_loads.
37551 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
37553 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
37555 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
37556 TARGET_LOONGSON_3A.
37557 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
37559 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
37562 * config.gcc (supported_defaults): Add lxc1-sxc1.
37563 (with_lxc1_sxc1): Add validation.
37564 (all_defaults): Add lxc1-sxc1.
37565 * config/mips/mips.opt (mlxc1-sxc1): New option.
37566 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
37568 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
37569 __mips_no_lxc1_sxc1.
37570 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
37571 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
37572 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
37574 2017-01-19 Richard Biener <rguenther@suse.de>
37576 PR tree-optimization/72488
37577 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
37578 sure to restore SSA info.
37579 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
37581 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
37583 PR rtl-optimization/79121
37584 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
37585 of the inner type when shifting an extended value.
37587 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
37590 * symtab.c (symtab_node::equal_address_to): Fix comparing of
37591 interposable aliases.
37593 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
37596 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
37597 Use the evmergelohi instruction.
37598 (mov_si<mode>_e500_subreg4_2_le): Likewise.
37599 (mov_sitf_e500_subreg8_2_be): Likewise.
37600 (mov_sitf_e500_subreg12_2_le): Likewise.
37601 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
37602 (mov_si<mode>_e500_subreg4_2_be): Likewise.
37603 (mov_sitf_e500_subreg8_2_le): Likewise.
37604 (mov_sitf_e500_subreg12_2_be): Likewise.
37606 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37608 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
37609 attribute from vecsimple to vecperm.
37610 (altivec_vbpermq2): Likewise.
37612 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37615 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
37617 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37618 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
37619 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
37620 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
37621 case where N arg is SIZE_MAX.
37622 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
37623 (cmpstrsi): Add pattern.
37625 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
37627 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
37628 __builtin_vec_revb builtins.
37629 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
37630 built-in functions to support generation of the ISA 3.0 XXBR<x>
37631 vector byte reverse instructions.
37632 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
37633 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
37634 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
37635 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
37636 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
37637 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
37638 (P9V_BUILTIN_VEC_REVB): Likewise.
37639 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
37640 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
37641 (p9_xxbrq_v16qi): Likewise.
37642 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
37643 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
37644 (p9_xxbrh_v8hi): Likewise.
37645 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
37646 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
37647 vec_revb built-in functions.
37649 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
37651 PR rtl-optimization/78952
37652 * config/i386/i386.md (any_extract): New code iterator.
37653 (*insvqi_2): Use any_extract for source operand.
37654 (*insvqi_3): Use any_shiftrt for source operand.
37656 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
37658 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
37660 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
37662 2017-01-18 Matthias Klose <doko@ubuntu.com>
37664 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
37666 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37668 * config/rs6000/altivec.h (vec_bperm): Change #define.
37669 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
37670 (altivec_vbpermq2): New define_insn.
37671 (altivec_vbpermd): Likewise.
37672 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
37673 function interface.
37674 (VBPERMD): Likewise.
37675 (VBPERM): New polymorphic function interface.
37676 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
37677 Add entries for P9V_BUILTIN_VEC_VBPERM.
37678 * doc/extend.texi: Add interfaces for vec_bperm.
37680 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37682 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
37683 first letter of error messages.
37684 (s390_resolve_overloaded_builtin): Likewise.
37685 * config/s390/s390.c (s390_expand_builtin): Likewise.
37686 (s390_invalid_arg_for_unprototyped_fn): Likewise.
37687 (s390_valid_target_attribute_inner_p): Likewise.
37688 * config/s390/s390.md ("tabort"): Likewise.
37690 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
37692 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
37693 (ISA_AVOID_DIV_HILO): New macro.
37694 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
37695 (ISA_HAS_DDIV): Likewise.
37697 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37699 * doc/invoke.texi (fabi-version): Correct number of occurrences.
37701 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37703 * doc/invoke.texi (fabi-version): Spelling fix.
37705 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37708 * doc/invoke.texi (fabi-version): Mention mangling fix for
37711 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37714 * doc/invoke.texi (fabi-version): Document discriminator mangling.
37716 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
37719 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
37720 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
37722 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
37724 (stack_protect_test): Ditto.
37725 * config/rs6000/rs6000.opt (mstack-protector-guard=,
37726 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
37728 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
37729 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
37730 -mstack-protector-guard-offset=.
37731 (RS/6000 and PowerPC Options): Ditto.
37733 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37735 * config/i386/i386.h (MASK_CLASS_P): New define.
37736 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
37737 there are no registers from different register sets also when
37738 mask registers are used. Update function comment.
37739 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
37740 to (*k/*r) and (*k/*km) alternatives.
37742 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
37744 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
37745 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
37746 (EH_RETURN_HANDLER_RTX): New define.
37747 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
37748 Force frame pointer in EH return functions.
37749 (aarch64_expand_epilogue): Add barrier for eh_return.
37750 (aarch64_final_eh_return_addr): Remove.
37751 (aarch64_eh_return_handler_rtx): New function.
37752 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
37754 (aarch64_eh_return_handler_rtx): New prototype.
37756 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37758 * config/rs6000/altivec.h (vec_rlmi): New #define.
37759 (vec_vrlnm): Likewise.
37760 (vec_rlnm): Likewise.
37761 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
37762 (UNSPEC_VRLNM): Likewise.
37763 (VIlong): New mode iterator.
37764 (altivec_vrl<VI_char>mi): New define_insn.
37765 (altivec_vrl<VI_char>nm): Likewise.
37766 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
37768 (VRLDNM): Likewise.
37769 (RLNM): New polymorphic function entry.
37770 (VRLWMI): New monomorphic function entry.
37771 (VRLDMI): Likewise.
37772 (RLMI): New polymorphic function entry.
37773 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
37774 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
37775 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
37778 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37781 * dwarf2out.c (field_byte_offset): Restore the
37782 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
37783 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
37784 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
37787 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37790 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
37792 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37795 * dwarf2out.c (add_data_member_location_attribute): For constant
37796 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
37797 instead of DW_AT_data_member_location, DW_AT_bit_offset and
37798 DW_AT_byte_size attributes.
37800 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37802 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
37803 after forcing to constant memory when the code model is medium.
37805 2017-01-17 Julia Koval <julia.koval@intel.com>
37808 * config/i386/avx512fintrin.h
37809 (_mm512_i32gather_ps): Change __addr type to void const*.
37810 (_mm512_mask_i32gather_ps): Ditto.
37811 (_mm512_i32gather_pd): Ditto.
37812 (_mm512_mask_i32gather_pd): Ditto.
37813 (_mm512_i64gather_ps): Ditto.
37814 (_mm512_mask_i64gather_ps): Ditto.
37815 (_mm512_i64gather_pd): Ditto.
37816 (_mm512_mask_i64gather_pd): Ditto.
37817 (_mm512_i32gather_epi32): Ditto.
37818 (_mm512_mask_i32gather_epi32): Ditto.
37819 (_mm512_i32gather_epi64): Ditto.
37820 (_mm512_mask_i32gather_epi64): Ditto.
37821 (_mm512_i64gather_epi32): Ditto.
37822 (_mm512_mask_i64gather_epi32): Ditto.
37823 (_mm512_i64gather_epi64): Ditto.
37824 (_mm512_mask_i64gather_epi64): Ditto.
37825 (_mm512_i32scatter_ps): Change __addr type to void*.
37826 (_mm512_mask_i32scatter_ps): Ditto.
37827 (_mm512_i32scatter_pd): Ditto.
37828 (_mm512_mask_i32scatter_pd): Ditto.
37829 (_mm512_i64scatter_ps): Ditto.
37830 (_mm512_mask_i64scatter_ps): Ditto.
37831 (_mm512_i64scatter_pd): Ditto.
37832 (_mm512_mask_i64scatter_pd): Ditto.
37833 (_mm512_i32scatter_epi32): Ditto.
37834 (_mm512_mask_i32scatter_epi32): Ditto.
37835 (_mm512_i32scatter_epi64): Ditto.
37836 (_mm512_mask_i32scatter_epi64): Ditto.
37837 (_mm512_i64scatter_epi32): Ditto.
37838 (_mm512_mask_i64scatter_epi32): Ditto.
37839 (_mm512_i64scatter_epi64): Ditto.
37840 (_mm512_mask_i64scatter_epi64): Ditto.
37841 * config/i386/avx512pfintrin.h
37842 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
37843 (_mm512_mask_prefetch_i32gather_ps): Ditto.
37844 (_mm512_mask_prefetch_i64gather_pd): Ditto.
37845 (_mm512_mask_prefetch_i64gather_ps): Ditto.
37846 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
37847 (_mm512_prefetch_i32scatter_ps): Ditto.
37848 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
37849 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
37850 (_mm512_prefetch_i64scatter_pd): Ditto.
37851 (_mm512_prefetch_i64scatter_ps): Ditto.
37852 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
37853 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
37854 * config/i386/avx512vlintrin.h
37855 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
37856 (_mm_mmask_i32gather_ps): Ditto.
37857 (_mm256_mmask_i32gather_pd): Ditto.
37858 (_mm_mmask_i32gather_pd): Ditto.
37859 (_mm256_mmask_i64gather_ps): Ditto.
37860 (_mm_mmask_i64gather_ps): Ditto.
37861 (_mm256_mmask_i64gather_pd): Ditto.
37862 (_mm_mmask_i64gather_pd): Ditto.
37863 (_mm256_mmask_i32gather_epi32): Ditto.
37864 (_mm_mmask_i32gather_epi32): Ditto.
37865 (_mm256_mmask_i32gather_epi64): Ditto.
37866 (_mm_mmask_i32gather_epi64): Ditto.
37867 (_mm256_mmask_i64gather_epi32): Ditto.
37868 (_mm_mmask_i64gather_epi32): Ditto.
37869 (_mm256_mmask_i64gather_epi64): Ditto.
37870 (_mm_mmask_i64gather_epi64): Ditto.
37871 (_mm256_i32scatter_ps): Change __addr type to void*.
37872 (_mm256_mask_i32scatter_ps): Ditto.
37873 (_mm_i32scatter_ps): Ditto.
37874 (_mm_mask_i32scatter_ps): Ditto.
37875 (_mm256_i32scatter_pd): Ditto.
37876 (_mm256_mask_i32scatter_pd): Ditto.
37877 (_mm_i32scatter_pd): Ditto.
37878 (_mm_mask_i32scatter_pd): Ditto.
37879 (_mm256_i64scatter_ps): Ditto.
37880 (_mm256_mask_i64scatter_ps): Ditto.
37881 (_mm_i64scatter_ps): Ditto.
37882 (_mm_mask_i64scatter_ps): Ditto.
37883 (_mm256_i64scatter_pd): Ditto.
37884 (_mm256_mask_i64scatter_pd): Ditto.
37885 (_mm_i64scatter_pd): Ditto.
37886 (_mm_mask_i64scatter_pd): Ditto.
37887 (_mm256_i32scatter_epi32): Ditto.
37888 (_mm256_mask_i32scatter_epi32): Ditto.
37889 (_mm_i32scatter_epi32): Ditto.
37890 (_mm_mask_i32scatter_epi32): Ditto.
37891 (_mm256_i32scatter_epi64): Ditto.
37892 (_mm256_mask_i32scatter_epi64): Ditto.
37893 (_mm_i32scatter_epi64): Ditto.
37894 (_mm_mask_i32scatter_epi64): Ditto.
37895 (_mm256_i64scatter_epi32): Ditto.
37896 (_mm256_mask_i64scatter_epi32): Ditto.
37897 (_mm_i64scatter_epi32): Ditto.
37898 (_mm_mask_i64scatter_epi32): Ditto.
37899 (_mm256_i64scatter_epi64): Ditto.
37900 (_mm256_mask_i64scatter_epi64): Ditto.
37901 (_mm_i64scatter_epi64): Ditto.
37902 (_mm_mask_i64scatter_epi64): Ditto.
37903 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
37904 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
37905 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
37906 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
37907 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
37908 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
37909 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
37910 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
37911 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
37912 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
37913 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
37914 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
37915 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
37916 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
37917 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
37918 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
37919 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
37920 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
37921 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
37922 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
37923 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
37924 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
37925 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
37926 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
37927 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
37928 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
37929 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
37930 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
37931 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
37932 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
37933 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
37934 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
37935 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
37936 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
37937 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
37938 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
37939 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
37940 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
37941 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
37942 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
37943 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
37944 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
37945 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
37946 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
37947 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
37948 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
37949 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
37950 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
37951 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
37952 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
37953 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
37954 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
37955 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
37956 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
37957 definitions accordingly.
37959 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
37960 Kuan-Lin Chen <kuanlinchentw@gmail.com>
37963 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
37966 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
37969 * ira-conflicts.c (ira_build_conflicts): Update total conflict
37970 hard regs for inner regno.
37972 2017-01-17 Martin Liska <mliska@suse.cz>
37975 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
37976 assumption and add comment.
37978 2017-01-17 Nathan Sidwell <nathan@acm.org>
37980 * ipa-visibility.c (localize_node): New function, broken out of ...
37981 (function_and_variable_visibility): ... here. Call it.
37983 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
37985 PR middle-end/77445
37986 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
37987 correctly set frequency of oudgoing edge.
37988 (duplicate_thread_path): Fix profile updating.
37990 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37993 * configure.ac: Add GCC_BASE_VER.
37994 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
37995 version from BASE-VER file.
37996 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
37997 (gcc.o): Depend on $(BASEVER).
37998 * common.opt (dumpfullversion): New option.
37999 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
38000 * doc/invoke.texi: Document -dumpfullversion.
38001 * doc/install.texi: Document --with-gcc-major-version-only.
38002 * configure: Regenerated.
38004 2017-01-17 Richard Biener <rguenther@suse.de>
38006 PR tree-optimization/71433
38007 * tree-vrp.c (register_new_assert_for): Merge same asserts
38008 on all incoming edges.
38009 (process_assert_insertions_for): Handle insertions at the
38012 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
38014 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
38015 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
38017 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
38020 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
38023 2017-01-17 Alan Modra <amodra@gmail.com>
38026 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
38027 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
38028 symbolic stack limit when pic.
38030 2017-01-16 Martin Sebor <msebor@redhat.com>
38032 PR tree-optimization/78608
38033 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
38035 2017-01-16 Jeff Law <law@redhat.com>
38038 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
38039 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
38040 for several include directories that may be relative to sysroot.
38041 * config/i386/x-mingw32 (gplus_includedir): Define.
38042 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
38043 (native_system_includedir): Likewise.
38044 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
38045 override if TARGET_SYSTEM_ROOT is defined.
38046 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
38048 PR tree-optimization/79090
38049 PR tree-optimization/33562
38050 PR tree-optimization/61912
38051 PR tree-optimization/77485
38052 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
38053 and computed trims into the dump file.
38055 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
38057 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
38059 2017-01-16 Jakub Jelinek <jakub@redhat.com>
38062 * gimplify.c (gimplify_init_constructor): If want_value and
38063 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
38067 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
38068 sequence. Formatting fixes.
38069 (doloop_optimize): Formatting fixes.
38072 * gcc.c (debug_level_greater_than_spec_func): New function.
38073 (static_spec_functions): Add debug-level-gt spec function.
38074 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
38076 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
38077 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
38078 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
38079 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
38080 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
38081 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
38083 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
38085 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
38086 QImode fixups to general and mask registers only.
38088 2017-01-16 Carl Love <cel@us.ibm.com>
38090 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
38091 for built-in functions
38092 vector signed char vec_nabs (vector signed char)
38093 vector signed short vec_nabs (vector signed short)
38094 vector signed int vec_nabs (vector signed int)
38095 vector signed long long vec_nabs (vector signed long long)
38096 vector float vec_nabs (vector float)
38097 vector double vec_nabs (vector double)
38098 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
38100 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
38101 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
38102 * doc/extend.texi: Update the documentation file for the new built-in
38105 2017-01-16 Martin Sebor <msebor@redhat.com>
38107 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
38110 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38112 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
38113 UNSPEC_VSX__XXSPLTD to require special splat handling.
38115 2017-01-16 David Malcolm <dmalcolm@redhat.com>
38118 * system.h: Poison strndup.
38120 2017-01-16 Alan Modra <amodra@gmail.com>
38123 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
38126 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
38128 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
38130 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
38132 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
38133 call recog here. Assert that INSN_CODE (insn) is non-negative.
38135 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
38138 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
38140 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
38141 in the currently scheduled RTL fragment.
38143 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
38145 PR rtl-optimization/78751
38146 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
38149 2017-01-14 Jeff Law <law@redhat.com>
38151 PR tree-optimization/79090
38152 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
38153 variable length stores.
38154 (compute_trims): Delete dead assignment to *trim_tail.
38155 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
38158 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
38160 PR rtl-optimization/78626
38161 PR rtl-optimization/78727
38162 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
38163 of a block, and split such blocks after everything else is finished.
38165 2017-01-14 Alan Modra <amodra@gmail.com>
38168 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
38169 target legitimate_combined_insn.
38170 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
38171 (rs6000_legitimate_combined_insn): New function.
38172 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
38174 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
38175 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
38176 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
38178 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
38180 * doc/frontends.texi (G++ and GCC): Remove references to Java.
38182 2017-01-13 Jeff Law <law@redhat.com>
38184 PR tree-optimization/33562
38185 PR tree-optimization/61912
38186 PR tree-optimization/77485
38187 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
38189 (delete_dead_assignment): Likewise.
38190 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
38191 statement to delete_dead_call and delete_dead_assignment.
38193 2017-01-13 David Malcolm <dmalcolm@redhat.com>
38196 * substring-locations.c (format_warning_va): Strengthen case 1 so
38197 that both endpoints of the substring must be within the format
38198 range for just the substring to be printed.
38200 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
38202 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
38203 * config/i386/i386.c (ix86_target_string): Add missing options
38204 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
38205 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
38206 flags_other and ix86_target_other to flags2_other. Display unknown
38208 (ix86_valid_target_attribute_inner_p): Add missing options and
38209 reorder options by implied ISAs, as in ix86_target_string.
38211 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
38213 * hash-table.h (hash_table::too_empty_p): New function.
38214 (hash_table::expand): Use it.
38215 (hash_table::traverse): Likewise.
38216 (hash_table::empty_slot): Use sizeof (value_type) instead of
38217 sizeof (PTR) to convert bytes to elements. Shrink the table
38218 if the current size is excessive for the current number of
38221 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
38223 * ira-costs.c (record_reg_classes): Break from the inner loop
38224 early once alt_fail is known to be true. Update outer loop
38225 handling accordingly.
38227 2017-01-13 Jeff Law <law@redhat.com>
38229 * tree-ssa-dse.c (decrement_count): New function.
38230 (increment_start_addr, maybe_trim_memstar_call): Likewise.
38231 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
38232 when we know the partially dead statement is a mem* function.
38234 PR tree-optimization/61912
38235 PR tree-optimization/77485
38236 * tree-ssa-dse.c: Include expr.h.
38237 (maybe_trim_constructor_store): New function.
38238 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
38240 PR tree-optimization/33562
38241 PR tree-optimization/61912
38242 PR tree-optimization/77485
38243 * doc/invoke.texi: Document new dse-max-object-size param.
38244 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
38245 * tree-ssa-dse.c: Include params.h.
38246 (dse_store_status): New enum.
38247 (initialize_ao_ref_for_dse): New, partially extracted from
38249 (valid_ao_ref_for_dse, normalize_ref): New.
38250 (setup_live_bytes_from_ref, compute_trims): Likewise.
38251 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
38252 (maybe_trim_partially_dead_store): Likewise.
38253 (maybe_trim_complex_store): Likewise.
38254 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
38255 Track what bytes live from the original store. Return tri-state
38256 for dead, partially dead or live.
38257 (dse_dom_walker): Add constructor, destructor and new private members.
38258 (delete_dead_call, delete_dead_assignment): New extracted from
38260 (dse_optimize_stmt): Make a member of dse_dom_walker.
38261 Use initialize_ao_ref_for_dse.
38263 PR tree-optimization/33562
38264 PR tree-optimization/61912
38265 PR tree-optimization/77485
38266 * sbitmap.h (bitmap_count_bits): Prototype.
38267 (bitmap_clear_range, bitmap_set_range): Likewise.
38268 * sbitmap.c (bitmap_clear_range): New function.
38269 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
38271 2017-01-13 Martin Liska <mliska@suse.cz>
38274 * function.c (set_cfun): Add new argument force.
38275 * function.h (set_cfun): Likewise.
38276 * ipa-inline-transform.c (inline_call): Use the function when
38277 strict alising from is dropped for function we inline to.
38279 2017-01-13 Richard Biener <rguenther@suse.de>
38281 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
38282 for dumping GIMPLE INTEGER_CSTs.
38284 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
38286 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
38287 to 201112L since C++17.
38289 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
38292 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
38293 if -fsanitize=kernel-address is present.
38295 2017-01-13 Richard Biener <rguenther@suse.de>
38297 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
38298 as _Literal ( type ) number in case usual suffixes do not
38299 preserve all information.
38301 2017-01-13 Richard Biener <rguenther@suse.de>
38303 PR tree-optimization/77283
38304 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
38305 and ssa-iterators.h.
38306 (is_feasible_trace): Implement a cost model based on joiner
38309 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
38312 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
38313 char or short to __float128/_Float128 directly.
38315 2017-01-12 Martin Sebor <msebor@redhat.com>
38317 to -Wformat-overflow.
38318 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
38319 (min_bytes_remaining): Same.
38320 (get_string_length): Same.
38321 (format_string): Same.
38322 (format_directive): Same.
38324 (pass_sprintf_length::handle_gimple_call): Same.
38326 2017-01-12 Jakub Jelinek <jakub@redhat.com>
38328 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
38329 info.nowrite calls with no lhs that can't throw. Return bool
38330 whether gsi_remove has been called or not.
38331 (pass_sprintf_length::handle_gimple_call): Return bool whether
38332 try_substitute_return_value called gsi_remove. Formatting fix.
38333 (pass_sprintf_length::execute): Don't use gsi_remove if
38334 handle_gimple_call returned true.
38337 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
38338 be removed due to side-effects, don't remove following barrier nor
38339 turn the successor edge into fallthru edge.
38341 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38344 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
38345 element-reversing loads and stores as not swappable.
38347 2017-01-12 Nathan Sidwell <nathan@acm.org>
38348 Nicolai Stange <nicstange@gmail.com>
38350 * combine.c (try_combine): Don't ignore result of overlap checking
38351 loop. Combine overlap & asm check into single loop.
38353 2017-01-12 Richard Biener <rguenther@suse.de>
38355 * tree-pretty-print.c (dump_generic_node): Provide -gimple
38356 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
38358 2017-01-12 Richard Biener <rguenther@suse.de>
38360 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
38361 and TS_TARGET_OPTION directly derive from TS_BASE.
38362 * tree-core.h (tree_optimization_option): Derive from tree_base.
38363 (tree_target_option): Likewise.
38365 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
38367 * config/i386/i386.c (memory_address_length): Increase len
38368 only when rip_relative_addr_p returns false.
38370 2017-01-11 Julia Koval <julia.koval@intel.com>
38372 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
38373 (OPTION_MASK_ISA_SGX_SET): New.
38374 (ix86_handle_option): Handle OPT_msgx.
38375 * config.gcc: Added sgxintrin.h.
38376 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
38377 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
38378 * config/i386/i386.c (ix86_target_string): Add -msgx.
38380 (ix86_option_override_internal): Handle new options.
38381 (ix86_valid_target_attribute_inner_p): Add sgx.
38382 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
38383 * config/i386/i386.opt: Add msgx.
38384 * config/i386/sgxintrin.h: New file.
38385 * config/i386/x86intrin.h: Add sgxintrin.h.
38387 2017-01-11 Jakub Jelinek <jakub@redhat.com>
38390 * fold-const.c (maybe_nonzero_address): Return 1 for function
38392 (tree_single_nonzero_warnv_p): Don't handle function local objects
38396 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
38399 2017-01-11 David Malcolm <dmalcolm@redhat.com>
38402 * opts.c: Include "spellcheck.h"
38403 (struct string_fragment): New struct.
38404 (struct edit_distance_traits<const string_fragment &>): New
38406 (get_closest_sanitizer_option): New function.
38407 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
38409 2017-01-11 Jakub Jelinek <jakub@redhat.com>
38411 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
38413 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
38414 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
38415 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
38416 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
38417 for initial die_offset if dwarf_split_debug_info.
38418 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
38419 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
38421 (output_skeleton_debug_sections): Formatting fix. Use
38422 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
38423 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
38425 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
38427 * config/arm/cortex-a53.md: Add bypasses for
38428 cortex_a53_r2f_cvt.
38429 (cortex_a53_r2f): Only use for transfers.
38430 (cortex_a53_f2r): Likewise.
38431 (cortex_a53_r2f_cvt): Add reservation for conversions.
38432 (cortex_a53_f2r_cvt): Likewise.
38434 2017-01-11 Tamar Christina <tamar.christina@arm.com>
38436 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
38437 to all inlined functions, change static to extern.
38439 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
38442 * config/arm/arm.c (legitimize_pic_address): Handle reference to
38444 (arm_assemble_integer): Likewise.
38446 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
38448 * config.gcc: Use new awk script to check CPU, FPU and architecture
38449 parameters for --with-... options.
38450 * config/arm/parsecpu.awk: New file
38451 * config/arm/arm-cpus.in: New file.
38452 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
38454 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
38456 * config/arm/t-arm: Update dependency rules.
38457 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
38458 of processing .def files.
38459 * config/arm/genopt.sh: Deleted.
38460 * config/arm/gentune.sh: Deleted.
38461 * config/arm/arm-cores.def: Deleted.
38462 * config/arm/arm-arches.def: Deleted.
38463 * config/arm/arm-fpus.def: Deleted.
38464 * config/arm/arm-tune.md: Regenerated.
38465 * config/arm/arm-tables.opt: Regenerated.
38466 * config/arm/arm-cpu.h: New generated file.
38467 * config/arm/arm-cpu-data.h: New generated file.
38468 * config/arm/arm-cpu-cdata.h: New generated file.
38470 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
38473 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
38475 (input_varpool_node): Unpack dynamically_initialized bit.
38477 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
38479 PR rtl-optimization/79032
38480 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
38481 the alignment of the adjusted memory reference against that of MODE,
38482 instead of the alignment of the original memory reference.
38484 2017-01-11 Martin Jambor <mjambor@suse.cz>
38486 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
38488 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
38489 decorated functions.
38491 2017-01-11 Richard Biener <rguenther@suse.de>
38493 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
38494 set range/nonnull info for PHI results. Do not set it on
38495 stmts marked for removal.
38497 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
38499 * expr.c (store_field): In the bitfield case, fetch the return value
38500 from the registers before applying a single big-endian adjustment.
38501 Always do a final load for a BLKmode value not larger than a word.
38503 2017-01-10 David Malcolm <dmalcolm@redhat.com>
38506 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
38507 that we correctly handle column numbers greater than
38508 LINE_MAP_MAX_COLUMN_NUMBER.
38510 2017-01-10 Martin Sebor <msebor@redhat.com>
38512 PR middle-end/78245
38513 * gimple-ssa-sprintf.c (get_destination_size): Call
38514 {init,fini}object_sizes.
38515 * tree-object-size.c (addr_object_size): Adjust.
38516 (pass_through_call): Adjust.
38517 (pass_object_sizes::execute): Adjust.
38518 * tree-object-size.h (fini_object_sizes): Declare.
38520 2017-01-10 Martin Sebor <msebor@redhat.com>
38522 PR tree-optimization/78775
38523 * builtins.c (get_size_range): Move...
38524 * calls.c: ...to here.
38525 (alloc_max_size): Accept zero argument.
38526 (operand_signed_p): Remove.
38527 (maybe_warn_alloc_args_overflow): Call get_size_range.
38528 * calls.h (get_size_range): Declare.
38530 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
38532 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
38533 from TI's devices.csv file as of September 2016.
38534 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
38536 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
38538 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
38539 * doc/invoke.texi: Likewise.
38540 * doc/md.texi: Likewise.
38541 * doc/objc.texi: Likewise.
38543 2017-01-10 Joshua Conner <joshconner@google.com>
38545 * config/arm/fuchsia-elf.h: New file.
38546 * config/fuchsia.h: New file.
38547 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
38548 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
38550 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
38552 2016-01-10 Richard Biener <rguenther@suse.de>
38554 PR tree-optimization/79034
38555 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
38556 Propagate out degenerate PHIs in the joiner.
38558 2017-01-10 Martin Liska <mliska@suse.cz>
38560 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
38561 (sort_congruence_classes_by_decl_uid): Likewise.
38562 (sort_congruence_class_groups_by_decl_uid): Likewise.
38563 (sem_item_optimizer::merge_classes): Sort class, groups in these
38564 classes and members in the groups by DECL_UID of declarations.
38565 This would make merge operations stable.
38567 2017-01-10 Martin Liska <mliska@suse.cz>
38569 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
38570 usage of m_classes_vec.
38571 (sem_item_optimizer::~sem_item_optimizer): Likewise.
38572 (sem_item_optimizer::get_group_by_hash): Likewise.
38573 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
38574 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
38575 (sem_item_optimizer::verify_classes): Likewise.
38576 (sem_item_optimizer::process_cong_reduction): Likewise.
38577 (sem_item_optimizer::dump_cong_classes): Likewise.
38578 (sem_item_optimizer::merge_classes): Likewise.
38579 * ipa-icf.h (congruence_class_hash): Rename from
38580 congruence_class_group_hash. Remove declaration of m_classes_vec.
38582 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
38584 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
38585 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
38586 * config.gcc: Add avx512vpopcntdqintrin.h.
38587 * config/i386/avx512vpopcntdqintrin.h: New.
38588 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
38589 * config/i386/i386-builtin-types.def: Add new types.
38590 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
38591 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
38592 __builtin_ia32_vpopcountq_v8di_mask): New.
38593 * config/i386/i386-c.c (ix86_target_macros_internal): Define
38594 __AVX512VPOPCNTDQ__.
38595 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
38596 (PTA_AVX512VPOPCNTDQ): Define.
38597 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
38598 TARGET_AVX512VPOPCNTDQ_P): Define.
38599 * config/i386/i386.opt: Add mavx512vpopcntdq.
38600 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
38601 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
38603 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38605 PR middle-end/77484
38606 * predict.def (PRED_CALL): Set to 67.
38608 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
38610 * expr.c (store_field): In the bitfield case, if the value comes from
38611 a function call and is of an aggregate type returned in registers, do
38612 not modify the field mode; extract the value in all cases if the mode
38613 is BLKmode and the size is not larger than a word.
38615 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
38618 * config/i386/cpuid.h: Fix undefined behavior.
38620 2017-01-04 Jeff Law <law@redhat.com>
38622 PR tree-optimization/79007
38623 PR tree-optimization/67955
38624 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
38625 conservative for pt.null when flag_non_call_exceptions is on.
38627 2017-01-09 Jakub Jelinek <jakub@redhat.com>
38629 PR translation/79019
38630 PR translation/79020
38631 * params.def (PARAM_INLINE_MIN_SPEEDUP,
38632 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
38633 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
38635 * config/avr/avr.opt (maccumulate-args): Likewise.
38636 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
38637 * common.opt (freport-bug): Likewise.
38638 * cif-code.def (CIF_FINAL_ERROR): Likewise.
38639 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
38640 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
38641 translatable string.
38642 * config/i386/i386.c (function_value_32): Likewise.
38643 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
38644 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
38646 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
38647 * common/config/msp430/msp430-common.c (msp430_handle_option):
38649 * symtab.c (symtab_node::verify_base): Likewise.
38650 * opts.c (set_debug_level): Likewise.
38651 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
38652 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
38653 missing whitespace to translatable strings.
38654 * config/avr/avr.md (bswapsi2): Fix typo in comment.
38655 * config/sh/superh.h: Likewise.
38656 * config/i386/xopintrin.h: Likewise.
38657 * config/i386/znver1.md: Likewise.
38658 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
38659 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
38660 * double-int.h (struct double_int): Likewise.
38661 * double-int.c (div_and_round_double): Likewise.
38662 * wide-int.cc: Likewise.
38663 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
38664 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
38665 * cfgcleanup.c (crossjumps_occured): Renamed to ...
38666 (crossjumps_occurred): ... this.
38667 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
38670 PR tree-optimization/78899
38671 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
38672 returning bool return struct loop *, NULL for failure and the new
38674 (versionable_outer_loop_p): Don't version outer loop if it has
38675 dont_vectorized bit set.
38676 (tree_if_conversion): When versioning outer loop, ensure
38677 tree_if_conversion is performed also on the inner loop of the
38678 non-vectorizable outer loop copy.
38679 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
38680 LOOP_VECTORIZED in inner loop of the scalar outer loop and
38681 prevent vectorization of it.
38682 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
38683 the outer loop vectorization of the non-scalar version is attempted
38684 before vectorization of the inner loop in scalar version. If
38685 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
38686 vectorization of its inner loop.
38687 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
38688 has 2 inner loops, rename also on edges from bb whose single pred
38689 is outer_loop->header. Fix typo in function comment.
38691 2017-01-09 Martin Sebor <msebor@redhat.com>
38694 * asan.c (asan_emit_stack_protection): Increase local buffer size
38695 to avoid snprintf truncation warning.
38697 2017-01-09 Andrew Pinski <apinski@cavium.com>
38699 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
38700 to reference thunderx2t99 for the tuning structure
38701 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
38703 (thunderx2t99_extra_costs): This.
38704 * config/aarch64/aarch64-tune.md: Regenerate.
38705 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
38706 (vulcan_addrcost_table): This.
38707 (vulcan_regmove_cost): Rename to ...
38708 (thunderx2t99_regmove_cost): This.
38709 (vulcan_vector_cost): Rename to ...
38710 (thunderx2t99_vector_cost): this.
38711 (vulcan_branch_cost): Rename to ...
38712 (thunderx2t99_branch_cost): This.
38713 (vulcan_tunings): Rename to ...
38714 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
38715 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
38717 2017-01-09 Martin Jambor <mjambor@suse.cz>
38721 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
38722 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
38723 (propagate_vr_accross_jump_function): Use the above function for all
38724 value range computations for pass-through jump functions and type
38725 converasion from explicit value range values.
38726 (ipcp_propagate_stage): Do not attempt to deduce types of formal
38727 parameters from TYPE_ARG_TYPES.
38728 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
38729 (ipa_write_node_info): Stream type of the actual argument.
38730 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
38732 2017-01-09 Martin Liska <mliska@suse.cz>
38735 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
38736 (lookup_compiler): Do not show error message with have_E.
38738 2017-01-09 Jakub Jelinek <jakub@redhat.com>
38740 PR tree-optimization/78938
38741 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
38742 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
38743 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
38744 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
38747 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38749 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
38752 2017-01-09 Richard Biener <rguenther@suse.de>
38754 PR tree-optimization/78997
38755 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
38756 name condition properly.
38758 2017-01-09 Richard Biener <rguenther@suse.de>
38761 * dwarf2out.c (is_cxx): New overload with context.
38762 (is_naming_typedef_decl): Use it.
38764 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38766 * invoke.texi (Option Summary): Correct spacing in option lists
38767 and add line breaks to fix over-long lines.
38769 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38771 PR middle-end/17660
38773 * extend.texi (Common Variable Attributes): Add xref to GCC
38774 Internals manual to explain mode attribute keywords.
38776 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38779 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
38780 and Preprocessor Options.
38781 (Options for Linking): Document -pthread here....
38782 (RS/6000 and PowerPC Options): ...not here.
38783 (Solaris 2 Options): ...or here.
38784 * doc/cppopts.texi: Document -pthread.
38786 2017-01-08 Martin Sebor <msebor@redhat.com>
38788 PR middle-end/77708
38789 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
38790 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
38791 New member functions.
38792 (format_directive): Used them.
38794 (pass_sprintf_length::handle_gimple_call): Same.
38795 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
38796 to avoid truncation for any argument.
38797 (extract_affine_mul): Same.
38798 * tree.c (get_file_function_name): Same.
38800 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38802 PR middle-end/77484
38803 * predict.def (PRED_INDIR_CALL): Set to 86.
38805 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38807 PR preprocessor/54124
38808 * doc/cppopts.texi: Reformat -d subtable to list the full name
38809 of the options. Add cross-reference to the docs for the general
38810 compiler -d options.
38811 * doc/invoke.texi (Developer Options): Add cross-reference to the
38812 preprocessor-specific -d option documentation.
38814 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38816 PR preprocessor/13498
38817 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
38818 redudant material, and reflect new command-line options.
38819 (System Headers): Likewise.
38821 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38823 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
38824 -isystem, and -idirafter. Copy-edit.
38825 * doc/cppopts.texi: Copy-edit. Remove contradiction about
38826 default for -ftrack-macro-expansion. Delete obsolete and
38827 badly-formatted implementation details about -fdebug-cpp output.
38828 * doc/cppwarnopts.texi: Copy-edit.
38830 2017-01-07 David Malcolm <dmalcolm@redhat.com>
38833 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
38834 that the transition from a max line width >= 1<<10 to narrower
38835 lines works correctly.
38837 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
38839 * doc/options.texi (PerFunction): New.
38840 * opt-functions.awk (switch_flags): Map both Optimization and
38841 PerFunction to CL_OPTIMIZATION.
38842 * opth-gen.awk: Test for PerFunction flag along with
38844 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
38845 it only when the latter is present. Skip those that don't in
38846 the hash function generator.
38847 * common.opt (fvar-tracking): Mark as PerFunction instead of
38849 (fvar-tracking-assignments): Likewise.
38850 (fvar-tracking-assignments-toggle): Likewise.
38851 (fvar-tracking-uninit): Likewise.
38853 2017-01-07 Jakub Jelinek <jakub@redhat.com>
38855 PR translation/79018
38856 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
38859 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
38862 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
38863 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
38864 TARGET_LONG_JUMP_TABLE_OFFSETS.
38865 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
38866 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
38867 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
38868 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
38869 * config/m68k/m68k.md (tablejump expander): Likewise.
38870 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
38871 TARGET_LONG_JUMP_TABLE_OFFSETS.
38872 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
38873 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
38875 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
38876 David Holsgrove <david.holsgrove@xilinx.com>
38878 * common/config/microblaze/microblaze-common.c
38879 (TARGET_EXCEPT_UNWIND_INFO): Remove.
38880 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
38882 * config/microblaze/microblaze.c (microblaze_must_save_register)
38883 (microblaze_expand_epilogue, microblaze_return_addr): Handle
38885 (microblaze_eh_return): New function.
38886 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
38887 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
38888 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
38889 * config/microblaze/microblaze.md (eh_return): New pattern.
38891 2017-01-06 Jakub Jelinek <jakub@redhat.com>
38893 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
38894 GCC_DIAGNOSTIC_STRINGIFY): Define.
38896 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
38898 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38900 * config/arm/arm.md (<mcrr>): New.
38902 * config/arm/arm.c (arm_arch5te): New.
38903 (arm_option_override): Set arm_arch5te.
38904 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
38906 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
38907 (arm_mcrr_qualifiers): ... this. New.
38908 (MRRC_QUALIFIERS): Define to...
38909 (arm_mrrc_qualifiers): ... this. New.
38910 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
38912 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
38913 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
38914 (MRRCI, mrrc, MRRC): New.
38915 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
38916 VUNSPEC_MRRC2): New.
38918 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38920 * config/arm/arm.md (<mcr>): New.
38922 * config/arm/arm.c (arm_coproc_builtin_available): Add
38923 support for mcr, mrc, mcr2 and mrc2.
38924 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
38925 (arm_mcr_qualifiers): ... this. New.
38926 (MRC_QUALIFIERS): Define to ...
38927 (arm_mrc_qualifiers): ... this. New.
38928 (MCR_QUALIFIERS): Define to ...
38929 (arm_mcr_qualifiers): ... this. New.
38930 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
38932 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
38933 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
38934 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
38935 VUNSPEC_MRC2): New.
38937 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38939 * config/arm/arm.md (*ldc): New.
38943 * config/arm/arm.c (arm_coproc_builtin_available): Add
38944 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
38945 (arm_coproc_ldc_stc_legitimate_address): New.
38946 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
38947 'qualifier_const_pointer'.
38948 (LDC_QUALIFIERS): Define to...
38949 (arm_ldc_qualifiers): ... this. New.
38950 (STC_QUALIFIERS): Define to...
38951 (arm_stc_qualifiers): ... this. New.
38952 * config/arm/arm-protos.h
38953 (arm_coproc_ldc_stc_legitimate_address): New.
38954 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
38955 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
38956 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
38957 stc2, stcl, stc2l): New.
38958 * config/arm/constraints.md (Uz): New.
38959 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
38960 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
38961 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
38962 VUNSPEC_STC2L): New.
38964 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38966 * config/arm/arm.md (<cdp>): New.
38967 * config/arm/arm.c (neon_const_bounds): Rename this ...
38968 (arm_const_bounds): ... this.
38969 (arm_coproc_builtin_available): New.
38970 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
38971 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
38972 (CDP_QUALIFIERS): Define to...
38973 (arm_cdp_qualifiers): ... this. New.
38975 (arm_expand_builtin_args): Add case for 6 arguments.
38976 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
38977 (arm_const_bounds): ... this.
38978 (arm_coproc_builtin_available): New.
38979 * config/arm/arm_acle.h (__arm_cdp): New.
38981 * config/arm/arm_acle_builtins.def (cdp): New.
38983 * config/arm/iterators.md (CDPI,CDP,cdp): New.
38984 * config/arm/neon.md: Rename all 'neon_const_bounds' to
38985 'arm_const_bounds'.
38986 * config/arm/types.md (coproc): New.
38987 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
38988 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
38989 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
38990 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
38992 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38994 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
38995 (UBINOP_QUALIFIERS): New.
38997 (acle_builtin_data): New. Change comment.
38998 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
38999 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
39000 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
39001 arm_acle_builtins.def.
39002 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
39003 (arm_init_acle_builtins): New.
39004 (CRC32_BUILTIN): Remove.
39005 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
39006 crc32cb, crc32ch and crc32cw.
39007 (arm_init_crc32_builtins): Remove.
39008 (arm_init_builtins): Use arm_init_acle_builtins rather
39009 than arm_init_crc32_builtins.
39010 (arm_expand_acle_builtin): New.
39011 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
39012 * config/arm/arm_acle_builtins.def: New.
39014 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39016 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
39017 (arm_builtin_datum): ... this.
39018 (arm_init_neon_builtin): Rename to ...
39019 (arm_init_builtin): ... this. Add a new parameters PREFIX
39020 and USE_SIG_IN_NAME.
39021 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
39022 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
39023 'arm_builtin_datum'.
39024 (arm_init_vfp_builtins): Likewise.
39025 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
39026 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
39027 (arm_expand_neon_args): Rename to ...
39028 (arm_expand_builtin_args): ... this. Rename builtin_arg
39029 enum values and differentiate between ARG_BUILTIN_MEMORY
39030 and ARG_BUILTIN_NEON_MEMORY.
39031 (arm_expand_neon_builtin_1): Rename to ...
39032 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
39033 values, arm_expand_builtin_args and add bool parameter NEON.
39034 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
39035 (arm_expand_vfp_builtin): Likewise.
39036 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
39038 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39040 PR middle-end/77484
39041 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
39042 * predict.c (tree_estimate_probability_bb): Reverse direction of
39043 polymorphic call predictor.
39045 2017-01-06 David Malcolm <dmalcolm@redhat.com>
39047 * passes.c (execute_one_pass): Split out pass-skipping logic into...
39048 (determine_pass_name_match): ...this new function and...
39049 (should_skip_pass_p): ...this new function.
39051 2017-01-06 Nathan Sidwell <nathan@acm.org>
39053 * ipa-visibility.c (function_and_variable_visibility): Reformat
39054 comments and long lines. Remove extrneous if.
39055 * symtab.c (symtab_node::make_decl_local): Fix code format.
39056 (symtab_node::set_section_for_node): Fix comment typo.
39058 2017-01-06 Martin Liska <mliska@suse.cz>
39061 * lra-constraints.c: Rename invariant to lra_invariant.
39062 * predict.c (set_even_probabilities): Initialize e to NULL.
39064 2017-01-05 Martin Sebor <msebor@redhat.com>
39066 PR tree-optimization/78910
39067 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
39068 (format_integer): Correct off-by-one error in the handling
39069 of precision with negative numbers in signed conversions..
39071 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
39073 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
39075 2017-01-05 Jakub Jelinek <jakub@redhat.com>
39077 PR tree-optimization/71016
39078 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
39079 factor_out_conditional_conversion. Formatting fix.
39080 (factor_out_conditional_conversion): Add cond_stmt argument.
39081 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
39082 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
39085 2017-01-05 David Malcolm <dmalcolm@redhat.com>
39087 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
39088 read-rtl-function.o, and selftest-rtl.o.
39089 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
39090 (selftest::aarch64_test_loading_full_dump): New function.
39091 (selftest::aarch64_run_selftests): New function.
39092 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
39093 selftest::aarch64_run_selftests.
39094 * config/i386/i386.c
39095 (selftest::ix86_test_loading_dump_fragment_1): New function.
39096 (selftest::ix86_test_loading_call_insn): New function.
39097 (selftest::ix86_test_loading_full_dump): New function.
39098 (selftest::ix86_test_loading_unspec): New function.
39099 (selftest::ix86_run_selftests): Call the new functions.
39100 * emit-rtl.c (maybe_set_max_label_num): New function.
39101 * emit-rtl.h (maybe_set_max_label_num): New decl.
39102 * function.c (instantiate_decls): Guard call to
39103 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
39104 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
39106 * gensupport.c (gen_reader::gen_reader): Pass "false"
39107 for new "compact" param of rtx_reader.
39108 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
39109 rather than an empty string for NULL strings.
39110 * read-md.c: Potentially include config.h rather than bconfig.h.
39111 Wrap include of errors.h with #ifdef GENERATOR_FILE.
39112 (have_error): New global, copied from errors.c.
39113 (md_reader::read_name): Rename to...
39114 (md_reader::read_name_1): ...this, adding "out_loc" param,
39115 and converting "missing name or number" to returning false, rather
39117 (md_reader::read_name): Reimplement in terms of read_name_1.
39118 (md_reader::read_name_or_nil): New function.
39119 (md_reader::read_string): Handle "(nil)" by returning NULL.
39120 (md_reader::md_reader): Add new param "compact".
39121 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
39122 (md_reader::read_file): New method.
39123 * read-md.h (md_reader::md_reader): Add new param "compact".
39124 (md_reader::read_file): New method.
39125 (md_reader::is_compact): New accessor.
39126 (md_reader::read_name): Convert return type from void to file_location.
39127 (md_reader::read_name_or_nil): New decl.
39128 (md_reader::read_name_1): New decl.
39129 (md_reader::m_compact): New field.
39130 (noop_reader::noop_reader): Pass "false" for new "compact" param
39132 (rtx_reader::rtx_reader): Add new "compact" param.
39133 (rtx_reader::read_rtx_operand): Make virtual and convert return
39134 type from void to rtx.
39135 (rtx_reader::read_until): New decl.
39136 (rtx_reader::handle_any_trailing_information): New virtual function.
39137 (rtx_reader::postprocess): New virtual function.
39138 (rtx_reader::finalize_string): New virtual function.
39139 (rtx_reader::m_in_call_function_usage): New field.
39140 (rtx_reader::m_reuse_rtx_by_id): New field.
39141 * read-rtl-function.c: New file.
39142 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
39143 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
39144 (selftest::verify_three_block_rtl_cfg): New decl.
39145 * read-rtl-function.h: New file.
39146 * read-rtl.c: Potentially include config.h rather than bconfig.h.
39147 For host, include function.h, memmodel.h, and emit-rtl.h.
39148 (one_time_initialization): New function.
39149 (struct compact_insn_name): New struct.
39150 (compact_insn_names): New array.
39151 (find_code): Handle insn codes in compact dumps.
39152 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
39153 (bind_subst_iter_and_attr): Likewise.
39154 (add_condition_to_string): Likewise.
39155 (add_condition_to_rtx): Likewise.
39156 (apply_attribute_uses): Likewise.
39157 (add_current_iterators): Likewise.
39158 (apply_iterators): Likewise.
39159 (initialize_iterators): Guard usage of apply_subst_iterator with
39160 #ifdef GENERATOR_FILE.
39161 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
39162 (md_reader::read_mapping): Likewise.
39163 (add_define_attr_for_define_subst): Likewise.
39164 (add_define_subst_attr): Likewise.
39165 (read_subst_mapping): Likewise.
39166 (check_code_iterator): Likewise.
39167 (rtx_reader::read_rtx): Likewise. Move one-time initialization
39169 (one_time_initialization): New function.
39170 (rtx_reader::read_until): New method.
39171 (read_flags): New function.
39172 (parse_reg_note_name): New function.
39173 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
39174 Handle reuse_rtx ids.
39175 Wrap iterator lookup within #ifdef GENERATOR_FILE.
39176 Add parsing support for RTL dumps, mirroring the special-cases in
39177 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
39178 values, and calling handle_any_trailing_information.
39179 (rtx_reader::read_rtx_operand): Convert return type from void
39180 to rtx, returning return_rtx. Handle case 'e'. Call
39181 finalize_string on XSTR and XTMPL fields.
39182 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
39183 "(nil)" values were omitted. Call the postprocess vfunc on the
39185 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
39186 class ctor. Initialize m_in_call_function_usage. Call
39187 one_time_initialization.
39188 * rtl-tests.c (selftest::test_uncond_jump): Call
39189 set_new_first_and_last_insn.
39190 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
39191 * selftest-rtl.c: New file.
39192 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
39193 (selftest::get_insn_by_uid): New decl.
39194 * selftest-run-tests.c (selftest::run_tests): Call
39195 read_rtl_function_c_tests.
39196 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
39197 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
39200 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
39202 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
39203 operands in a special way. Assert that pos+len <= mode precision.
39205 2017-01-05 Jakub Jelinek <jakub@redhat.com>
39207 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
39208 3 argument Alias with unlimited for the negative form.
39209 (fno-vect-cost-model): Removed.
39211 2017-01-05 Martin Liska <mliska@suse.cz>
39213 * hsa-gen.c (gen_hsa_divmod): New function.
39214 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
39216 2017-01-05 Martin Liska <mliska@suse.cz>
39219 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
39222 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39224 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
39225 small constant length operands.
39227 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39229 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
39230 between loop iterations.
39232 2017-01-05 Martin Liska <mliska@suse.cz>
39235 * gimplify.c (gimplify_decl_expr): Compare to
39236 asan_poisoned_variables instread of checking flags.
39237 (gimplify_target_expr): Likewise.
39238 (gimplify_expr): Likewise.
39239 (gimplify_function_tree): Conditionally initialize
39240 asan_poisoned_variables.
39242 2017-01-04 Jeff Law <law@redhat.com>
39244 PR tree-optimizatin/78812
39245 * rtl.h (contains_mem_rtx_p): Prototype.
39246 * ifcvt.c (containts_mem_rtx_p): Move from here to...
39247 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
39248 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
39249 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
39250 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
39252 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39254 * input.c (assert_char_at_range): Default-initialize actual_range.
39256 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39258 * df-scan.c (df_ref_create_structure): Make regno unsigned,
39259 to match the caller.
39261 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39263 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
39264 insns after final jump in test to emit dummy move.
39266 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39268 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
39269 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
39271 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39273 * multiple_target.c (create_dispatcher_calls): Init e_next.
39274 * tree-ssa-loop-split.c (split_loop): Init border.
39275 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
39278 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
39283 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
39284 (altivec_register_operand): Do not return true if the operand
39285 contains a SUBREG mixing SImode and SFmode.
39286 (vsx_register_operand): Likewise.
39287 (vsx_reg_sfsubreg_ok): New predicate.
39288 (vfloat_operand): Do not return true if the operand contains a
39289 SUBREG mixing SImode and SFmode.
39290 (vint_operand): Likewise.
39291 (vlogical_operand): Likewise.
39292 (gpc_reg_operand): Likewise.
39293 (int_reg_operand): Likewise.
39294 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
39295 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
39296 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
39298 (rs6000_emit_move_si_sf_subreg): New helper function.
39299 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
39300 fixup SUBREGs involving SImode and SFmode.
39301 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
39302 numbers for the new peephole2 optimization.
39303 (peephole2 for SFmode unions): New peephole2 to optimize cases in
39304 the GLIBC math library that do AND/IOR/XOR operations on single
39305 precision floating point.
39306 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
39307 target macros to say whether we need to avoid SUBREGs mixing
39309 (TARGET_ALLOW_SF_SUBREG): Likewise.
39310 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
39311 (UNSPEC_SI_FROM_SF): Likewise.
39312 (iorxor): Change spacing.
39313 (and_ior_xor): New iterator for AND, IOR, and XOR.
39314 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
39315 (movdi_from_sf_zero_ext): Likewise.
39316 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
39317 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
39318 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
39319 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
39320 (fms<mode>4): Likewise.
39321 (fnma<mode>4): Likewise.
39322 (fnms<mode>4): Likewise.
39323 (nfma<mode>4): Likewise.
39324 (nfms<mode>4): Likewise.
39326 2017-01-04 Marek Polacek <polacek@redhat.com>
39329 * doc/invoke.texi: Document -Wpointer-compare.
39331 2017-01-04 Jakub Jelinek <jakub@redhat.com>
39333 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
39336 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
39337 descriptions for -gdwarf-5 and emit them as uleb128 instead of
39340 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
39343 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
39344 documentation of the powerpc_popcntb_ok attribute.
39345 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
39346 code to issue warning messages if a requested CPU configuration is
39347 not supported by the binary (assembler and loader) toolchain.
39348 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
39349 made to define a built-in function that has been disabled.
39350 (paired_init_builtins): Add assertion to prevent ICE if attempt is
39351 made to define a built-in function that has been disabled.
39352 (altivec_init_builtins): Add comment explaining why definition
39353 of the DST built-in functions is not preceded by an assertion
39354 check. Add assertions to prevent ICE if attempts are made to
39355 define an altivec predicate or an abs* built-in function that has
39357 (htm_init_builtins): Add comment explaining why definition of the
39358 htm built-in functions is not preceded by an assertion check.
39360 2017-01-04 Jeff Law <law@redhat.com>
39362 PR tree-optimizatin/67955
39363 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
39364 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
39365 the points-to solution does not include pt_null. Use DECL_PT_UID
39368 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
39370 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
39371 Use gen_int_mode instead of gen_lopwart for const_int operands.
39373 2017-01-04 Jakub Jelinek <jakub@redhat.com>
39375 PR tree-optimization/71563
39376 * match.pd: Simplify X << Y into X if Y is known to be 0 or
39377 out of range value - has low bits known to be zero.
39379 2017-01-04 Alan Modra <amodra@gmail.com>
39381 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
39382 * configure: Regenerate.
39383 * config.in: Regenerate.
39385 2017-01-04 Jakub Jelinek <jakub@redhat.com>
39388 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
39389 a substring of the message, but strcmp with the whole message. Ifdef
39390 ENABLE_NLS, translate the message first using dgettext.
39392 2017-01-03 Jeff Law <law@redhat.com>
39394 PR tree-optimizatin/78856
39395 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
39396 (mark_threaded_blocks): Remove code to truncate thread paths that
39397 cross multiple loop headers. Instead invalidate the cached loop
39398 iteration information and handle case of a thread path walking
39399 into an irreducible region.
39401 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
39404 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
39405 assertions. Add support for doing the signbit if the IEEE 128-bit
39406 floating point value is in a GPR.
39407 * config/rs6000/rs6000.md (Fsignbit): Delete.
39408 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
39409 Update the length attribute if the value is in a GPR.
39410 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
39411 the sign or zero extension instruction, since the value is always 0/1.
39412 (signbit<mode>2_dm2): Delete using <Fsignbit>.
39415 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
39416 extracting SImode to a GPR register so that we can generate a
39417 store, limit the vector to be in a traditional Altivec register
39418 for the vextuwrx instruction.
39420 2017-01-03 Ian Lance Taylor <iant@google.com>
39422 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
39424 2017-01-03 Martin Sebor <msebor@redhat.com>
39426 PR tree-optimization/78696
39427 * gimple-ssa-sprintf.c (format_floating): Correct handling of
39428 precision. Use MPFR for %f for greater fidelity. Correct handling
39430 (pass_sprintf_length::compute_format_length): Set width and precision
39431 specified by asrerisk to void_node for vararg functions.
39432 (try_substitute_return_value): Adjust dump output.
39434 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
39436 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
39438 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
39440 * doc/invoke.texi (SPARC options): Document -mlra as the default.
39441 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
39442 -mlra/-mno-lra was passed to the compiler.
39444 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
39446 PR rtl-optimization/65618
39447 * emit-rtl.c (try_split): Move initialization of "before" and
39448 "after" to just before the call to emit_insn_after_setloc.
39450 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
39452 * doc/md.texi (Standard Names): Remove reference to Java frontend.
39454 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
39456 * dwarf2out.c (gen_enumeration_type_die): When
39457 -gno-strict-dwarf, add a DW_AT_encoding attribute.
39459 2017-01-03 Jakub Jelinek <jakub@redhat.com>
39461 PR tree-optimization/78965
39462 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
39463 Change first argument from const call_info & to call_info &. For %n
39464 set info.nowrite to false.
39466 PR middle-end/78901
39467 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
39468 possibly throwing calls.
39470 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
39471 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
39472 and fns handling, rather than in a separate case SSA_NAME.
39474 2017-01-02 Jeff Law <law@redhat.com>
39476 * config/darwin-driver.c (darwin_driver_init): Const-correctness
39477 fixes for first_period and second_period variables.
39479 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
39482 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
39483 (*insvqi_1): New insn pattern.
39484 (*insvqi_1_mem_rex64): Ditto.
39485 (*insvqi_2): Ditto.
39486 (*insvqi_3): Rename from *insvqi.
39488 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
39490 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
39492 * doc/cfg.texi (Edges): Remove reference to Java.
39493 (Maintaining the CFG): Ditto.
39495 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39497 PR middle-end/77674
39498 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
39499 transparent aliases.
39501 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39503 PR middle-end/77484
39504 * predict.def (PRED_CALL): Update hitrate.
39505 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
39506 * predict.c (tree_estimate_probability_bb): Split CALL predictor
39507 into direct/indirect/polymorphic variants.
39509 2017-01-01 Jakub Jelinek <jakub@redhat.com>
39511 Update copyright years.
39513 * gcc.c (process_command): Update copyright notice dates.
39514 * gcov-dump.c (print_version): Ditto.
39515 * gcov.c (print_version): Ditto.
39516 * gcov-tool.c (print_version): Ditto.
39517 * gengtype.c (create_file): Ditto.
39518 * doc/cpp.texi: Bump @copying's copyright year.
39519 * doc/cppinternals.texi: Ditto.
39520 * doc/gcc.texi: Ditto.
39521 * doc/gccint.texi: Ditto.
39522 * doc/gcov.texi: Ditto.
39523 * doc/install.texi: Ditto.
39524 * doc/invoke.texi: Ditto.
39526 Copyright (C) 2017 Free Software Foundation, Inc.
39528 Copying and distribution of this file, with or without modification,
39529 are permitted in any medium without royalty provided the copyright
39530 notice and this notice are preserved.