1 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4 * config/s390/s390.c (s390_option_override): Avoid strncpy.
6 2017-12-18 Richard Biener <rguenther@suse.de>
8 PR tree-optimization/81877
9 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
10 (outermost_indep_loop): Adjust.
11 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
12 (can_sm_ref_p): Adjust.
14 2017-12-18 Richard Biener <rguenther@suse.de>
17 * tree.c (array_at_struct_end_p): Return true if the underlying
18 object has space for at least one element in excess of what
19 the array domain specifies.
21 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
23 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
24 example to avoid overfull hbox.
25 * doc/invoke.texi (Option Summary): Add missing @gol.
26 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
29 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
30 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
33 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
34 (AArch64 Options): Document -mverbose-cost-dump.
35 (ARM Options): Likewise, plus -mflip-thumb.
37 2017-12-17 Martin Sebor <msebor@redhat.com>
40 * gimple-ssa-warn-restrict.c
41 (builtin_memref::offset_out_of_bounds): Correct the handling of
44 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
46 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
48 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
50 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
52 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
54 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
57 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
59 * doc/invoke.texi: Fix some typos.
61 2017-12-16 Martin Sebor <msebor@redhat.com>
63 PR tree-optimization/78918
64 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
65 * builtins.c (check_sizes): Rename...
66 (check_access): ...to this. Rename function arguments for clarity.
67 (check_memop_sizes): Adjust names.
68 (expand_builtin_memchr, expand_builtin_memcpy): Same.
69 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
70 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
71 (check_strncat_sizes, expand_builtin_strncat): Same.
72 (expand_builtin_strncpy, expand_builtin_memset): Same.
73 (expand_builtin_bzero, expand_builtin_memcmp): Same.
74 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
75 (maybe_emit_sprintf_chk_warning): Same.
76 (expand_builtin_strcpy): Adjust.
77 (expand_builtin_stpcpy): Same.
78 (expand_builtin_with_bounds): Detect out-of-bounds accesses
79 in pointer-checking forms of memcpy, memmove, and mempcpy.
80 (gcall_to_tree_minimal, max_object_size): Define new functions.
81 * builtins.h (max_object_size): Declare.
82 * calls.c (alloc_max_size): Call max_object_size instead of
83 hardcoding ssizetype limit.
84 (get_size_range): Handle new argument.
85 * calls.h (get_size_range): Add a new argument.
86 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
87 * doc/invoke.texi (-Wrestrict): Adjust, add example.
88 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
90 (gimple_fold_builtin_memory_chk): Same.
91 (gimple_fold_builtin_stxcpy_chk): New function.
92 * gimple-ssa-warn-restrict.c: New source.
93 * gimple-ssa-warn-restrict.h: New header.
94 * gimple.c (gimple_build_call_from_tree): Propagate location.
95 * passes.def (pass_warn_restrict): Add new pass.
96 * tree-pass.h (make_pass_warn_restrict): Declare.
97 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
99 (handle_builtin_strcat): Same.
100 (strlen_optimize_stmt): Rename...
101 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
102 stpncpy, strncpy, and their checking forms.
104 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
106 PR rtl-optimization/82849
107 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
108 and get_max_loop_iterations_int.
110 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
111 Alan Hayward <alan.hayward@arm.com>
112 David Sherwood <david.sherwood@arm.com>
114 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
115 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
116 size as a poly_uint64.
117 (mode_for_vector, mode_for_int_vector): Take the number of vector
118 elements as a poly_uint64.
119 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
120 size as a poly_uint64.
121 (mode_for_vector, mode_for_int_vector): Take the number of vector
122 elements as a poly_uint64.
124 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
125 Alan Hayward <alan.hayward@arm.com>
126 David Sherwood <david.sherwood@arm.com>
128 * machmode.h (MACRO_MODE): New macro.
129 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
130 * targhooks.c (default_libcall_value, default_secondary_reload)
131 (default_memory_move_cost, default_register_move_cost)
132 (default_class_max_nregs): Likewise.
134 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
135 Alan Hayward <alan.hayward@arm.com>
136 David Sherwood <david.sherwood@arm.com>
138 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
139 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
141 * genautomata.c (main): Likewise.
142 * genconditions.c (write_header): Likewise.
143 * genemit.c (main): Likewise.
144 * genextract.c (print_header): Likewise.
145 * genopinit.c (main): Likewise.
146 * genoutput.c (output_prologue): Likewise.
147 * genpeep.c (main): Likewise.
148 * genpreds.c (write_insn_preds_c): Likewise.
149 * genrecog.c (writer_header): Likewise.
150 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
151 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
152 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
153 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
154 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
155 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
156 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
157 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
158 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
159 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
160 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
161 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
162 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
163 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
164 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
165 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
166 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
167 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
168 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
169 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
170 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
171 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
172 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
173 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
174 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
175 * config/darwin.c (IN_TARGET_CODE): Likewise.
176 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
177 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
178 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
179 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
180 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
181 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
182 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
183 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
184 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
185 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
186 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
187 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
188 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
189 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
190 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
191 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
192 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
193 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
194 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
195 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
196 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
197 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
198 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
199 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
200 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
201 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
202 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
203 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
204 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
205 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
206 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
207 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
208 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
209 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
210 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
211 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
212 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
213 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
214 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
215 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
216 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
217 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
218 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
219 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
220 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
221 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
222 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
223 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
224 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
225 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
226 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
227 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
228 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
229 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
230 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
231 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
232 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
233 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
234 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
235 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
236 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
237 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
238 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
239 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
240 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
241 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
242 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
243 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
244 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
245 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
246 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
247 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
248 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
249 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
250 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
251 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
252 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
253 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
254 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
255 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
256 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
257 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
258 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
259 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
260 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
261 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
262 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
263 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
264 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
265 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
266 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
267 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
268 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
269 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
270 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
271 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
272 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
273 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
274 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
275 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
276 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
277 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
278 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
279 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
280 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
281 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
283 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
284 Alan Hayward <alan.hayward@arm.com>
285 David Sherwood <david.sherwood@arm.com>
287 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
290 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
291 Alan Hayward <alan.hayward@arm.com>
292 David Sherwood <david.sherwood@arm.com>
294 * doc/generic.texi (VEC_SERIES_EXPR): Document.
295 * doc/md.texi (vec_series@var{m}): Document.
296 * tree.def (VEC_SERIES_EXPR): New tree code.
297 * tree.h (build_vec_series): Declare.
298 * tree.c (build_vec_series): New function.
299 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
300 * tree-pretty-print.c (dump_generic_node): Likewise.
301 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
302 * tree-inline.c (estimate_operator_cost): Likewise.
303 * expr.c (expand_expr_real_2): Likewise.
304 * optabs-tree.c (optab_for_tree_code): Likewise.
305 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
306 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
307 * expmed.c (make_tree): Handle VEC_SERIES.
308 * optabs.def (vec_series_optab): New optab.
309 * optabs.h (expand_vec_series_expr): Declare.
310 * optabs.c (expand_vec_series_expr): New function.
311 * tree-vect-generic.c (expand_vector_operations_1): Check that
312 the operands also have vector type.
314 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
315 Alan Hayward <alan.hawyard@arm.com>
316 David Sherwood <david.sherwood@arm.com>
318 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
319 (VEC_COND_EXPR): Add missing @tindex.
320 * doc/md.texi (vec_duplicate@var{m}): Document.
321 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
322 * tree.c (build_vector_from_val): Add stubbed-out handling of
323 variable-length vectors, using VEC_DUPLICATE_EXPR.
324 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
325 * cfgexpand.c (expand_debug_expr): Likewise.
326 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
327 * tree-inline.c (estimate_operator_cost): Likewise.
328 * tree-pretty-print.c (dump_generic_node): Likewise.
329 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
330 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
331 (test_vec_duplicate_folding): New function.
332 (fold_const_c_tests): Call it.
333 * optabs.def (vec_duplicate_optab): New optab.
334 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
335 * optabs.h (expand_vector_broadcast): Declare.
336 * optabs.c (expand_vector_broadcast): Make non-static. Try using
338 * expr.c (store_constructor): Try using vec_duplicate_optab for
340 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
342 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
345 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
346 div/mod latencies a bit.
348 2017-12-15 Jeff Law <law@redhat.com>
350 PR tree-optimization/36550
351 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
352 (mark_threaded_blocks): Rewrite code to avoid block copying when
353 optimizing for size. Don't pessimize blocks which will be
354 copied, but all the statements will be dead.
356 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
358 PR tree-optimization/81165
359 * tree-ssa-threadupdate.c (uses_in_bb): New.
360 (estimate_threading_killed_stmts): New.
361 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
362 * tree-ssa-threadedge.c
363 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
366 2017-12-15 Jeff Law <law@redhat.com>
368 PR tree-optimization/83410
369 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
370 threads when parallelizing loops.
372 2017-12-15 Jakub Jelinek <jakub@redhat.com>
374 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
376 * config/alpha/alpha.c (vms_attribute_table): Swap
377 affects_type_identity and handler fields, adjust comments.
378 * config/mips/mips.c (mips_attribute_table): Likewise.
379 * config/visium/visium.c (visium_attribute_table): Likewise.
380 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
381 * config/microblaze/microblaze.c (microblaze_attribute_table):
383 * config/spu/spu.c (spu_attribute_table): Likewise.
384 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
385 * config/arc/arc.c (arc_attribute_table): Likewise.
386 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
387 * config/v850/v850.c (v850_handle_interrupt_attribute,
388 v850_handle_data_area_attribute): Formatting fixes.
389 (v850_attribute_table): Swap affects_type_identity and handler
390 fields, adjust comments.
391 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
392 * config/arm/arm.c (arm_attribute_table): Likewise.
393 * config/avr/avr.c (avr_attribute_table): Likewise.
394 * config/s390/s390.c (s390_attribute_table): Likewise.
395 * config/sh/sh.c (sh_attribute_table): Likewise.
396 * config/i386/i386.c (ix86_handle_cconv_attribute,
397 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
398 (ix86_attribute_table): Swap affects_type_identity and handler
399 fields, adjust comments.
400 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
401 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
402 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
403 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
404 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
405 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
406 * config/rx/rx.c (rx_attribute_table): Likewise.
407 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
408 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
409 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
410 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
411 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
412 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
413 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
414 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
415 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
416 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
417 * doc/plugins.texi (user_attr): Likewise. Add NULL for
419 * attribs.c (empty_attribute_table): Swap affects_type_identity and
421 (register_scoped_attributes, decl_attributes): Formatting fixes.
423 PR tree-optimization/83269
424 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
425 subtraction in arg0's type if type is signed and arg0 is unsigned.
429 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
430 convert? on @0 instead of convert. Check type of @1, not @0.
431 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
432 convert? on @0 instead of convert. Check type of @1, not @0.
433 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
434 only optimize if either both @1 and @2 types are narrower
435 precision, or both are wider or equal precision, and in the former
436 case only if both have undefined overflow.
438 2017-12-15 Richard Biener <rguenther@suse.de>
441 * internal-fn.def (IFN_NOP): Add.
442 * internal-fn.c (expand_NOP): Do nothing.
443 * lto-streamer-in.c (input_function): Instead of removing
444 sanitizer calls replace them with IFN_NOP calls.
446 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
447 Alan Hayward <alan.hayward@arm.com>
448 David Sherwood <david.sherwood@arm.com>
450 * dse.c (store_info, read_info_type): Replace begin and end with
452 (print_range): New function.
453 (set_all_positions_unneeded, any_positions_needed_p)
454 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
456 (record_store): Likewise. Optimize the case in which all positions
458 (get_stored_val): Replace read_begin and read_end with read_offset
460 (replace_read): Update call accordingly.
462 2017-12-15 Bin Cheng <bin.cheng@arm.com>
464 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
465 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
466 (loop_cand::loop_cand): Initialize above members.
467 (loop_cand::supported_operations): Delete.
468 (loop_cand::can_interchange_p): Inline above function.
469 (loop_cand::classify_simple_reduction): Record number of constant
470 initialized simple reductions.
471 (should_interchange_loops): New parameters. Check stmt cost of loops
473 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
474 Update call to should_interchange_loops.
475 (should_interchange_loop_nest): Update call to
476 should_interchange_loops.
478 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
481 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
482 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
483 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
485 2017-12-15 Julia Koval <julia.koval@intel.com>
487 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
488 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
489 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
490 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
491 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
493 2017-12-15 Julia Koval <julia.koval@intel.com>
495 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
496 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
497 * config/i386/sse.md (vaesenc_<mode>): New pattern.
498 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
499 _mm_aesenc_epi128): New intrinsics.
501 2017-12-15 Julia Koval <julia.koval@intel.com>
503 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
504 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
505 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
506 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
507 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
509 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
511 * gimple-ssa-strength-reduction.c (analyze_increments):
512 Distinguish replacement costs for constant strides from those for
515 2017-12-14 Jakub Jelinek <jakub@redhat.com>
517 * var-tracking.c (variable_tracking_main_1): Formatting fix.
519 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
521 * doc/invoke.texi: Document -Wcast-function-type.
522 * recog.h (stored_funcptr): Change signature.
523 * tree-dump.c (dump_node): Avoid warning.
524 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
526 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
529 * gimple-fold.c (get_range_strlen): Add the handling of non-member
532 2017-12-14 David Malcolm <dmalcolm@redhat.com>
534 PR tree-optimization/83312
535 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
536 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
537 "val" param, and to cope with arbitrary basic blocks.
538 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
539 handle NULL_TREE for "val", dropping "bb" param.
540 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
541 drop "bb" param. Handle NULL_TREE for "val".
542 (find_case_label_for_value): Make "switch_stmt" param const.
543 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
545 (vrp_prop::check_all_array_refs): Reimplement as...
546 (check_array_bounds_dom_walker::before_dom_children): ...this new
547 vfunc. Replace linear search through BB block list, excluding
548 those with non-executable in-edges via dominator walk.
550 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
552 * config/arm/arm.opt (mverbose-cost-dump): New option.
553 * config/arm/arm.c (arm_rtx_costs): Use it.
555 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
558 * reload1.c (emit_input_reload_insns): Skip debug markers.
560 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
562 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
566 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
567 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
568 * config/sh/sh.c: Skip debug insns besides notes.
569 * config/sh/sh.md: Likewise.
570 * config/sh/sh_treg_combine.cc: Likewise.
571 * config/sh/sync.md: Likewise.
573 2017-12-14 Tom de Vries <tom@codesourcery.com>
575 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
576 Add item for weak_undefined.
578 2017-12-14 Richard Biener <rguenther@suse.de>
580 PR tree-optimization/67842
581 * sese.h (bb_in_region): Remove #if 0'ed code.
583 2017-12-14 Richard Biener <rguenther@suse.de>
585 PR tree-optimization/83326
586 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
587 may_be_zero parameter and handle it by not marking the first
588 peeled copy as not exiting the loop.
589 (try_peel_loop): Likewise.
590 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
591 to handle the case of constant or zero iterations and perform
592 loop header copying on-the-fly.
594 2017-12-14 Richard Biener <rguenther@suse.de>
596 PR tree-optimization/83418
597 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
598 Instead of asserting we don't get unfolded comparisons deal with
601 2017-12-14 Jakub Jelinek <jakub@redhat.com>
604 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
605 basic blocks. Assert debug bind insns don't appear outside of bbs,
606 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
609 PR tree-optimization/83198
610 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
611 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
612 value if arg is a REAL_CST with incompatible type.
614 2017-12-14 Sudakshina Das <sudi.das@arm.com>
615 Bin Cheng <bin.cheng@arm.com>
618 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
620 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
623 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
625 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
626 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
628 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
629 Alan Hayward <alan.hayward@arm.com>
630 David Sherwood <david.sherwood@arm.com>
632 * poly-int.h: New file.
633 * poly-int-types.h: Likewise.
634 * coretypes.h: Include them.
635 (POLY_INT_CONVERSION): Define.
636 * target.def (estimated_poly_value): New hook.
637 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
638 * doc/tm.texi: Regenerate.
639 * doc/poly-int.texi: New file.
640 * doc/gccint.texi: Include it.
641 * doc/rtl.texi: Describe restrictions on subreg modes.
642 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
643 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
644 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
645 * targhooks.h (default_estimated_poly_value): Declare.
646 * targhooks.c (default_estimated_poly_value): New function.
647 * target.h (estimated_poly_value): Likewise.
648 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
649 (wi::unary_traits): Delete.
650 (wi::binary_traits::signed_shift_result_type): Define for
651 offset_int << HOST_WIDE_INT, etc.
652 (generic_wide_int::operator <<=): Define for all types and use
653 wi::lshift instead of <<.
654 (wi::hwi_with_prec): Add a default constructor.
655 (wi::ints_for): New class.
656 (operator <<): Define for all wide-int types.
657 (operator /): New function.
658 (operator %): Likewise.
659 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
660 (ASSERT_MAYBE_NE_AT): New macros.
662 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
663 Dominik Vogt <vogt@linux.vnet.ibm.com>
666 * emit-rtl.c (init_emit): Remove ??? comment.
667 * explow.c (get_dynamic_stack_size): Take known alignment of stack
668 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
669 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
670 alignment of 3 virtual registers to BITS_PER_WORD.
672 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
674 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
676 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
677 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
679 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
682 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
684 2017-12-13 Jakub Jelinek <jakub@redhat.com>
686 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
687 with the exception of debug begin stmt markers appear before
691 * final.c (rest_of_handle_final): Call variable_tracking_main only
692 if !flag_var_tracking.
694 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
695 Jakub Jelinek <jakub@redhat.com>
699 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
700 labels debug stmts that can only appear after labels.
702 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
704 PR rtl-optimization/82398
705 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
706 EXPR_USEFULNESS in priority comparison.
708 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
710 PR rtl-optimization/83393
711 * combine.c (move_deaths): If reg_stat points to a too new insn in
712 last_death, do not use it: find the proper insn instead.
714 2017-12-12 Jeff Law <law@redhat.com>
716 PR tree-optimization/83298
717 PR tree-optimization/83362
718 PR tree-optimization/83383
719 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
720 push_value_range a public interface. Add new argument to
721 record_ranges_from_stmt.
722 * gimple-ssa-evrp-analyze.c
723 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
724 Update comments. Handle recording temporary equivalences.
725 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
726 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
727 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
728 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
729 gimple-ssa-evrp-analyze.h.
730 (record_temporary_equivalences_from_phis): Add new argument. When
731 the PHI arg is an SSA_NAME, set the result's range to the range
733 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
735 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
736 Pass it down to children as needed.
737 (thread_outgoing_edges): Likewise.
738 (thread_across_edge): Likewise. Push/pop range state as needed.
739 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
741 2017-12-12 Julia Koval <julia.koval@intel.com>
743 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
744 (PTA_CANNONLAKE): Remove PTA_CLWB.
746 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
749 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
750 (vcondv2div2df): Likewise.
751 (vconduv2dfv2di): Likewise.
753 2017-12-12 Jakub Jelinek <jakub@redhat.com>
755 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
756 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
757 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
758 ATTR_MATHFN_FPROUNDING_ERRNO.
760 2017-12-12 Richard Biener <rguenther@suse.de>
762 PR tree-optimization/83385
763 * graphite-scop-detection.c (get_order, order): Remove.
764 (bb_to_rpo): New global.
766 (build_scops): Sort pbbs in RPO order.
768 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
770 * combine.c (simplify_set): Do not transform subregs to zero_extends
771 if the destination mode is a vector mode.
773 2017-12-12 Jakub Jelinek <jakub@redhat.com>
775 PR tree-optimization/83359
776 * tree-cfg.h (fold_loop_internal_call): Declare.
777 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
778 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
779 (find_loop_dist_alias): New function.
780 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
781 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
784 PR tree-optimization/80631
785 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
786 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
787 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
788 hardcoding zero as the value if COND_EXPR is never true. For
789 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
790 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
791 hardcoding MAX_EXPR as the reduction operation.
792 (is_nonwrapping_integer_induction): Allow negative step.
793 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
794 vect_create_epilog_for_reduction, if no value is suitable, don't
795 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
797 2017-12-12 Richard Biener <rguenther@suse.de>
799 PR tree-optimization/81889
800 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
801 range info from the non-wrapping IV instead of just the range
804 2017-12-12 Julia Koval <julia.koval@intel.com>
806 * config.gcc: Add vaesintrin.h.
807 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
809 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
810 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
812 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
813 * config/i386/immintrin.h: Include vaesintrin.h.
814 * config/i386/sse.md (vaesdec_<mode>): New pattern.
815 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
816 _mm_aesdec_epi128): New intrinsics.
818 2017-12-12 Julia Koval <julia.koval@intel.com>
820 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
821 OPTION_MASK_ISA_VAES_UNSET): New.
822 (ix86_handle_option): Handle -mvaes.
823 * config/i386/cpuid.h: Define bit_VAES.
824 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
825 * config/i386/i386-c.c (__VAES__): New.
826 * config/i386/i386.c (ix86_target_string): Add -mvaes.
827 (ix86_valid_target_attribute_inner_p): Ditto.
828 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
829 * config/i386/i386.opt: Add -mvaes.
830 * doc/invoke.texi: Ditto.
832 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
834 * debug.h (gcc_debug_hooks): Add inline_entry.
835 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
836 * debug.c (do_nothing_debug_hooks): Likewise.
837 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
838 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
839 (dwarf2_lineno_debug_hooks): Likewise.
841 * common.opt (gstatement-frontiers): New, setting
842 debug_nonbind_markers_p.
843 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
844 * toplev.c (process_options): Autodetect value for debug statement
846 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
847 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
849 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
850 markers. Integrate source bind into debug stmt expand loop.
851 (pass_expand::execute): Check debug marker limit. Avoid deep
852 TER and expand debug locations for debug bind insns only.
853 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
855 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
856 * final.c (reemit_insn_block_notes): Take current block from
857 nonbind markers. Declare note where it's first set.
858 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
859 begin stmt markers if enabled.
860 (notice_source_line): Handle nonbind markers. Fail if their
861 location is unknown or that of builtins.
862 (rest_of_handle_final): Convert begin stmt markers to notes if
863 var-tracking didn't run.
864 (rest_of_clean_state): Skip begin stmt markers.
865 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
867 * function.c (allocate_struct_function): Set begin_stmt_markers.
868 * function.h (struct function): Add debug_marker_count counter
869 and debug_nonbind_markers flag.
870 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
871 * gimple-low.c (lower_function_body): Adjust
872 debug_nonbind_markers.
873 (lower_stmt): Drop or skip gimple debug stmts.
874 (lower_try_catch): Skip debug stmts.
875 * gimple.c (gimple_build_debug_begin_stmt): New.
876 (gimple_copy): Increment debug_marker_count if copying one.
877 * gimple.h (gimple_build_debug_begin_stmt): Declare.
878 * gimplify.c (rexpr_location): New.
879 (rexpr_has_location): New.
880 (warn_switch_unreachable_r): Handle gimple debug stmts.
881 (shortcut_cond_r): Call expr_location.
883 (find_goto_label): New.
884 (shortcut_cond_expr): Call expr_has_location, expr_location, and
886 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
888 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
889 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
890 (LANG_HOOKS_INITIALIZER): ... this.
891 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
892 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
894 (update_ebb_live_info): Skip debug insn markers.
895 * lra.c (debug_insn_static_data): Rename to...
896 (debug_bind_static_data): ... this.
897 (debug_marker_static_data): New.
898 (lra_set_insn_recog_data): Select one of the above depending
900 (lra_update_isn_regno_info): Don't assume debug insns have
902 (push_insns): Skip debug insns.
903 * lto-streamer-in.c (input_function): Drop debug stmts
904 depending on active options. Adjust debug_nonbind_markers.
905 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
906 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
907 begin stmt marker notes.
908 (print_insn): Likewise.
909 * recog.c (extract_insn): Recognize rtl for debug markers.
910 * rtl.def (DEBUG_MARKER): New.
911 * tree-inline.c: Include params.h.
912 (remap_gimple_stmt): Handle nonbind markers.
913 (maybe_move_debug_stmts_to_successors): Likewise.
914 (copy_debug_stmt): Likewise.
915 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
916 markers regardless of no side effects.
917 (tsi_link_before): Don't update container's side effects when adding
919 (tsi_link_after): Likewise.
920 (expr_first): Skip begin stmt markers.
921 (expr_last): Likewise.
922 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
923 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
924 Disregard nonbind markers.
925 * tree.c (make_node_stat): Don't set side effects for begin stmt
927 (build1_stat): Likewise.
928 * tree.def (DEBUG_BEGIN_STMT): New.
929 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
930 * var-tracking.c (delete_debug_insns): Renamed to...
931 (delete_vta_debug_insns): ... this.
932 (reemit_marker_as_note): New.
933 (vt_initialize): Reemit markers.
934 (delete_vta_debug_insns): Likewise.
935 (vt_debug_insns_local): Reemit or delete markers.
936 (variable_tracking_main_1): Likewise.
937 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
938 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
939 (gimple_debug_nonbind_marker_p): New.
940 (gimple_build_debug_bind): Adjust.
941 (gimple_build_debug_begin_stmt): New.
942 * doc/invoke.texi (max-debug-marker-count): New param.
943 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
944 (debug_parameter_ref, debug_marker): New.
945 (NOTE_INSN_BEGIN_STMT): New.
946 (DEBUG_INSN): Describe begin stmt markers.
948 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
949 without debug insns, we wouldn't, but clean up debug insns
950 after a control flow insn nevertheless.
952 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
953 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
954 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
955 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
956 (rtl_tidy_fallthru_edge): Likewise.
957 (rtl_verify_fallthru): Likewise.
958 (rtl_verify_bb_layout): Likewise.
959 (skip_insns_after_block): Likewise.
960 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
961 * dwarf2out.c: Include print-rtl.h.
962 (dwarf2out_next_real_insn): New.
963 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
964 Dump debug binds in asm comments.
965 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
966 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
967 callers to use gsi_start_nondebug_bb instead.
968 (gsi_after_labels): Skip gimple debug stmts.
969 (gsi_start_nondebug): New.
970 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
971 (proper_loop_form_for_interchange): Adjust.
972 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
973 * gimple.h (gimple_seq_last_nondebug_stmt): New.
974 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
975 (collect_fallthrough_labels): Likewise.
976 (should_warn_for_implicit_fallthrough): Likewise.
977 (warn_implicit_fallthrough_r): Likewise.
978 (expand_FALLTHROUGH_r): Likewise.
979 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
980 (graphite_copy_stmts_from_block): Skip nonbind markers.
981 * haifa-sched.c (sched_extend_bb): Skip debug insns.
982 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
983 * jump.c (clean_barriers): Skip debug insns.
984 * omp-expand.c (expand_parallel_call): Skip debug insns.
985 (expand_task_call): Likewise.
986 (remove_exit_barrier): Likewise.
987 (expand_omp_taskreg): Likewise.
988 (expand_omp_for_init_counts): Likewise.
989 (expand_omp_for_generic): Likewise.
990 (expand_omp_for_static_nochunk): Likewise.
991 (expand_omp_for_static_chunk): Likewise.
992 (expand_omp_simd): Likewise.
993 (expand_omp_taskloop_for_outer): Likewise.
994 (expand_omp_taskloop_for_inner): Likewise.
995 (expand_oacc_for): Likewise.
996 (expand_omp_sections): Likewise.
997 (expand_omp_single): Likewise.
998 (expand_omp_synch): Likewise.
999 (expand_omp_atomic_load): Likewise.
1000 (expand_omp_atomic_store): Likewise.
1001 (expand_omp_atomic_fetch_op): Likewise.
1002 (expand_omp_atomic_pipeline): Likewise.
1003 (expand_omp_atomic_mutex): Likewise.
1004 (expand_omp_target): Likewise.
1005 (grid_expand_omp_for_loop): Likewise.
1006 (grid_expand_target_grid_body): Likewise.
1007 (build_omp_regions_1): Likewise.
1008 * omp-low.c (check_combined_parallel): Skip debug stmts.
1009 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
1010 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
1011 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
1013 * tree-cfg.c (make_blobs_1): Skip debug stmts.
1014 (make_edges): Likewise.
1015 (cleanup_dead_labels): Likewise.
1016 (gimple_can_merge_blocks_p): Likewise.
1017 (stmt_starts_bb_p): Likewise.
1018 (gimple_block_label): Likewise.
1019 (gimple_redirect_edge_and_branch): Likewise.
1020 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
1022 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
1024 * tree-pretty-print (print_declaration): Omit initializer in slim
1026 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
1028 (eliminate_unnecessary_stmts): Stabilize block removal order.
1029 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
1030 * var-tracking.c (get_first_insn): New.
1031 (vt_emit_notes): Call it.
1032 (vt_initialize): Walk any insns before the first BB.
1033 (delete_debug_insns): Likewise.
1035 * gimple.h (enum gimple_debug_subcode): Add
1036 GIMPLE_DEBUG_BEGIN_STMT.
1037 (gimple_debug_begin_stmt_p): New.
1038 (gimple_debug_nonbind_marker_p): New.
1039 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
1040 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
1041 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
1042 * insn-notes.def (BEGIN_STMT): New.
1043 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
1044 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
1045 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
1046 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
1047 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
1048 (INSN_DEBUG_MARKER_KIND): New.
1049 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
1050 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
1051 (INSN_VAR_LOCATION_PTR): New.
1052 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
1054 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
1055 deps only if debug bind insns are enabled.
1056 (pass_expand::execute): Avoid deep TER and expand
1057 debug locations for debug bind insns only.
1058 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
1059 debug stmts special handling down to debug bind stmts.
1060 * combine.c (try_combine): Narrow debug insns special handling
1061 down to debug bind insns.
1062 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
1063 Narrow debug insns preexisting special handling down to debug
1065 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
1066 handling down to debug bind insns.
1067 * function.c (instantiate_virtual_regs): Skip debug markers,
1068 adjust handling of debug binds.
1069 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
1070 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
1071 * haifa-sched.c (schedule_insn): Narrow special handling of debug
1072 insns to debug bind insns.
1073 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
1074 special handling of debug stmts to debug bind stmts.
1075 * ipa-split.c (split_function): Likewise.
1076 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
1077 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
1079 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
1080 * regrename.c (build_def_use): Likewise.
1081 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
1082 (pass_cprop_hardreg): Narrow special casing of debug insns to
1084 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
1085 * reload1.c (reload): Likewise.
1086 * sese.c (sese_insert_phis_for_liveouts): Narrow special
1087 casing of debug stmts to debug bind stmts.
1088 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
1089 * ssa-iterators.h (num_imm_uses): Likewise.
1090 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
1091 debug stmts to debug bind stmts.
1092 * tree-inline.c (tree_function_versioning): Narrow special casing
1093 of debug stmts to debug bind stmts.
1094 * tree-loop-distribution.c (generate_loops_for_partition):
1095 Narrow special casing of debug stmts to debug bind stmts.
1096 * tree-sra.c (analyze_access_subtree): Narrow special casing
1097 of debug stmts to debug bind stmts.
1098 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
1099 stmts to debug bind stmts.
1100 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
1101 casing of debug stmts to debug bind stmts.
1102 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
1103 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
1104 casing of debug stmts to debug bind stmts.
1105 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
1107 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
1108 debug stmts to debug bind stmts.
1109 (gimple_replace_ssa_lhs): Likewise.
1110 (insert_debug_temp_for_var_def): Likewise.
1111 (insert_debug_temps_for_defs): Likewise.
1112 (reset_debug_uses): Likewise.
1113 * tree-ssanames.c (release_ssa_name_fn): Likewise.
1114 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
1115 (adjust_debug_stmts): Likewise.
1116 (adjust_phi_and_debug_stmts): Likewise.
1117 (vect_do_peeling): Likewise.
1118 * tree-vect-loop.c (vect_transform_loop): Likewise.
1119 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
1120 * var-tracking.c (adjust_mems): Narrow special casing of debug
1121 insns to debug bind insns.
1122 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
1123 (compute_bb_dataflow, vt_find_locations): Likewise.
1124 (vt_expand_loc, emit_notes_for_changes): Likewise.
1125 (vt_init_cfa_base): Likewise.
1126 (vt_emit_notes): Likewise.
1127 (vt_initialize): Likewise.
1128 (vt_finalize): Likewise.
1130 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
1131 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
1132 (next_nonnote_nondebug_insn_bb): New.
1133 (prev_nonnote_nondebug_insn_bb): New.
1134 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
1135 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
1136 (prev_nonnote_nondebug_insn_bb): Declare.
1137 (next_nonnote_nondebug_insn_bb): Declare.
1138 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
1139 * cfgrtl.c (get_last_bb_insn): Likewise.
1140 * lra.c (push_insns): Likewise.
1142 2017-12-11 David Malcolm <dmalcolm@redhat.com>
1145 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
1147 * selftest.c (selftest::test_runner::test_runner): New ctor.
1148 (selftest::test_runner::~test_runner): New dtor.
1149 * selftest.h (class selftest::test_runner): New class.
1151 2017-12-11 Carl Love <cel@us.ibm.com>
1153 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
1154 vec_extract_fp32_from_shortl]): Add #defines.
1155 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
1156 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
1157 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
1158 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
1159 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
1160 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
1161 * doc/extend.texi: Add documentation for the added builtins.
1163 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
1165 PR rtl-optimization/80693
1166 PR rtl-optimization/81019
1167 PR rtl-optimization/81020
1168 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
1169 are not mentioned in i3. Place the REG_UNUSED note on i2,
1170 possibly modified to REG_DEAD, if it did not originate in i3.
1172 2017-12-11 Jakub Jelinek <jakub@redhat.com>
1174 * recog.c (store_data_bypass_p_1): New function.
1175 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
1176 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
1178 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
1180 PR rtl-optimization/83361
1181 * ifcvt.c (if_convert): Call fixup_partitions.
1183 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
1185 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1186 early folding of splat_u{8,16,32}.
1188 2017-12-11 Jakub Jelinek <jakub@redhat.com>
1190 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
1191 output_operand_lossage first argument with capital letter.
1192 (aarch64_override_options): Don't start error and sorry first argument
1193 with capital letter.
1195 2017-12-11 Andi Kleen <ak@linux.intel.com>
1197 PR gcov-profile/83355
1198 * auto-profile.c (string_table::get_index_by_decl): Don't
1199 recurse when abstract origin points to itself.
1201 2017-12-11 Bin Cheng <bin.cheng@arm.com>
1203 PR tree-optimization/83320
1204 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
1205 (prune_datarefs_not_in_loop): Ditto.
1207 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
1209 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
1211 2017-12-10 Jakub Jelinek <jakub@redhat.com>
1213 PR tree-optimization/83337
1214 * gimple-loop-interchange.cc (compute_access_stride): Handle
1215 bitfield DRs properly.
1217 2017-12-09 Jakub Jelinek <jakub@redhat.com>
1219 PR tree-optimization/83338
1220 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
1221 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
1222 vector of unsigned integers to vector of signed integers.
1224 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
1226 PR rtl-optimization/83317
1227 * lra-constraints.c (process_address_1): Add insn code check.
1229 2017-12-08 Michael Matz <matz@suse.de>
1231 Fix PR tree-optimization/83323
1232 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
1233 head-controlled loops and loop BBs.
1234 * common.opt (funroll-and-jam): Remove, instead ...
1235 (floop-unroll-and-jam): ... reuse this option.
1236 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
1237 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
1238 (-floop-unroll-and-jam): ... this option.
1240 2017-12-08 Jakub Jelinek <jakub@redhat.com>
1242 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
1243 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
1244 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
1246 2017-12-08 Richard Biener <rguenther@suse.de>
1249 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
1250 handle accesses outside of zero-sized vars.
1252 2017-12-08 Martin Jambor <mjambor@suse.cz>
1254 PR tree-optimization/83141
1255 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
1256 test for MEM_REFs implicitely changing types with padding. Remove
1258 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
1260 2017-12-08 Jakub Jelinek <jakub@redhat.com>
1262 * config/arc/arc.c (arc_attribute_table): Add exclusions to
1264 * config/avr/avr.c (avr_attribute_table): Likewise.
1265 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
1266 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
1267 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
1268 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
1269 of struct attribute_spec.
1270 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1272 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
1275 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
1277 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
1280 * profile-count.c (profile_count::from_gcov_type): Move from
1281 profile-count.h; handle overflow.
1282 * profile-count.h (profile_count::from_gcov_type): Move offline.
1284 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
1286 PR rtl-optimization/83304
1287 * combine.c (move_deaths): If we do not know where a register died,
1290 2017-12-08 Richard Biener <rguenther@suse.de>
1292 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
1293 Provide -fopt-info-loop feedback when we interchange in a nest.
1295 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
1297 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
1298 for armv6 ARM CPU IDs.
1300 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
1302 * common/config/arm/arm-common.c: Include <algorithm>.
1303 (INCLUDE_VECTOR): Define.
1304 (compare_opt_names): New function.
1305 (arm_rewrite_selected_arch): Only strip out extensions that can be
1306 expressed through -mfpu. Sort the remaining extensions
1309 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
1311 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
1312 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
1313 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
1314 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
1316 2017-06-08 Tristan Gingold <gindold@adacore.com>
1319 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
1320 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
1322 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1324 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
1325 of struct attribute_spec.
1327 2017-12-08 Julia Koval <julia.koval@intel.com>
1329 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
1330 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
1331 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
1332 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
1333 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
1334 _mm_maskz_dpwssds_epi32): Ditto.
1336 2017-12-08 Richard Biener <rguenther@suse.de>
1338 PR tree-optimization/81303
1339 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
1340 conditions try to create a comparison vector type matching
1341 the data vector type.
1342 (vectorizable_condition): Adjust.
1343 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
1344 Leave invariant conditions alone in case we can vectorize those.
1346 2017-12-08 Julia Koval <julia.koval@intel.com>
1348 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
1349 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
1350 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
1351 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
1352 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
1354 2017-12-08 Julia Koval <julia.koval@intel.com>
1356 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
1357 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
1358 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
1359 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
1360 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
1361 _mm_maskz_dpbusds_epi32): New intrinsics.
1363 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
1365 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
1367 (ld<bh>io_signed): Likewise.
1368 (st<bhw>io): Likewise.
1369 * config/nios2/predicates.md (ldstio_memory_operand): Allow
1370 SMALL_INT12 constant integer operand.
1372 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
1374 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
1376 (nios2_large_constant_allowed): ...this. Adjust uses.
1377 (nios2_plus_symbolic_constant_p): Rename to...
1378 (nios2_plus_large_constant_p): ...this. Adjust uses.
1379 (nios2_legitimate_address_p): Correct CONST_INT handling.
1380 (nios2_symbolic_memory_operand_p): Rename to...
1381 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
1382 (nios2_large_constant_p): Check for large constant integers too.
1383 (nios2_split_large_constant): Handle constant integers.
1384 (nios2_split_symbolic_memory_operand): Rename to...
1385 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
1386 (nios2_legitimize_constant_address): Handle constant integers.
1387 (r0rel_constant_p): Handle small constant integers.
1388 (nios2_print_operand_address): Handle r0-relative integer addresses.
1389 * config/nios2/nios2-protos.h: Adjust for renamed functions.
1390 * config/nios2/nios2.md: Adjust for renamed functions.
1392 2017-12-07 Andrew Waterman <andrew@sifive.com>
1394 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
1395 (TARGET_HAVE_SRODATA_SECTION): New define.
1396 (riscv_select_section): New function.
1398 2017-12-08 Joseph Myers <joseph@codesourcery.com>
1399 Alexander Monakov <amonakov@ispras.ru>
1400 Jakub Jelinek <jakub@redhat.com>
1403 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
1405 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1407 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
1408 a bool success value. Don't call output_operand_lossage here.
1409 (aarch64_print_ldpstp_address): Return a bool success value.
1410 (aarch64_print_operand_address): Call output_addr_const if
1411 aarch64_print_address_internal fails.
1412 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
1413 'y'; call output_operand_lossage instead. Call output_operand_lossage
1414 if aarch64_print_ldpstp_address fails.
1416 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1418 * tree-vector-builder.h
1419 (tree_vector_builder::binary_encoded_nelts): Declare.
1420 * tree-vector-builder.c
1421 (tree_vector_builder::binary_encoded_nelts): New function.
1422 * fold-const.c (negate_expr_p): Likewise.
1423 (operand_equal_p, fold_checksum_tree): Likewise.
1424 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
1425 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
1426 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
1427 (uniform_vector_p): Likewise.
1428 * varasm.c (const_hash_1, compare_constant): Likewise.
1429 * tree-ssa-ccp.c: Include tree-vector-builder.h.
1430 (valid_lattice_transition): Operate directly on the VECTOR_CST
1432 * ipa-icf.c: Include tree-vector-builder.h.
1433 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
1434 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
1436 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1438 * tree.c (build_vector): Delete.
1439 * tree.h (build_vector): Make static and move into the self-testing
1442 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1444 * vector-builder.h (vector_builder::derived): New const overload.
1445 (vector_builder::elt): New function.
1446 * tree-vector-builder.h (tree_vector_builder::type): New function.
1447 (tree_vector_builder::apply_step): Declare.
1448 * tree-vector-builder.c (tree_vector_builder::apply_step): New
1450 * gimple-fold.h (tree_vector_builder): Declare.
1451 (gimple_build_vector): Take a tree_vector_builder instead of a
1452 type and vector of elements.
1453 * gimple-fold.c (gimple_build_vector): Likewise.
1454 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
1456 (get_initial_defs_for_reduction): Likewise.
1457 (vectorizable_induction): Likewise.
1459 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1461 * tree-vector-builder.h
1462 (tree_vector_builder::new_binary_operation): Declare.
1463 * tree-vector-builder.c
1464 (tree_vector_builder::new_binary_operation): New function.
1465 * fold-const.c (fold_relational_const): Use it.
1466 (const_binop): Likewise. Check that both input vectors have
1467 the same number of elements, thus excluding things like WIDEN_SUM.
1468 Check whether it is possible to operate directly on the encodings
1471 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1473 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
1474 new_unary_operation, operating only on the encoded elements.
1475 (const_unop): Likewise.
1476 (exact_inverse): Likewise.
1477 (distributes_over_addition_p): New function.
1478 (const_binop): Use tree_vector_builder and new_unary_operation
1479 for combinations of VECTOR_CST and INTEGER_CST. Operate only
1480 on the encoded elements unless the encoding is strided and the
1481 operation does not distribute over addition.
1482 (fold_convert_const): Use tree_vector_builder and
1483 new_unary_operation. Operate only on the encoded elements
1484 for truncating integer conversions, or for non-stepped encodings.
1486 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1488 * config/sparc/sparc.c: Include tree-vector-builder.h.
1489 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
1490 * expmed.c: Include tree-vector-builder.h.
1491 (make_tree): Use tree_vector_builder instead of build_vector.
1492 * fold-const.c: Include tree-vector-builder.h.
1493 (const_binop): Use tree_vector_builder instead of build_vector.
1494 (const_unop): Likewise.
1495 (native_interpret_vector): Likewise.
1496 (fold_vec_perm): Likewise.
1497 (fold_ternary_loc): Likewise.
1498 * gimple-fold.c: Include tree-vector-builder.h.
1499 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
1501 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
1502 (simplify_vector_constructor): Use tree_vector_builder instead
1504 * tree-vect-generic.c: Include tree-vector-builder.h.
1505 (add_rshift): Use tree_vector_builder instead of build_vector.
1506 (expand_vector_divmod): Likewise.
1507 (optimize_vector_constructor): Likewise.
1508 * tree-vect-loop.c: Include tree-vector-builder.h.
1509 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
1510 of build_vector. Explicitly use a stepped encoding for
1512 * tree-vect-slp.c: Include tree-vector-builder.h.
1513 (vect_get_constant_vectors): Use tree_vector_builder instead
1515 (vect_transform_slp_perm_load): Likewise.
1516 (vect_schedule_slp_instance): Likewise.
1517 * tree-vect-stmts.c: Include tree-vector-builder.h.
1518 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
1519 (vect_gen_perm_mask_any): Likewise.
1520 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
1521 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
1523 (build_vector_from_val): Likewise. Explicitly use a duplicate
1526 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
1528 * doc/generic.texi (VECTOR_CST): Describe new representation of
1530 * vector-builder.h: New file.
1531 * tree-vector-builder.h: Likewise.
1532 * tree-vector-builder.c: Likewise.
1533 * Makefile.in (OBJS): Add tree-vector-builder.o.
1534 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
1535 * tree-core.h (tree_base): Add a vector_cst field to the u union.
1536 (tree_vector): Change the number of elements to
1537 vector_cst_encoded_nelts.
1538 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
1539 (VECTOR_CST_ELTS): Delete.
1540 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
1541 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
1542 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
1543 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
1544 (VECTOR_CST_ENCODED_ELT): Likewise.
1545 (vector_cst_encoded_nelts): New function.
1546 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
1547 VECTOR_CST_NELTS_PER_PATTERN as arguments.
1548 (vector_cst_int_elt, vector_cst_elt): Declare.
1549 * tree.c: Include tree-vector-builder.h.
1550 (tree_code_size): Abort if passed VECTOR_CST.
1551 (tree_size): Update for new VECTOR_CST layout.
1552 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
1553 VECTOR_CST_NELTS_PER_PATTERN as arguments.
1554 (build_vector): Use tree_vector_builder.
1555 (vector_cst_int_elt, vector_cst_elt): New functions.
1556 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
1557 encoded elements and then create the vector in the canonical form.
1558 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
1559 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
1560 (tree_c_tests): Call test_vector_cst_patterns.
1561 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
1563 (hash_tree): Likewise.
1564 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
1565 (streamer_write_tree_header): Likewise.
1566 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
1567 (streamer_alloc_tree): Likewise. Update call to make_vector.
1568 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
1570 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1572 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
1573 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
1574 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
1576 2017-12-07 Bin Cheng <bin.cheng@arm.com>
1577 Richard Biener <rguenther@suse.de>
1579 PR tree-optimization/81303
1580 * Makefile.in (gimple-loop-interchange.o): New object file.
1581 * common.opt (floop-interchange): Reuse the option from graphite.
1582 * doc/invoke.texi (-floop-interchange): Ditto. New document for
1583 -floop-interchange and mention it for -O3.
1584 * opts.c (default_options_table): Enable -floop-interchange at -O3.
1585 * gimple-loop-interchange.cc: New file.
1586 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
1587 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
1588 * passes.def (pass_linterchange): New pass.
1589 * timevar.def (TV_LINTERCHANGE): New time var.
1590 * tree-pass.h (make_pass_linterchange): New declaration.
1591 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
1592 interchange. Record IV before/after increment in new parameters.
1593 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
1594 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
1596 (check_reduction_path): ...New function here.
1597 * tree-vectorizer.h (check_reduction_path): New declaration.
1599 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
1602 PR rtl-optimization/80818
1603 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
1604 always early clobbered.
1605 * lra-lives.c (process_bb_lives): Check input hard regs for early
1606 clobbered non-operand hard reg.
1608 2017-12-07 Jakub Jelinek <jakub@redhat.com>
1611 * tree-cfg.c (verify_gimple_assign_binary): Don't require
1612 types_compatible_p, just that TYPE_MODE is the same.
1614 2017-12-07 Martin Sebor <msebor@redhat.com>
1617 * attribs.c (empty_attribute_table): Initialize new member of
1618 struct attribute_spec.
1619 (decl_attributes): Add argument. Handle mutually exclusive
1620 combinations of attributes.
1621 (selftests::test_attribute_exclusions): New function.
1622 (selftests::attribute_c_tests): Ditto.
1623 * attribs.h (decl_attributes): Add default argument.
1624 * selftest.h (attribute_c_tests): Declare.
1625 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
1626 * tree-core.h (attribute_spec::exclusions, exclude): New type and
1628 * doc/extend.texi (Common Function Attributes): Update const and pure.
1629 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
1630 of struct attribute_spec.
1631 * config/arc/arc.c (arc_attribute_table): Same.
1632 * config/arm/arm.c (arm_attribute_table): Same.
1633 * config/avr/avr.c ( avr_attribute_table): Same.
1634 * config/bfin/bfin.c (bfin_attribute_table): Same.
1635 * config/cr16/cr16.c (cr16_attribute_table): Same.
1636 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
1637 * config/h8300/h8300.c (h8300_attribute_table): Same.
1638 * config/i386/i386.c (ix86_attribute_table): Same.
1639 * config/ia64/ia64.c (ia64_attribute_table): Same.
1640 * config/m32c/m32c.c (m32c_attribute_table): Same.
1641 * config/m32r/m32r.c (m32r_attribute_table): Same.
1642 * config/m68k/m68k.c (m68k_attribute_table): Same.
1643 * config/mcore/mcore.c (mcore_attribute_table): Same.
1644 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
1645 * config/mips/mips.c (mips_attribute_table): Same.
1646 * config/msp430/msp430.c (msp430_attribute_table): Same.
1647 * config/nds32/nds32.c (nds32_attribute_table): Same.
1648 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
1649 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
1650 * config/rl78/rl78.c (rl78__attribute_table): Same.
1651 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
1652 * onfig/rx/rx.c (rx_attribute_table): Same.
1653 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
1654 * config/sh/sh.c (sh_attribute_table): Same.
1655 * config/sparc/sparc.c (sparc_attribute_table): Same.
1656 * config/spu/spu.c (spu_attribute_table): Same.
1657 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
1658 * config/v850/v850.c (v850_attribute_table): Same.
1659 * config/visium/visium.c (visium_attribute_table): Same.
1661 2017-12-07 Tamar Christina <tamar.christina@arm.com>
1664 * config/arm/arm.c (INCLUDE_STRING): Define.
1665 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
1666 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
1669 2017-12-07 Michael Matz <matz@suse.de>
1671 Add unroll and jam pass
1673 * gimple-loop-jam.c: New file.
1674 * Makefile.in (OBJS): Add gimple-loop-jam.o.
1675 * common.opt (funroll-and-jam): New option.
1676 * opts.c (default_options_table): Add unroll-and-jam at -O3.
1677 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
1678 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
1679 * passes.def: Add pass_loop_jam.
1680 * timevar.def (TV_LOOP_JAM): Add.
1681 * tree-pass.h (make_pass_loop_jam): Declare.
1682 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
1683 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
1684 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
1685 to flow_loop_tree_node_add.
1686 (duplicate_subloops, copy_loops_to): Append to sibling list.
1687 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
1688 * doc/invoke.texi (-funroll-and-jam): Document new option.
1689 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
1691 2017-12-07 Richard Biener <rguenther@suse.de>
1693 PR tree-optimization/83296
1694 PR tree-optimization/67769
1695 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
1696 flow sensitive info in an unrelated BB.
1697 (value_replacement): Use reset_flow_sensitive_info.
1698 (minmax_replacement): Reset flow sensitive info on the def
1699 we move. Do not reset flow sensitive info in the whole BB
1700 we move the stmt to.
1701 (abs_replacement): Likewise.
1703 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1706 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1707 rs6000_cpu to the given -mcpu=, or to the default processor.
1709 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1711 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
1712 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
1713 instead of rs6000_cpu_attr.
1714 (rs6000_adjust_cost): Ditto.
1715 (is_microcoded_insn): Ditto.
1716 (rs6000_adjust_priority): Ditto.
1717 (rs6000_issue_rate): Ditto.
1718 (rs6000_use_sched_lookahead): Ditto.
1719 (rs6000_use_sched_lookahead_guard): Ditto.
1720 (rs6000_sched_reorder): Ditto.
1721 (force_new_group): Ditto.
1722 * config/rs6000/rs6000.md (cpu attribute): Ditto.
1723 (group_ending_nop): Ditto.
1725 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1727 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
1728 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
1729 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
1730 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
1732 (rs6000_reassociation_width): Ditto.
1733 (rs6000_emit_epilogue): Ditto.
1734 (rs6000_adjust_cost): Ditto.
1735 (is_microcoded_insn): Ditto.
1736 (is_cracked_insn): Ditto.
1737 (rs6000_adjust_priority): Ditto.
1738 (rs6000_sched_reorder): Ditto.
1739 (rs6000_sched_reorder2): Ditto.
1740 (insn_must_be_first_in_group): Ditto.
1741 (insn_must_be_last_in_group): Ditto.
1742 (rs6000_register_move_cost): Ditto.
1743 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
1746 2017-12-07 Julia Koval <julia.koval@intel.com>
1748 * config.gcc: Add vaesintrin.h.
1749 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
1751 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
1752 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
1754 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1755 * config/i386/immintrin.h: Include vaesintrin.h.
1756 * config/i386/sse.md (vaesdec_<mode>): New pattern.
1757 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
1758 _mm_aesdec_epi128): New intrinsics.
1760 2017-12-06 David Malcolm <dmalcolm@redhat.com>
1762 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
1763 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
1764 c-family/c-spellcheck.cc.
1765 (best_macro_match::best_macro_match): Likewise.
1767 (struct edit_distance_traits<cpp_hashnode *>): Move to
1768 c-family/c-spellcheck.h.
1769 (class best_macro_match): Likewise.
1771 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1773 PR tree-optimization/83293
1774 * gimple-ssa-strength-reduction.c (insert_initializers): Use
1775 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
1776 might insert into empty bb.
1779 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
1780 simplify for plus with :c added, and pointer_plus without that.
1781 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
1782 with undefined overflow and the conversion is not widening,
1783 perform negation in utype and only convert to type afterwards.
1784 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
1785 simplify for plus with :c added, and pointer_plus without that.
1786 If type is integral with undefined overflow and the conversion is
1787 not widening, perform minus in utype and only convert to type
1788 afterwards. Move the last pointer_diff_expr simplify into the
1791 2017-12-06 Martin Sebor <msebor@redhat.com>
1793 PR tree-optimization/82646
1794 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
1795 strncpy, not maxlen.
1797 2017-12-06 Martin Sebor <msebor@redhat.com>
1799 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
1802 PR tree-optimization/83075
1803 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
1804 strncat/strncpy don't change length of source string.
1806 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
1809 2017-11-29 Martin Aberg <maberg@gaisler.com>
1811 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1812 to prevent b2bst errata sequence.
1813 (sqrtdf2_fix): Likewise.
1815 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1817 PR tree-optimization/81945
1818 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
1819 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
1820 to dest_cfun has orig_loop_num set, either remap it to the new
1821 loop number if the loop got moved too, or clear it.
1823 2017-12-05 Steve Ellcey <sellcey@cavium.com>
1825 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
1827 (thunderx2t99_nothing): New insn reservation.
1828 (thunderx2t99_mrs): New insn reservation.
1829 (thunderx2t99_multiple): New insn reservation.
1830 (thunderx2t99_alu_basi): Add bfx to reservation.
1831 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
1833 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1836 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
1838 2017-12-05 Bin Cheng <bin.cheng@arm.com>
1840 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
1841 tree-ssa-pre.c::remove_dead_inserted_code.
1842 * tree-ssa-dce.h: New file.
1843 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
1844 (remove_dead_inserted_code): Move and rename to function
1845 tree-ssa-dce.c::simple_dce_from_worklist.
1846 (pass_pre::execute): Update use.
1848 2017-12-05 Richard Biener <rguenther@suse.de>
1850 PR tree-optimization/83277
1851 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
1852 to code-gen liveout vars.
1854 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
1856 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
1857 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
1858 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
1861 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
1863 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
1865 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
1866 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
1869 2017-12-05 Richard Biener <rguenther@suse.de>
1871 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
1872 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
1873 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
1874 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
1876 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1878 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
1880 2017-12-05 Martin Liska <mliska@suse.cz>
1881 Jakub Jelinek <jakub@redhat.com>
1883 * doc/invoke.texi: Document the options.
1884 * flag-types.h (enum sanitize_code): Add
1885 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1886 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
1887 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1888 * opts.c: Define new sanitizer options.
1889 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
1890 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
1892 2017-12-05 Julia Koval <julia.koval@intel.com>
1894 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
1895 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
1896 (ix86_handle_option): Handle -mavx512vnni.
1897 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
1898 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
1899 * config/i386/i386-c (__AVX512VNNI__): New.
1900 * config/i386/i386.c (ix86_target_string): Handle new option.
1901 (ix86_valid_target_attribute_inner_p): Handle new option.
1902 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
1903 * config/i386/i386.opt (mavx512vnni): New option.
1905 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1908 * athlon.md: Disable for generic.
1909 * haswell.md: Enable for generic.
1910 * i386.c (ix86_sched_init_global): Add core hooks for generic.
1911 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
1913 (ix86_adjust_cost): Move generic to haswell path.
1915 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
1917 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
1918 instead of MEM_P in a couple more places. Fix formatting issues.
1920 2017-12-04 Jim Wilson <jimw@sifive.com>
1922 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
1923 instead of GP_REG_LAST-1.
1924 (riscv_adjust_libcall_cfi_prologue): Likewise.
1925 (riscv_adjust_libcall_cri_epilogue): Likewise.
1926 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
1929 2017-12-04 Luis Machado <luis.machado@linaro.org>
1931 * ipa-pure-const.c (check_decl): Add missing newline.
1932 (state_from_flags): Likewise.
1934 2017-12-04 Jeff Law <law@redhat.com>
1936 PR tree-optimizatin/78496
1937 * gimple-ssa-evrp-analyze.h
1938 (evrp_range_analyzer::get_vr_values): Simplify.
1939 * gimple-ssa-evrp-analyze.c: Corresponding changes.
1940 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
1941 and gimple-ssa-evrp-analyze.h.
1942 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
1943 (simplify_stmt_for_jump_threading): Copy a blob of code from
1944 tree-vrp.c to use ranges to simplify statements.
1945 (dom_opt_dom_walker::before_dom_children): Call
1946 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
1947 (dom_opt_dom_walker::after_dom_children): Similarly for
1948 evrp_range_analyzer::leave.
1949 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
1952 * gimple-ssa-evrp-analyze.c
1953 (evrp_range_analyzer::extract_range_from_stmt): Always use
1954 vr_values::update_value_range so preexisting range info is
1955 medged with new range info, even if the new range is VR_VARYING.
1957 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1959 * combine.c: Adjust comment.
1960 (use_crosses_set_p): Delete.
1961 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
1962 (try_combine): Ditto.
1964 2017-12-04 Richard Biener <rguenther@suse.de>
1966 PR tree-optimization/83255
1967 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
1968 Re-add zero-iteration check.
1970 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1972 PR rtl-optimization/83245
1973 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
1974 hard registers as earlyclobber, also if not in an asm.
1976 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1980 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1983 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1984 rs6000_cpu based on cpu_index, not tune_index.
1986 2017-12-04 Richard Biener <rguenther@suse.de>
1988 PR tree-optimization/83238
1989 * graphite-scop-detection.c (scop_detection::merge_sese): Make
1990 code match comment, rejecting invalid SESE regions.
1992 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
1994 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
1995 require base operand is a REG_POINTER prior to reload on targets
1996 with non-equivalent space registers.
1998 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
2000 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
2001 (update_specialized_profile): Fix updating of counts.
2002 (perhaps_add_new_callers): Likewise.
2004 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
2007 * x86-tune.def: Remove obsolette FIXMEs.
2008 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
2009 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
2010 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
2012 (X86_TUNE_PAD_RETURNS): Disable for generic.
2014 2017-12-02 Jakub Jelinek <jakub@redhat.com>
2016 PR tree-optimization/83170
2017 PR tree-optimization/83241
2018 * gimple-ssa-store-merging.c
2019 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
2020 gimple_vuse (ins_stmt) in case it has changed.
2021 (imm_store_chain_info::output_merged_store): Likewise.
2023 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
2027 * tree-cfg.c (pass_warn_function_return::execute): Handle
2028 __builtin_ubsan_handle_missing_return like __builtin_unreachable
2029 with BUILTINS_LOCATION.
2033 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
2034 is BLKmode for vector field with vector raw mode, use TYPE_MODE
2035 instead of DECL_MODE.
2037 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
2038 last argument to rtx pointer.
2039 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
2040 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
2041 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
2042 depending on the chosen ISAs.
2043 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
2044 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
2045 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
2047 * config/i386/sse.md (mov<mode>_internal): Likewise.
2048 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
2050 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2052 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
2054 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2056 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
2057 parameters from prototype.
2058 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
2059 parameters. Don't print a comment.
2060 (emit_fusion_gpr_load): Adjust.
2061 (emit_fusion_load_store): Adjust.
2062 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
2063 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
2064 comment on the second line.
2066 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2069 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
2070 rs6000_cpu based on cpu_index, not tune_index.
2072 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2074 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
2075 which_alternative instead of which_alternative + 1.
2076 (output_asm_insn): Print an extra tab if the template is short.
2078 2017-12-01 Jim Wilson <jimw@sifive.com>
2080 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
2082 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
2083 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
2085 * doc/tm.texi: Regenerate.
2087 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
2090 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
2091 whether we can allocate pseudos before trying to fix an address.
2092 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
2093 memory address is indexed or indirect.
2094 (floatuns_<mode>si2_hw2): Likewise.
2096 2017-12-01 Jason Merrill <jason@redhat.com>
2098 * Makefile.in (TAGS): Add c-family/*.cc.
2100 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
2102 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
2103 (call_value_insn): Likewise.
2104 (sibcall_insn): Likewise.
2105 (sibcall_value_insn): Likewise.
2106 (movsi_aarch64): Likewise.
2107 (movdi_aarch64): Likewise.
2108 (add_losym_): Likewise.
2109 (ldr_got_small_): Likewise.
2110 (ldr_got_small_sidi): Likewise.
2111 (ldr_got_small_28k_): Likewise.
2112 (ldr_got_small_28k_sidi): Likewise.
2113 * config/aarch64/aarch64.c (aarch64_print_address_internal):
2114 Move output_addr_const to symbolic case. Add error check.
2116 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2118 * config/s390/predicates.md (plus16_Q_operand): New predicate.
2119 * config/s390/s390.md: Disable MVC merging peephole if it would
2120 disable operand forwarding.
2121 (new peephole2): Split MVCs if it would turn them into up to 2
2124 2017-12-01 Richard Biener <rguenther@suse.de>
2126 PR tree-optimization/83232
2127 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
2128 detection of same access. Instead of breaking the group here
2129 do not consider the duplicate. Add comment explaining real fix.
2131 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
2133 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
2135 2017-12-01 Sudakshina Das <sudi.das@arm.com>
2137 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
2139 2017-12-01 Jakub Jelinek <jakub@redhat.com>
2141 * function.h (struct function): Remove cilk_frame_decl,
2142 is_cilk_function and calls_cilk_spawn fields.
2143 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
2145 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
2146 cilk_elemental field.
2147 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
2148 * target.def: Adjust comment.
2149 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
2150 Don't test cilk_elemental.
2152 PR tree-optimization/83233
2153 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
2154 bswap_stat name for the struct.
2157 * tree.h (SWITCH_BREAK_LABEL_P): Define.
2158 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
2159 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
2160 SWITCH_BREAK_LABEL_P set on the label.
2161 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
2162 added for default case if it was missing and not all cases covered.
2163 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
2164 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
2166 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
2167 to avoid -Wimplicit-fallthrough warning.
2168 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
2169 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
2171 PR tree-optimization/83221
2172 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
2174 (init_reassoc): Formatting fix.
2177 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
2178 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
2182 * tree-cfg.c: Include asan.h.
2183 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
2185 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2187 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
2189 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2191 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
2192 vpcmpeqd instruction.
2194 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2196 * config/i386/i386.c (standard_sse_constant_opcode): Fix
2197 registers type for 128bit mode.
2199 2017-11-30 Jakub Jelinek <jakub@redhat.com>
2201 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
2203 * gengtype-state.c (read_state_pair): Likewise.
2204 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
2205 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
2206 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
2207 * ifcvt.c (noce_try_store_flag_constants): Likewise.
2208 * tree-ssa-ccp.c (ccp_finalize): Likewise.
2209 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
2210 * builtins.c (fold_builtin_3): Likewise.
2211 * graphite-scop-detection.c
2212 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
2213 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
2216 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
2219 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
2220 parameter. Return true if ignore_decl_rtl_set_p is true and other
2221 conditions are satisfied.
2222 * asan.h (asan_protect_global): Add new parameter.
2223 * varasm.c (categorize_decl_for_section): Pass true as second parameter
2224 to asan_protect_global calls.
2226 2017-11-30 Jim Wilson <jimw@sifive.com>
2228 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
2229 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
2230 -mstrict-align, add info on default value. Delete redundant lines for
2231 -mabi. Add missing -mexplicit-relocs docs.
2233 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2235 * config/arc/arc.md (trap): New pattern.
2237 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2239 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
2240 instruction to end into a delay slot.
2241 * config/arc/arc.md (cond_delay_insn): Check if the instruction
2242 can be placed into a delay slot against reg_note.
2244 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2246 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
2247 labels number of usages.
2249 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2251 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
2253 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
2255 2017-11-30 Jakub Jelinek <jakub@redhat.com>
2258 * internal-fn.c (expand_mul_overflow): Optimize unsigned
2259 multiplication by power of 2 constant into two shifts + comparison.
2261 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
2264 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
2266 2017-11-30 Richard Biener <rguenther@suse.de>
2268 PR tree-optimization/83202
2269 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2270 allow_peel argument and guard peeling.
2271 (canonicalize_loop_induction_variables): Likewise.
2272 (canonicalize_induction_variables): Pass false.
2273 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
2274 peeling from cunrolli.
2276 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
2278 * combine.c (try_combine): Print a message to dump file whenever
2279 I0, I1, or I2 cannot be combined into I3.
2281 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
2283 PR rtl-optimization/83156
2284 PR rtl-optimization/82621
2285 * combine.c (try_combine): Don't split an I2 if one of the dests is
2286 set again before I3. Allow unused dests.
2288 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
2290 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
2292 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
2294 PR rtl-optimization/80818
2295 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
2296 recursively. Use insn code for clobber.
2297 (lra_set_insn_recog_data): Pass the new arg to
2298 collect_non_operand_hard_regs.
2299 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
2301 (lra_update_insn_regno_info): Pass insn to
2302 add_regs_to_insn_regno_info.
2304 2017-11-29 Jim Wilson <jimw@sifive.com>
2305 Andrew Waterman <andrew@sifive.com>
2307 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
2308 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
2309 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
2310 use SINGLE_SHIFT_COST.
2311 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
2312 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
2314 2017-11-29 Julia Koval <julia.koval@intel.com>
2316 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
2317 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
2318 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
2319 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
2320 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
2321 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
2322 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
2323 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
2324 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
2325 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
2326 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
2327 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
2328 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
2329 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
2330 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
2331 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
2332 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
2333 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
2334 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
2335 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
2336 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
2337 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
2338 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
2339 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
2340 __builtin_ia32_vpshldv_v2di_mask,
2341 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
2342 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
2343 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
2345 2017-11-29 Julia Koval <julia.koval@intel.com>
2347 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
2348 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
2349 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
2350 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
2351 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
2352 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
2353 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
2354 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
2355 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
2356 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
2357 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
2358 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
2359 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
2360 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
2361 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
2362 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
2363 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
2364 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
2365 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2366 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
2367 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
2369 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2371 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
2372 movsi_pic_gotdata_op instruction as a load for the UT699 errata
2375 2017-11-29 Martin Aberg <maberg@gaisler.com>
2377 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
2378 to prevent b2bst errata sequence.
2379 (sqrtdf2_fix): Likewise.
2381 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2383 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
2384 (div_sqrt_insn_p): New function.
2385 (sparc_do_work_around_errata): Insert NOP instructions to
2386 prevent sequences that could trigger the TN-0013 errata for
2387 certain LEON3 processors.
2388 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
2389 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
2390 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
2391 (in_branch_delay): Prevent div and sqrt in delay slot if
2393 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
2395 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2397 * config/sparc/sparc.c (atomic_insn_p): New function.
2398 (sparc_do_work_around_errata): Insert NOP instructions to
2399 prevent sequences that could trigger the TN-0010 errata for
2401 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
2402 instruction referable in atomic_insns_p.
2404 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2406 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
2407 (atomic_compare_and_swap_leon3_1): Likewise.
2410 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2412 * config/sparc/sparc.c (fpop_insn_p): New function.
2413 (sparc_do_work_around_errata): Insert NOP instructions to
2414 prevent sequences that could trigger the TN-0012 errata for
2416 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
2417 * config/sparc/sparc.md (fix_gr712rc): New attribute.
2418 (in_branch_annul_delay): Prevent floating-point instructions
2419 in delay slot of annulled integer branch.
2421 2017-11-29 Richard Biener <rguenther@suse.de>
2423 PR tree-optimization/83202
2424 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
2426 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
2427 nodes vectorized to the same stmts multiple times.
2428 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
2429 (vect_analyze_slp_instance): Adjust.
2430 (scalar_stmts_to_slp_tree_map_t): New typedef.
2431 (vect_schedule_slp_instance): Add a map recording the SLP node
2432 representing the vectorized stmts for a set of scalar stmts.
2433 Avoid code-generating redundancies.
2434 (vect_schedule_slp): Allocate map and pass it down.
2436 2017-11-29 Nathan Sidwell <nathan@acm.org>
2439 * tree.c (build_complex_type): Fix canonicalization. Only fill in
2442 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
2444 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
2445 cases for printing LDP/STP memory addresses.
2446 (aarch64_print_address_internal): Renamed from
2447 aarch64_print_operand_address, added parameter, add Pmode check.
2448 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
2449 (aarch64_print_operand_address): Indirect to
2450 aarch64_print_address_internal.
2451 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
2452 'y' operand output specifier.
2454 2017-11-29 Jakub Jelinek <jakub@redhat.com>
2457 * tree.c (build_simple_mem_ref_loc): Handle
2458 get_addr_base_and_unit_offset returning a MEM_REF.
2461 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
2464 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
2467 2017-11-28 David Malcolm <dmalcolm@redhat.com>
2469 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
2471 (layout::move_to_column): Likewise.
2473 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2476 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
2477 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
2479 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2480 Martin Jambor <mjambor@suse.cz>
2483 * tree.h (expr_type_first_operand_type_p): Declare
2484 * tree.c (expr_type_first_operand_type_p): New function.
2485 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
2486 (ipa_value_from_jfunc): Adjust declaration.
2487 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
2488 Use it as result type for arithmetics, unless it is NULL in which case
2489 be more conservative.
2490 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
2491 ipa_get_jf_pass_through_result.
2492 (propagate_vals_across_pass_through): Likewise.
2493 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
2494 is to propagate_vals_across_pass_through.
2495 (propagate_constants_across_call): Pass PARM_TYPE to
2496 propagate_scalar_across_jump_function.
2497 (find_more_scalar_values_for_callers_subset): Pass parameter type to
2498 ipa_value_from_jfunc.
2499 (cgraph_edge_brings_all_scalars_for_node): Likewise.
2500 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
2501 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
2502 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
2503 target_type, pass it to ipa_value_from_jfunc.
2504 (update_indirect_edges_after_inlining): Pass parameter type to
2505 try_make_edge_direct_simple_call.
2507 2017-11-28 Jeff Law <law@redhat.com>
2509 * gimple-ssa-evrp-analyze.c
2510 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
2511 refine ranges if scev_initialized_p returns true.
2512 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
2514 2017-11-28 Julia Koval <julia.koval@intel.com>
2516 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
2517 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
2518 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
2519 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
2520 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
2521 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
2522 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
2523 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
2524 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
2525 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
2526 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
2527 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
2528 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
2529 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
2530 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
2531 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
2532 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
2533 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
2534 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
2535 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
2536 __builtin_ia32_vpshrd_v2di_mask): New builtins.
2537 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
2539 2017-11-28 Julia Koval <julia.koval@intel.com>
2541 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
2542 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
2543 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
2544 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
2545 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
2546 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
2547 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
2548 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
2549 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
2550 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
2551 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
2552 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
2553 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
2554 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
2555 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
2556 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
2557 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
2558 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
2559 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
2560 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
2561 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
2562 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
2563 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
2564 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
2565 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
2566 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
2567 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
2568 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
2569 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
2570 __builtin_ia32_vpshld_v2di_mask): New builtins.
2571 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2572 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
2574 2017-11-28 Richard Biener <rguenther@suse.de>
2576 PR tree-optimization/80776
2577 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
2579 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
2581 (evrp_range_analyzer::record_ranges_from_incoming_edges):
2582 If the incoming edge is an effective fallthru because the other
2583 edge only reaches a __builtin_unreachable () then record ranges
2584 derived from the controlling condition in SSA info.
2585 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
2586 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
2588 2017-11-28 Olivier Hainque <hainque@adacore.com>
2590 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
2591 on mingw build hosts.
2593 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2596 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
2598 2017-11-28 Richard Biener <rguenther@suse.de>
2601 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
2602 copies generated from memcpy use a character array as reference
2605 2017-11-28 Julia Koval <julia.koval@intel.com>
2606 Sebastian Peryt <sebastian.peryt@intel.com>
2608 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
2609 c-family/c-cilkplus.o, c-family/array-notation-common.o,
2610 cilk-common.o, cilk.h, cilk-common.c): Remove.
2612 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
2613 * builtins.c (is_builtin_name): Remove cilkplus condition.
2614 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
2615 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
2616 cilk-builtins.def, cilkplus.def): Remove.
2617 * cif-code.def (CILK_SPAWN): Remove.
2618 * cilk-builtins.def: Delete.
2619 * cilk-common.c: Ditto.
2621 * cilkplus.def: Ditto.
2622 * config/darwin.h (fcilkplus): Delete.
2623 * cppbuiltin.c: Ditto.
2624 * doc/extend.texi: Remove cilkplus doc.
2625 * doc/generic.texi: Ditto.
2626 * doc/invoke.texi: Ditto.
2627 * doc/passes.texi: Ditto.
2628 * gcc.c (fcilkplus): Remove.
2629 * gengtype.c (cilk.h): Remove.
2630 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
2632 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
2634 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
2636 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
2637 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
2638 cilkplus conditions.
2639 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
2640 inline_read_section): Ditto.
2641 * ipa-inline-analysis.c (cilk.h): Remove.
2642 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
2643 * lto-wrapper.c (merge_and_complain, append_compiler_options,
2644 append_linker_options): Remove condition for fcilkplus.
2645 * lto/lto-lang.c (cilk.h): Remove.
2646 (lto_init): Remove condition for fcilkplus.
2647 * omp-expand.c (expand_cilk_for_call): Delete.
2648 (expand_omp_taskreg, expand_omp_for_static_chunk,
2649 expand_omp_for): Remove cilkplus
2651 (expand_cilk_for): Delete.
2652 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
2653 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
2654 execute_lower_omp, diagnose_sb_0): Ditto.
2655 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
2656 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2657 * tree-nested.c: Ditto.
2658 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
2659 (dump_generic_node): Ditto.
2660 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2661 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
2663 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
2665 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2667 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
2668 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
2669 for vector mode and !TARGET_SIMD.
2671 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2673 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
2675 * tree.h (SWITCH_LABELS): Remove.
2676 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
2677 assert SWITCH_BODY is non-NULL.
2678 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
2680 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
2682 PR tree-optimization/80788
2683 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
2684 has TREE_OVERFLOW set, call drop_tree_overflow.
2686 2017-11-28 Richard Biener <rguenther@suse.de>
2688 PR tree-optimization/83158
2689 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
2691 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
2694 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
2695 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
2697 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
2699 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
2700 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
2702 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2705 * varasm.c (compare_constant): Take the mode of the constants into
2706 account when comparing floating point constants.
2708 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
2710 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
2711 from explicit instantiation of debug_helper.
2712 * vec.h (DEFINE_DEBUG_VEC): Ditto.
2714 2017-11-27 Richard Biener <rguenther@suse.de>
2716 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
2719 2017-11-27 Richard Biener <rguenther@suse.de>
2721 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
2722 (type_hash_canon): Also clear int_cst_hash_table entry for
2724 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
2726 2017-11-27 Tamar Christina <tamar.christina@arm.com>
2728 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2729 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2730 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2732 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2734 * hash-map.h (gt_cleare_cache): Avoid UB.
2736 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
2738 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
2739 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
2740 (decide_unroll_runtime_iterations): Likewise.
2741 (decide_unroll_stupid): Likewise.
2743 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2746 * config/i386/i386.md: Add a loop with incssp.
2748 2017-11-27 Martin Jambor <mjambor@suse.cz>
2750 PR tree-optimization/81248
2751 * tree-sra.c (splice_param_accesses): Remove size check.
2752 (decide_one_param_reduction): Fix size check.
2753 * gimple-pretty-print.c (dump_profile): Silence warning.
2754 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
2756 2017-11-27 Jakub Jelinek <jakub@redhat.com>
2759 * dbxout.c (lastlineno): New variable.
2760 (dbx_debug_hooks): Use dbxout_switch_text_section as
2761 switch_text_section debug hook.
2762 (dbxout_function_end): Switch to current_function_section
2763 rather than function_section. If crtl->has_bb_partition,
2764 output just one N_FUN, depending on in_cold_section_p.
2765 (dbxout_source_line): Remember last lineno in lastlineno.
2766 (dbxout_switch_text_section): New function.
2767 (dbxout_function_decl): Adjust dbxout_block caller.
2768 (dbx_block_with_cold_children): New function.
2769 (dbxout_block): Return true if any LBRAC/RBRAC have been
2770 emitted. Use dbx_block_with_cold_children at depth == 0
2771 in second partition. Add PARENT_BLOCKNUM argument, pass
2772 it optionally adjusted to children. Output LBRAC/RBRAC
2773 around recursive call only if the block is in the current
2774 partition, if not and anything was output, emit empty
2776 * final.c (final_scan_insn): Compute cold_function_name
2777 before calling switch_text_section debug hook. Call
2778 that hook even if dwarf2out_do_frame if not emitting
2782 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
2783 TREE_READONLY decls.
2785 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2787 PR rtl-optimization/82488
2788 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
2790 2017-11-26 Julia Koval <julia.koval@intel.com>
2792 * config/i386/i386.c (processor_target_table): Add skylake_cost for
2794 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
2797 2017-11-26 Julia Koval <julia.koval@intel.com>
2799 * config/i386/driver-i386.c (host_detect_local_cpu):
2800 Detect skylake-avx512.
2802 2017-11-26 Julia Koval <julia.koval@intel.com>
2804 * config.gcc: Add -march=cannonlake.
2805 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
2806 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
2807 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
2808 (PTA_CANNONLAKE): New.
2809 (processor_target_table): Add cannonlake.
2810 (ix86_option_override_internal): Ditto.
2811 (fold_builtin_cpu): Ditto.
2812 (get_builtin_code_for_version): Handle cannonlake.
2813 (M_INTEL_COREI7_CANNONLAKE): New.
2814 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
2815 * doc/invoke.texi: Add -march=cannonlake.
2817 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
2819 * plugin.c (add_new_plugin): Use platform-specific library extensions.
2820 (try_init_one_plugin): Alternative implementation for MinGW.
2821 * Makefile.in (plugin_implib): New.
2822 (gengtype-lex.c): Fix broken AIX workaround.
2823 * configure: Regenerate.
2824 * doc/plugins.texi: Document support for MinGW.
2826 2017-11-25 Jakub Jelinek <jakub@redhat.com>
2828 PR rtl-optimization/81553
2829 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
2830 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
2831 is a shift where C1 has different mode than the whole shift, use C1's
2832 mode for MULT rather than the shift's mode.
2835 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
2836 builtins not enabled in the currently selected ISA.
2838 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
2840 PR tree-optimization/71026
2841 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
2842 (insert_reciprocals): Change to insert reciprocals before a division
2843 by a square and to insert the square of a reciprocal.
2844 (execute_cse_reciprocals_1): Change to consider division by a square.
2845 (register_division_in): Add importance parameter.
2847 2017-11-24 Richard Biener <rguenther@suse.de>
2849 PR tree-optimization/82402
2850 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
2851 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
2853 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
2855 * match.pd (0-ptr): New transformation.
2857 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
2860 * ipa-inline.c (inline_small_functions): Set current badnes correctly
2861 when skipping checking.
2863 2017-11-24 Richard Biener <rguenther@suse.de>
2865 PR tree-optimization/83128
2866 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
2867 (vn_reference_lookup_3): Likewise.
2869 2017-11-24 Jakub Jelinek <jakub@redhat.com>
2872 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
2873 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
2876 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
2877 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
2878 check for the latter with gimple_call_builtin_p. Do not handle
2879 BUILT_IN_STPNCPY_CHK which is not a pass through call.
2881 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
2883 * config/arm/arm_neon.h: Fix pragma GCC push_options before
2886 2017-11-23 Julia Koval <julia.koval@intel.com>
2888 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
2889 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
2890 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
2891 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
2892 _mm512_maskz_expandloadu_epi16): New intrinsics.
2893 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
2894 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
2895 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
2896 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
2897 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
2898 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
2899 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
2900 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
2901 _mm256_maskz_expandloadu_epi8): New intrinsics.
2902 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
2903 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
2904 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
2905 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
2906 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
2907 * config/i386/sse.md (VI248_VLBW): New iterator.
2908 (expand<mode>_mask, expand<mode>_maskz): New patterns.
2910 2017-11-23 Julia Koval <julia.koval@intel.com>
2912 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
2913 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
2914 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
2915 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
2916 _mm512_mask_compressstoreu_epi16): New.
2917 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
2918 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
2919 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
2920 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
2921 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
2922 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
2923 _mm256_mask_compressstoreu_epi8): New.
2924 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
2925 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
2926 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
2927 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
2928 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
2929 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
2930 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
2931 __builtin_ia32_compresshi128_mask,
2932 __builtin_ia32_compressstoreuqi512_mask,
2933 __builtin_ia32_compressstoreuhi512_mask,
2934 __builtin_ia32_compressstoreuqi256_mask,
2935 __builtin_ia32_compressstoreuqi128_mask,
2936 __builtin_ia32_compressstoreuhi256_mask,
2937 __builtin_ia32_compressstoreuhi128_mask): New builtins.
2938 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
2940 (ix86_expand_special_args_builtin): Handle new types.
2941 (s4fma_expand): Handle new builtin array.
2942 * config/i386/immintrin.h: Include new headers.
2943 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
2944 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
2946 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2948 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
2949 fixes. Declare temp and g variables at the top in order to avoid
2950 {} in most of the cases.
2952 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
2954 * match.pd (ptr-0): New transformation.
2956 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2958 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
2959 (struct addr_mode_cost_table): New.
2960 (struct tune_params): Add field addr_mode_costs.
2961 * config/arm/arm.c (generic_addr_mode_costs): New.
2962 (arm_slowmul_tune): Initialise addr_mode_costs field.
2963 (arm_fastmul_tune): Likewise.
2964 (arm_strongarm_tune): Likewise.
2965 (arm_xscale_tune): Likewise.
2966 (arm_9e_tune): Likewise.
2967 (arm_marvell_pj4_tune): Likewise.
2968 (arm_v6t2_tune): Likewise.
2969 (arm_cortex_tune): Likewise.
2970 (arm_cortex_a8_tune): Likewise.
2971 (arm_cortex_a7_tune): Likewise.
2972 (arm_cortex_a15_tune): Likewise.
2973 (arm_cortex_a35_tune): Likewise.
2974 (arm_cortex_a53_tune): Likewise.
2975 (arm_cortex_a57_tune): Likewise.
2976 (arm_exynosm1_tune): Likewise.
2977 (arm_xgene1_tune): Likewise.
2978 (arm_cortex_a5_tune): Likewise.
2979 (arm_cortex_a9_tune): Likewise.
2980 (arm_cortex_a12_tune): Likewise.
2981 (arm_cortex_a73_tune): Likewise.
2982 (arm_v7m_tune): Likewise.
2983 (arm_cortex_m7_tune): Likewise.
2984 (arm_v6m_tune): Likewise.
2985 (arm_fa726te_tune): Likewise.
2986 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
2988 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2990 * config/arm/arm.c (arm_mem_costs): New function.
2991 (arm_rtx_costs_internal): Use arm_mem_costs.
2993 2017-11-23 Mark Wielaard <mark@klomp.org>
2995 * dwarf2out.c (init_sections_and_labels): Use generation to create
2996 unique ranges_section_label and ranges_base_label. Return generation.
2997 (output_rnglists): Add generation argument. Use generation to create
2998 unique ranges labels.
2999 (dwarf2out_finish): Get generation from init_sections_and_labels
3000 and pass generation to output_rnglists.
3002 2017-11-23 Mike Stump <mikestump@comcast.net>
3003 Eric Botcazou <ebotcazou@adacore.com>
3005 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
3006 * cfgloop.h (struct loop): Add unroll field.
3007 * function.h (struct function): Add has_unroll bitfield.
3008 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
3009 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
3010 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
3011 (pass_rtl_unroll_loops::gate): Likewise.
3012 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
3013 for which loop->unroll==1.
3014 (decide_unroll_constant_iterations): Use note for consistency and
3015 take loop->unroll into account. Return early if loop->unroll is set.
3016 Fix thinko in existing test.
3017 (decide_unroll_runtime_iterations): Use note for consistency and
3018 take loop->unroll into account.
3019 (decide_unroll_stupid): Likewise.
3020 * lto-streamer-in.c (input_cfg): Read loop->unroll.
3021 * lto-streamer-out.c (output_cfg): Write loop->unroll.
3022 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
3024 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
3025 (print_loop): Print loop->unroll if set.
3026 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
3027 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
3028 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
3030 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
3031 loop->unroll is set and smaller than the trip count. Otherwise bypass
3032 entirely the heuristics if loop->unroll is set. Remove dead note.
3033 Fix off-by-one bug in other note.
3034 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
3035 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
3037 (tree_unroll_loops_completely): Make static.
3038 (pass_complete_unroll::execute): Use correct type for variable.
3039 (pass_complete_unrolli::execute): Fix formatting.
3040 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
3042 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
3044 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
3045 enable when TARGET_PREFER_AVX128 is set.
3047 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3049 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
3051 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
3052 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
3054 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3056 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
3057 frequencies. Fix estimation of aggregate parameters.
3059 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3061 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
3064 2017-11-23 Tom de Vries <tom@codesourcery.com>
3066 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
3068 2017-11-23 Tom de Vries <tom@codesourcery.com>
3070 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
3073 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
3076 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
3077 sibcall_value_pcrel_fdpic): Use local variable instead of
3079 (calli_tbr_rel): Add missing operand 2.
3080 (call_valuei_tbr_rel): Add missing operand 3.
3082 2017-11-23 Jakub Jelinek <jakub@redhat.com>
3085 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
3086 bitpos/bitsize covering the whole destination, use store_expr only if
3087 the complex mode is the same. Otherwise, use expand_normal and if
3088 it returns CONCAT, subreg each part separately instead of trying to
3089 subreg the whole result.
3091 2017-11-23 Richard Biener <rguenther@suse.de>
3093 PR tree-optimization/23094
3094 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
3095 come here from walking over backedges in the first iteration.
3096 (vn_reference_lookup_3): Skip clobbers that store the same value.
3098 2017-11-23 Richard Biener <rguenther@suse.de>
3100 PR tree-optimization/81403
3101 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
3102 a block we need a leader relative to.
3103 (phi_translate_1): For nary processing require a leader from
3104 get_representative_for given we run expression simplification
3105 using match-and-simplify. Remove previous fix.
3107 2017-11-22 Jeff Law <law@redhat.com>
3109 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
3110 Use new method allocate_value_range rather than accessing the
3111 vrp_value_range_pool data member directly.
3112 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
3113 to use extract_range_from_stmt method to avoid need for
3114 extract_range_from_assignment method.
3115 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
3116 method rather than setting values_propgated data member directly.
3117 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
3118 and values propagated data members and extract_range_from_assignment
3119 method. Reorder private data members to conform to standards.
3120 Add new methods set_lattice_propagation_complete and
3121 allocate_value_range.
3123 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
3125 PR rtl-optimization/83030
3126 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
3127 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
3128 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
3129 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
3130 CROSSING_JUMP_P flag is already set before setting it.
3131 * cfgrtl.c (fixup_partition_crossing): Likewise.
3132 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
3135 2017-11-22 Jakub Jelinek <jakub@redhat.com>
3137 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
3138 Handle the case where both arguments are using gen_const_vec_series.
3140 2017-11-22 David Malcolm <dmalcolm@redhat.com>
3143 * pretty-print.c (pp_format): Move quoting implementation to
3144 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
3145 to pass address of "quote" local.
3146 (pp_begin_quote): New function.
3147 (pp_end_quote): New function.
3148 * pretty-print.h (printer_fn): Convert penultimate param from bool
3150 (pp_begin_quote): New decl.
3151 (pp_end_quote): New decl.
3152 * tree-diagnostic.c (default_tree_printer): Convert penultimate
3153 param from bool to bool *.
3154 * tree-diagnostic.h (default_tree_printer): Likewise.
3156 2017-11-22 Jeff Law <law@redhat.com>
3158 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
3159 Initialize vr_values.
3160 (evrp_range_analyzer::try_find_new_range): Call methods attached to
3161 vr_values via vr_values class instance rather than delegators.
3162 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
3163 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
3164 (evrp_range_analyzer::push_value_range): Likewise.
3165 (evrp_range_analyzer::pop_value_range): Likewise.
3166 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
3167 most delegators. Those remaining are exposed as public interfaces.
3168 Make vr_values a pointer and private.
3169 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
3171 (evrp_range_analyzer::get_vr_value): New method.
3172 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
3173 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
3174 (class evrp_dom_walker): Attach evrp_folder class, initialize
3175 it in the ctor. Remove temporary delegators.
3176 (evrp_dom_walker::before_dom_children): Call methods in attached
3177 evrp_range_analyzer class via class instance pointer. Use
3178 free value_range_constant_singleton to remove need for
3179 op_with_constant_singleton_value delegator method. Do not
3180 create a vrp_prop class instance for every call! Narrow
3181 scope of a couple variables.
3182 (evrp_dom_walker::cleanup): Call methods in attached
3183 evrp_range_analyzer class via class instance pointer.
3184 * vr-values.h (class vr_values): Privatize many methods and
3187 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3189 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
3190 * cfgexpand.c (expand_debug_expr): Remove handling for them.
3191 * expr.c (expand_expr_real_2): Likewise.
3192 * fold-const.c (const_unop): Likewise.
3193 * optabs-tree.c (optab_for_tree_code): Likewise.
3194 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3195 * tree-inline.c (estimate_operator_cost): Likewise.
3196 * tree-pretty-print.c (dump_generic_node): Likewise.
3197 (op_code_prio): Likewise.
3198 (op_symbol_code): Likewise.
3199 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
3200 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
3201 * internal-fn.c (direct_internal_fn_optab): New function.
3202 (direct_internal_fn_array, direct_internal_fn_supported_p
3203 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
3204 * fold-const-call.c (fold_const_reduction): New function.
3205 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
3207 * tree-vect-loop.c: Include internal-fn.h.
3208 (reduction_code_for_scalar_code): Rename to...
3209 (reduction_fn_for_scalar_code): ...this and return an internal
3211 (vect_model_reduction_cost): Take an internal_fn rather than
3213 (vect_create_epilog_for_reduction): Likewise. Build calls rather
3215 (vectorizable_reduction): Use internal functions rather than tree
3216 codes for the reduction operation. Update calls to the functions
3218 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
3219 Use calls to internal functions rather than REDUC tree codes.
3220 * config/aarch64/aarch64-simd.md: Update comment accordingly.
3222 2017-11-22 Olivier Hainque <hainque@adacore.com>
3224 * config/vxworks.c (vxworks_override_options): Pick default
3225 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
3226 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
3227 DWARF_GNAT_ENCODINGS_DEFAULT.
3228 * config/vxworksae.h: Likewise.
3230 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
3232 PR tree-optimization/83104
3233 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
3236 2017-11-22 Marek Polacek <polacek@redhat.com>
3237 H.J. Lu <hongjiu.lu@intel.com>
3238 Jason Merrill <jason@redhat.com>
3243 * calls.c (initialize_argument_information): Call
3244 warn_parameter_passing_abi target hook.
3245 (store_one_arg): Use 0 for empty record size. Don't push 0 size
3246 argument onto stack.
3247 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
3248 * common.opt: Update -fabi-version description.
3249 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
3250 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
3252 (ix86_is_empty_record): New function.
3253 (ix86_warn_parameter_passing_abi): New function.
3254 (TARGET_EMPTY_RECORD_P): Redefine.
3255 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
3256 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
3257 * doc/tm.texi: Regenerated.
3258 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
3259 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
3260 * dwarf2out.c (get_ultimate_context): Move to tree.c.
3261 * explow.c (hard_function_value): Call arg_int_size_in_bytes
3262 instead of int_size_in_bytes.
3263 * expr.c (copy_blkmode_to_reg): Likewise.
3264 * function.c (aggregate_value_p): Return 0 for empty types.
3265 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
3267 (locate_and_pad_parm): Call arg size_in_bytes instead
3269 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
3270 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
3271 * target.def (empty_record_p, warn_parameter_passing_abi): New target
3273 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
3274 (std_gimplify_va_arg_expr): Skip empty records. Call
3275 arg_size_in_bytes instead size_in_bytes.
3276 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
3277 * tree-core.h (tree_type_common): Add empty_flag.
3278 (tree_decl_common): Update comments.
3279 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
3281 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
3282 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
3284 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
3285 * tree.c (default_is_empty_type): New function.
3286 (default_is_empty_record): New function.
3287 (arg_int_size_in_bytes): New function.
3288 (arg_size_in_bytes): New function.
3289 (get_ultimate_context): New function.
3290 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
3291 TRANSLATION_UNIT_WARN_EMPTY_P.
3292 (default_is_empty_record, arg_int_size_in_bytes,
3293 arg_size_in_bytes, get_ultimate_context): Declare.
3295 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
3297 * config/arm/arm.c (cmse_clear_registers): New function.
3298 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
3299 code by call to cmse_clear_registers.
3300 (cmse_nonsecure_entry_clear_before_return): Likewise.
3302 2017-11-22 Tamar Christina <tamar.christina@arm.com>
3304 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
3305 (vdot_s32, vdotq_s32): New.
3306 (vdot_lane_u32, vdotq_lane_u32): New.
3307 (vdot_lane_s32, vdotq_lane_s32): New.
3310 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3313 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
3314 for unsigned values with fewer HWIs than the precision.
3315 (test_overflow): New function.
3316 (wide_int_cc_tests): Call it.
3318 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3320 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
3321 has the same mode class as Pmode.
3323 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
3325 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
3326 padding_bits_to_clear_ptr.
3327 (cmse_nonsecure_entry_clear_before_return): Likewise.
3329 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
3331 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
3332 auto_sbitap instead of integer bitfield to control register needing
3335 2017-11-22 Jakub Jelinek <jakub@redhat.com>
3337 PR tree-optimization/83044
3338 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
3339 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
3340 that need the upper bound. Subtract offset from
3341 get_addr_base_and_unit_offset only if positive and subtract it
3342 before division by eltsize rather than after it.
3345 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
3346 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
3347 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
3351 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
3352 expand_binop with *mul_widen_optab, make sure at least one of the
3353 operands doesn't have VOIDmode.
3356 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
3358 PR rtl-optimization/82044
3359 PR tree-optimization/82042
3360 * dse.c (record_store): Check for overflow.
3361 (check_mem_read_rtx): Properly check for overflow if width == -1, call
3362 add_wild_read instead of clear_rhs_from_active_local_stores on
3363 overflow and log it into dump_file.
3365 2017-11-22 Richard Biener <rguenther@suse.de>
3367 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
3368 fake edges to exit when looking for a place to insert.
3369 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
3371 (insert_into_preds_of_block): Commit edge insertion immediately,
3372 assert that doesn't require new BBs.
3373 (fini_pre): Release expressions.
3374 (pass_pre::execute): Shuffle things around a bit, if the fn
3375 is too large do not compute AVAIL either as this is really the
3378 2017-11-22 Richard Biener <rguenther@suse.de>
3380 PR tree-optimization/83089
3381 * tree-if-conv.c (pass_if_conversion::execute): If anything
3382 changed reset SCEV and free the number of iteration estimates.
3384 2017-11-21 Martin Sebor <msebor@redhat.com>
3386 PR tree-optimization/82945
3387 * calls.h (warn_nonstring_bound): Remove unused function.
3389 2017-11-21 Martin Sebor <msebor@redhat.com>
3391 PR tree-optimization/82945
3392 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
3393 * calls.h (maybe_warn_nonstring_arg): Declare new function.
3394 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
3396 (initialize_argument_information): Call maybe_warn_nonstring_arg.
3397 * calls.h (get_attr_nonstring_decl): Declare new function.
3398 * doc/extend.texi (attribute nonstring): Update.
3399 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
3400 get_attr_nonstring_decl and handle it.
3401 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
3402 detection of nul-termination.
3403 (strlen_to_stridx): Change to a pointer.
3404 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
3405 (pass_strlen::execute): Same.
3407 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
3409 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
3410 for the new option -mprefer-vector-width=[none|128|256|512].
3411 * config/i386/i386.c (ix86_target_string): remove old style options
3412 -mprefer-avx256 and make -mprefer-avx128 as alias.
3413 (ix86_option_override_internal): Apply defaults for the
3414 -mprefer-vector-width=[128|256] option.
3415 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
3416 Implement macros to work with -mprefer-vector-width=.
3417 * config/i386/i386.opt: Implemented option
3418 -mprefer-vector-width=[none|128|256|512].
3419 * doc/invoke.texi: Documentation for
3420 -mprefer-vector-width=[none|128|256|512].
3422 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
3424 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
3426 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
3428 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
3430 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
3432 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
3434 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
3435 * expr.c (expand_expr_real_2): Likewise.
3436 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
3437 fold_binary_loc): Likewise.
3438 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
3439 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
3440 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
3441 MINUS_EXPR transformations.
3442 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
3443 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
3444 * tree-inline.c (estimate_operator_cost): Likewise.
3445 * tree-pretty-print.c (dump_generic_node, op_code_prio,
3446 op_symbol_code): Likewise.
3447 * tree-vect-stmts.c (vectorizable_operation): Likewise.
3448 * vr-values.c (extract_range_from_binary_expr): Likewise.
3449 * varasm.c (initializer_constant_valid_p_1): Likewise.
3450 * tree.def: New tree code POINTER_DIFF_EXPR.
3452 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
3454 * config/i386/i386.md (*bswap<mode>2_movbe): Add
3455 integer suffix to movbe mnemonic.
3456 (*bswaphi2_movbe): Ditto.
3457 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
3459 2017-11-21 Jakub Jelinek <jakub@redhat.com>
3462 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
3463 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
3464 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
3466 2017-11-21 Martin Liska <mliska@suse.cz>
3468 * tree-inline.c (expand_call_inline): Remove not needed
3471 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
3472 Jakub Jelinek <jakub@redhat.com>
3475 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
3476 Remove static keyword from f variable.
3478 2017-11-21 Jakub Jelinek <jakub@redhat.com>
3480 PR tree-optimization/83086
3481 * gimple-ssa-store-merging.c
3482 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
3483 rather than n.base_addr.
3485 2017-11-21 Martin Liska <mliska@suse.cz>
3487 PR rtl-optimization/82044
3488 PR tree-optimization/82042
3489 * dse.c (check_mem_read_rtx): Check for overflow.
3491 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
3493 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
3496 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
3498 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
3499 and its corresponding call arg location note.
3501 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
3502 Andrew Burgess <andrew.burgess@embecosm.com>
3504 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
3506 (arc_return_slot_offset): Likewise.
3507 (arc_eh_return_address_location): New declaration.
3508 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
3509 (MUST_SAVE_REGISTER): Add exception handler case.
3510 (MUST_SAVE_RETURN_ADDR): Likewise.
3511 (arc_frame_pointer_required): Likewise.
3512 (arc_frame_pointer_needed): New function.
3513 (arc_compute_frame_size): Changed.
3514 (arc_expand_prologue): Likewise.
3515 (arc_expand_epilogue): Likewise.
3516 (arc_initial_elimination_offset): Likewise.
3517 (arc_return_slot_offset): Delete.
3518 (arc_eh_return_address_location): New function.
3519 (arc_builtin_setjmp_frame_value): Likewise.
3520 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
3521 (EH_RETURN_STACKADJ_RTX): Define.
3522 (EH_RETURN_HANDLER_RTX): Likewise.
3523 * config/arc/arc.md (eh_return): Delete.
3525 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
3527 * print-tree.h (debug_vec_tree): Remove prototype.
3528 * gdbinit.in (pvt): Remove macro.
3530 2017-11-21 Jakub Jelinek <jakub@redhat.com>
3532 PR tree-optimization/83047
3533 * gimple-ssa-store-merging.c
3534 (imm_store_chain_info::output_merged_store): If the loads with the
3535 same vuse are in different basic blocks, for load_gsi pick a load
3536 location that is dominated by the other loads.
3539 * config/i386/i386.c (ix86_memmodel_check): Start
3540 -Winvalid-memory-model diagnostics with lowercase letter.
3543 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
3544 set in_cold_section_p to the partition containing loc_list->first.
3545 When seeing loc_list->last_before_switch node, update secname and
3546 perform range_across_switch second partition handling only after that.
3549 * run-rtl-passes.c: Include debug.h.
3550 (run_rtl_passes): Call debug_hooks->assembly_start.
3551 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
3555 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
3556 OPTAB_DIRECT in calls to expand_simple_binop.
3558 2017-11-20 David Malcolm <dmalcolm@redhat.com>
3561 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
3563 2017-11-20 David Malcolm <dmalcolm@redhat.com>
3566 * spellcheck.h (best_match::blithely_get_best_candidate): New
3569 2017-11-20 Jakub Jelinek <jakub@redhat.com>
3571 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
3572 with lower case letter.
3574 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
3576 * config/i386/i386.md (bswaphi2): New expander.
3577 (*bswaphi2_movbe): New insn pattern.
3578 (bswaphi -> rorhi pepehole2): New peephole pattern.
3580 2017-11-20 Jeff Law <law@redhat.com>
3582 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
3583 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
3584 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
3585 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
3586 gimple-ssa-evrp-analyze.h.
3588 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
3589 set BB_VISITED here.
3590 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
3592 2017-11-20 Bin Cheng <bin.cheng@arm.com>
3594 * tree-predcom.c: Add general comment on Store-Store chains.
3595 (split_data_refs_to_components): Postpone clearing eliminate_store_p
3597 (get_chain_last_ref_at): Rename into...
3598 (get_chain_last_write_at): ...this.
3599 (get_chain_last_write_before_load): New function.
3600 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
3601 CT_STORE_STORE when write reference is added.
3602 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
3603 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
3604 (initialize_root_vars_store_elim_1): Ditto.
3605 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
3606 definition is created.
3607 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
3608 chain by replacing it with dominant stored value.
3610 2017-11-20 Bin Cheng <bin.cheng@arm.com>
3612 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
3614 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
3616 * vr-values.c (extract_range_from_binary_expr): Use a full range
3619 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
3621 * config/arm/arm.md (R4_REGNUM): Define constant.
3622 (nonsecure_call_internal): Remove r4 clobber.
3623 (nonsecure_call_value_internal): Likewise.
3624 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
3625 clobber and resequence match_operands.
3626 (nonsecure_call_value_reg_thumb1_v5): Likewise.
3627 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
3628 (nonsecure_call_value_reg_thumb2): Likewise.
3630 2017-11-20 Jakub Jelinek <jakub@redhat.com>
3632 PR tree-optimization/78821
3633 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
3634 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
3635 to the address of the base rather than the base itself.
3636 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
3637 (find_bswap_or_nop_finalize): New function.
3638 (find_bswap_or_nop): Use it.
3639 (bswap_replace): Return a tree rather than bool, change first
3640 argument from gimple * to gimple_stmt_iterator, allow inserting
3641 into an empty sequence, allow ins_stmt to be NULL - then emit
3642 all stmts into gsi. Fix up MEM_REF address gimplification.
3643 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
3644 (struct store_immediate_info): Add N and INS_STMT non-static
3646 (store_immediate_info::store_immediate_info): Initialize them
3647 from newly added ctor args.
3648 (merged_store_group::apply_stores): Formatting fixes. Sort by
3650 (stmts_may_clobber_ref_p): For stores call also
3651 refs_anti_dependent_p.
3652 (gather_bswap_load_refs): New function.
3653 (imm_store_chain_info::try_coalesce_bswap): New method.
3654 (imm_store_chain_info::coalesce_immediate_stores): Use it.
3655 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
3656 (imm_store_chain_info::output_merged_store): Fail if number of
3657 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
3658 and NOP_EXPR rhs_code.
3659 (pass_store_merging::process_store): Compute n and ins_stmt, if
3660 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
3661 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
3663 (pass_store_merging::execute): Calculate dominators.
3665 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
3666 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
3667 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
3668 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3669 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
3670 class pass_optimize_bswap, bswap_replace,
3671 pass_optimize_bswap::execute): Moved to ...
3672 * gimple-ssa-store-merging.c: ... this file.
3673 Include optabs-tree.h.
3674 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
3675 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3676 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
3677 anonymous namespace, remove static keywords.
3678 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
3679 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
3681 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
3684 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
3686 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
3688 * vec.h (debug_helper): New function.
3689 (DEFINE_DEBUG_VEC): New macro.
3690 * hash-set.h (debug_helper): New function.
3691 (DEFINE_DEBUG_HASH_SET): New macro.
3692 * cfg.c (debug_slim (edge)): New function.
3693 Call DEFINE_DEBUG_VEC for edges.
3694 Call DEFINE_DEBUG_HASH_SET for edges.
3695 * cfghooks.c (debug_slim (basic_block)): New function.
3696 Call DEFINE_DEBUG_VEC for basic blocks.
3697 Call DEFINE_DEBUG_HASH_SET for basic blocks.
3698 * print-tree.c (debug_slim): New function to handle trees.
3699 Call DEFINE_DEBUG_VEC for trees.
3700 Call DEFINE_DEBUG_HASH_SET for trees.
3701 (debug (vec<tree, va_gc>) &): Remove.
3702 (debug (<vec<tree, va_gc>) *): Remove.
3703 * print-rtl.c (debug_slim): New function to handle const_rtx.
3704 Call DEFINE_DEBUG_VEC for rtx_def.
3705 Call DEFINE_DEBUG_VEC for rtx_insn.
3706 Call DEFINE_DEBUG_HASH_SET for rtx_def.
3707 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
3708 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
3709 (debug (vec<rtx_insn *> *ptr): Remove.
3710 (debug_insn_vector): Remove.
3711 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
3713 2017-11-20 Tom de Vries <tom@codesourcery.com>
3715 PR rtl-optimization/82020
3716 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
3717 IF_THEN_ELSE condition.
3719 2017-11-19 Jeff Law <law@redhat.com>
3721 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
3722 of degenerates resulting from ignoring an edge.
3724 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3727 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
3729 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3732 * profile-count.c (profile_count::to_sreal_scale): Fix return value
3733 for uninitialied counts.
3735 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3738 * tree-inline.c (estimate_num_insns): Set to 1 at least.
3740 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3743 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
3745 2017-11-19 Tom de Vries <tom@codesourcery.com>
3747 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
3750 2017-11-19 Tom de Vries <tom@codesourcery.com>
3752 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
3755 2017-11-19 Tom de Vries <tom@codesourcery.com>
3757 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
3760 2017-11-19 Tom de Vries <tom@codesourcery.com>
3762 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
3764 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
3766 2017-11-19 Tom de Vries <tom@codesourcery.com>
3768 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
3770 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
3771 semicolon after MCORE_EXPORT_NAME call.
3773 2017-11-19 Tom de Vries <tom@codesourcery.com>
3776 * vmsdbgout.c (vmsdbgout_early_finish): New function.
3777 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
3779 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
3781 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
3782 (cgraph_edge::clone): Cleanup updating of profile.
3783 * ipa-cp.c (update_profiling_info): Likewise.
3784 * ipa-inline-transform.c (inline_transform): Likewise.
3785 * ipa-inline.c (inline_small_functions): Add missing space to dump.
3786 * ipa-split.c (execute_split_functions): Do not split when function
3788 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
3789 * profile-count.c (profile_count::dump): Add global0.
3790 (profile_count::to_cgraph_frequency): Do not ICE when entry is
3792 (profile_count::to_sreal_scale): Likewise.
3793 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
3794 (profile_count::combine_with_ipa_count): New function.
3795 * profile-count.h (profile_guessed_global0adjusted): New.
3796 (profile_count::adjusted_zero): New.
3797 (profile_count::global0adjusted): New.
3798 (profile_count::combine_with_ipa_count): New.
3799 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
3800 correct profile of return block of split functions.
3801 (copy_cfg_body): Remove unused profile_count.
3802 (copy_body): Likewise.
3803 (expand_call_inline): Update.
3804 (tree_function_versioning): Update.
3806 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
3808 * hash-set.h (hash_set::empty): New.
3809 * tree-ssa-threadbackward.h: Delete.
3810 * tree-ssa-threadbackward.c (class thread_jumps): New.
3811 Move max_threaded_paths into class.
3812 (fsm_find_thread_path): Remove arguments that are now in class.
3813 (profitable_jump_thread_path): Rename to...
3814 (thread_jumps::profitable_jump_thread_path): ...this.
3815 (convert_and_register_jump_thread_path): Rename to...
3816 (thread_jumps::convert_and_register_current_path): ...this.
3817 (check_subpath_and_update_thread_path): Rename to...
3818 (thread_jumps::check_subpath_and_update_thread_path): ...this.
3819 (register_jump_thread_path_if_profitable): Rename to...
3820 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
3821 (handle_phi): Rename to...
3822 (thread_jumps::handle_phi): ...this.
3823 (handle_assignment): Rename to...
3824 (thread_jumps::handle_assignment): ...this.
3825 (fsm_find_control_statement_thread_paths): Rename to...
3826 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
3827 (find_jump_threads_backwards): Rename to...
3828 (thread_jumps::find_jump_threads_backwards): ...this.
3829 Initialize path local data.
3830 (pass_thread_jumps::execute): Call find_jump_threads_backwards
3831 from within thread_jumps class.
3832 (pass_early_thread_jumps::execute): Same.
3834 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3836 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
3838 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3840 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
3841 to canonical location.
3843 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3846 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
3848 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
3850 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
3852 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
3853 rcrt1.o%s/grcrt1.o%s for -static-pie.
3855 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3857 * i386.c (ix86_multiplication_cost, ix86_division_cost,
3858 ix86_shift_rotate_cost): Break out from ...
3859 (ix86_rtx_costs): ... here.
3860 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
3863 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3865 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
3866 when entry block was promoted unlikely.
3867 (estimate_bb_frequencies): Increase frequency scale.
3868 * profile-count.h (profile_count): Export precision info.
3870 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3872 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
3873 disturbin profile of entry block.
3875 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3877 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
3880 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3882 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
3883 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
3885 2017-11-17 Jeff Law <law@redhat.com>
3887 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
3888 from evrp_dom_walker class. Various methods moved into new class.
3889 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
3890 (evrp_range_analyzer::enter): New method.
3891 (evrp_range_analyzer::leave): New method.
3892 (evrp_dom_walker): Remove delegators no longer needed by this class.
3893 Replace vr_values data member with evrp_range_analyzer
3895 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
3896 method extracted from evrp_dom_walker::before_dom_children.
3897 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
3898 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
3900 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
3901 Add private copy constructor and move assignment operators.
3902 Privatize methods and class data where trivially possible.
3903 (evrp_dom_walker::cleanup): New function, extracted from
3904 execute_early_vrp. Simplify access to class data.
3906 * vr-values.h (get_output_for_vrp): Prototype.
3907 * vr-values.c (get_output_for_vrp): New function extracted from
3908 vrp_visit_assignment_or_call and extract_range_from_stmt.
3909 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
3911 2017-11-17 Luis Machado <luis.machado@linaro.org>
3913 * config/aarch64/aarch64.c
3914 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
3915 (qdf24xx_tunings) <autoprefetcher_model>: Set to
3916 tune_params::AUTOPREFETCHER_WEAK.
3918 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3921 * config/arm/arm.c (arm_valid_target_attribute_rec):
3922 Parse "arch=" and "+<ext>".
3923 (arm_valid_target_attribute_tree): Re-init global options.
3924 (arm_option_override): Make non-static.
3925 (arm_options_perform_arch_sanity_checks): Make errors fatal.
3926 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
3927 (__ARM_FEATURE_CRC32): Support undef.
3928 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
3929 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
3931 2017-11-17 David Malcolm <dmalcolm@redhat.com>
3933 * gdbinit.in (break-on-diagnostic): New command.
3935 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3937 * config/i386/i386.c (ix86_expand_epilogue): Change simple
3938 return to indirect jump for EH return if control-flow
3939 protection is enabled. Change explicit 'false' argument in
3940 pro_epilogue_adjust_stack with a value of
3942 * config/i386/i386.md (simple_return_indirect_internal):
3943 Remove SImode restriction to support 64-bit.
3945 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3947 * combine.c (added_notes_insn): New.
3948 (try_combine): Handle added_notes_insn like added_links_insn.
3949 Rewrite return value code.
3950 (distribute_notes): Set added_notes_insn to the earliest insn we added
3953 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3955 PR rtl-optimization/82621
3956 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
3957 dest of one of those SETs is unused.
3959 2017-11-17 Richard Biener <rguenther@suse.de>
3962 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
3963 * tree-pretty-print.c (dump_generic_node): Handle
3964 annot_expr_parallel_kind.
3965 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
3966 * gimplify.c (gimple_boolify): Likewise.
3968 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3970 * config.gcc (extra_headers): Add cet.h for x86 targets.
3971 * config/i386/cet.h: New file.
3972 * doc/install.texi: Add --enable-cet/--disable-cet.
3974 2017-11-17 Richard Biener <rguenther@suse.de>
3976 PR tree-optimization/83017
3977 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
3978 (gen_parallel_loop): Properly count iterations.
3979 (parallelize_loops): Handle loop->can_be_parallel independent
3980 of flag_loop_parallelize_all. Make static profitability test match
3982 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
3983 * invoke.texi (parloops-min-per-thread): Document.
3985 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
3987 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
3988 upstreaming review comments.
3990 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3992 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
3993 with fast unaligned access.
3994 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
3996 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
3998 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
3999 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
4001 2017-11-17 Richard Biener <rguenther@suse.de>
4003 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
4004 folding of references.
4006 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
4009 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
4010 of replacing call to strncmp with corresponding call to strcmp when
4013 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
4015 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
4016 option prefer-avx256 for skylake-avx512 configuration.
4017 * config/i386/i386.c (ix86_option_override_internal): Ditto.
4018 (get_builtin_code_for_version): Ditto.
4020 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
4021 Monk Chiang <sh.chiang04@gmail.com>
4023 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
4024 (FIXED_REGISTERS): Reserve more register numbers.
4025 (CALL_USED_REGISTERS): Likewise.
4026 (REG_ALLOC_ORDER): Likewise.
4027 (REG_CLASS_CONTENTS): Likewise.
4028 (REGISTER_NAMES): Likewise.
4030 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
4031 Kito Cheng <kito.cheng@gmail.com>
4033 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
4035 * config/nds32/iterators.md: Add vector mode iterators and attributes.
4037 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
4039 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
4040 parameter list for vec_splats.
4042 2017-11-16 Joseph Myers <joseph@codesourcery.com>
4044 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
4046 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
4048 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
4051 * doc/install.texi: Document incompatibility of Perl >=5.6.26
4052 with the required version of automake 1.11.6.
4054 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
4056 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
4058 (DU_C2_3_power9): Correct reservation combinations.
4059 (FP_DIV_power9, VEC_DIV_power9): New.
4060 (power9-alu): Split out rotate/shift...
4061 (power9-rot): ...to here, correct dispatch resource.
4062 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
4064 (power9-fp): Correct latency.
4065 (power9-sdiv): Add div/sqrt resource.
4066 (power9-ddiv): Correct latency, add div/sqrt resource.
4067 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
4068 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
4070 (power9-qpdiv, power9-qpmul): Adjust resource usage.
4072 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
4074 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
4075 switch statement mapping KF built-ins to TF built-ins if we don't
4076 have the proper ISA 3.0 assembler support.
4078 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4080 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
4081 (lower_emutls_function_body): Do not compute it.
4083 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4085 * ipa-split.c (split_bb_info): Turn time to sreal.
4086 (split_point): Likewise.
4087 (dump_split_point): Likewise.
4088 (fine_split_points): Likewise.
4089 (execute_split_functions): Only zero split_bbs; turn time to sreals.
4091 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4093 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
4096 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4098 * predict.c (combine_predictions_for_bb): Preserve zero predicted
4100 (expensive_function_p): Remove useless assert.
4101 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
4103 2017-11-16 Martin Sebor <msebor@redhat.com>
4105 PR tree-optimization/82588
4106 PR tree-optimization/82583
4107 * tree-vrp.c (check_array_ref): Handle flexible array members,
4108 string literals, and inner indices.
4109 (search_for_addr_array): Add detail to diagnostics.
4111 2017-11-16 Nathan Sidwell <nathan@acm.org>
4115 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
4116 (SET_DECL_ASSEMBLER_NAME): Forward to
4117 overwrite_decl_assembler_name.
4118 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
4119 (overwrite_decl_assembler_name): Declare.
4120 * tree.c (overwrite_decl_assembler_name): New.
4121 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
4122 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
4123 (LANG_HOOKS_INITIALIZER): Add it.
4124 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
4125 * langhooks.c (lhd_set_decl_assembler_name): Use
4126 SET_DECL_ASSEMBLER_NAME.
4127 (lhd_overwrite_decl_assembler_name): Default implementation.
4129 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
4130 Jackson Woodruff <jackson.woodruff@arm.com>
4132 PR tree-optimization/71026
4133 * match.pd: Canonicalize constant multiplies in division.
4135 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
4137 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
4138 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
4139 Remove OPT_fomit_frame_pointer entry.
4140 * common/config/arc/arc-common.c: Likewise.
4141 * common/config/arm/arm-common.c: Likewise.
4142 * common/config/avr/avr-common.c: Likewise.
4143 * common/config/c6x/c6x-common.c: Likewise.
4144 * common/config/cr16/cr16-common.c: Likewise.
4145 * common/config/cris/cris-common.c: Likewise.
4146 * common/config/epiphany/epiphany-common.c: Likewise.
4147 * common/config/fr30/fr30-common.c: Likewise.
4148 * common/config/frv/frv-common.c: Likewise.
4149 * common/config/ia64/ia64-common.c: Likewise.
4150 * common/config/iq2000/iq2000-common.c: Likewise.
4151 * common/config/lm32/lm32-common.c: Likewise.
4152 * common/config/m32r/m32r-common.c: Likewise.
4153 * common/config/mcore/mcore-common.c: Likewise.
4154 * common/config/microblaze/microblaze-common.c: Likewise.
4155 * common/config/mips/mips-common.c: Likewise.
4156 * common/config/mmix/mmix-common.c: Likewise.
4157 * common/config/mn10300/mn10300-common.c: Likewise.
4158 * common/config/nios2/nios2-common.c: Likewise.
4159 * common/config/pa/pa-common.c: Likewise.
4160 * common/config/pdp11/pdp11-common.c: Likewise.
4161 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
4162 * common/config/riscv/riscv-common.c: Likewise.
4163 * common/config/rs6000/rs6000-common.c: Likewise.
4164 * common/config/rx/rx-common.c: Likewise.
4165 * common/config/s390/s390-common.c: Likewise.
4166 * common/config/sh/sh-common.c: Likewise.
4167 * common/config/sparc/sparc-common.c: Likewise.
4168 * common/config/tilegx/tilegx-common.c: Likewise.
4169 * common/config/tilepro/tilepro-common.c: Likewise.
4170 * common/config/v850/v850-common.c: Likewise.
4171 * common/config/visium/visium-common.c: Likewise.
4172 * common/config/xstormy16/xstormy16-common.c: Likewise.
4173 * common/config/xtensa/xtensa-common.c: Likewise.
4174 * invoke.texi (-fomit-frame-pointer): Update documentation.
4176 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4178 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
4180 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4182 * cfg.c (scale_bbs_frequencies_int,
4183 cale_bbs_frequencies_gcov_type): Remove.
4184 * cfg.h (scale_bbs_frequencies_int,
4185 cale_bbs_frequencies_gcov_type): Remove.
4187 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4189 * tree-ssa-loop-manip.c
4190 (scale_dominated_blocks_in_loop): Update to profile counts.
4191 (tree_transform_and_unroll_loop): Likewise.
4193 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4195 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
4196 scale_bbs_frequencies_int.
4198 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4200 * final.c (compute_alignments): Use counts rather than frequencies.
4202 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4204 * cfgloopanal.c: Include sreal.h
4205 (average_num_loop_insns): Use counts and sreal for accounting.
4207 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4209 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
4212 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4213 Kito Cheng <kito.cheng@gmail.com>
4215 * config/nds32/constraints.md: Provide more constraints.
4216 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
4217 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
4218 support constraints usage.
4220 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4222 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
4224 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4225 Kito Cheng <kito.cheng@gmail.com>
4227 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
4228 * config/nds32/nds32.opt: Refine the layout.
4229 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
4230 TARGET_EXT_STRING): Support new options.
4231 * config/nds32/nds32.h: Likewise.
4232 * config/nds32/nds32.md: Likewise.
4233 * config/nds32/nds32-predicates.c: Likewise.
4234 * config/nds32/constraints.md: Likewise.
4235 * common/config/nds32/nds32-common.c: Likewise.
4237 2017-11-16 Julia Koval <julia.koval@intel.com>
4240 * config/i386/gfniintrin.h: Add sse check.
4241 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
4243 2017-11-16 Julia Koval <julia.koval@intel.com>
4245 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
4246 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
4247 (ix86_handle_option): Handle -mavx512vbmi2.
4248 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
4249 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4250 * config/i386/i386-c.c (__AVX512VBMI2__): New.
4251 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
4252 (ix86_valid_target_attribute_inner_p): Ditto.
4253 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
4254 * config/i386/i386.opt (mavx512vbmi2): New option.
4255 * doc/invoke.texi: Add new option.
4257 2017-11-16 Julia Koval <julia.koval@intel.com>
4259 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
4260 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
4261 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
4262 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
4263 _mm512_gf2p8mul_epi8): New intrinsics.
4264 * config/i386/i386-builtin-types.def
4265 (V64QI_FTYPE_V64QI_V64QI): New type.
4266 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
4267 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
4268 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
4269 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
4270 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
4271 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4273 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
4275 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
4276 explicitly as a stream of bytes.
4278 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4280 * config/rs6000/altivec.h (vec_xst_be): New #define.
4281 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
4282 and externalize from *altivec_vperm_<mode>_internal.
4283 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
4285 (XL_BE_V8HI): Likewise.
4286 (XL_BE_V4SI): Likewise.
4287 (XL_BE_V4SI): Likewise.
4288 (XL_BE_V2DI): Likewise.
4289 (XL_BE_V4SF): Likewise.
4290 (XL_BE_V2DF): Likewise.
4291 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
4292 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
4293 all array entries with these keys: VSX_BUILTIN_VEC_XL,
4294 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
4295 VSX_BUILTIN_VEC_XST_BE.
4296 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
4297 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
4299 (altivec_init_builtins): Replace conditional calls to def_builtin
4300 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
4301 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
4302 with unconditional calls. Remove calls to def_builtin for
4303 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
4304 __builtin_vec_xst_be.
4305 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
4306 to define_expand, and add alternate RTL generation for P8.
4307 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
4308 vsx_ld_elemrev_v8hi.
4309 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
4310 add alternate RTL generation for P8.
4311 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
4312 vsx_ld_elemrev_v16qi.
4313 (vsx_st_elemrev_v8hi): Convert define_insn
4314 to define_expand, and add alternate RTL generation for P8.
4315 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
4316 vsx_st_elemrev_v8hi.
4317 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
4318 add alternate RTL generation for P8.
4319 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
4320 vsx_st_elemrev_v16qi.
4322 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
4325 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
4326 TARGET_AVX512ER check.
4327 (ix86_option_override_internal): Set MASK_VZEROUPPER if
4328 neither -mzeroupper nor -mno-zeroupper is used and
4329 TARGET_EMIT_VZEROUPPER is set.
4330 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
4331 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
4333 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
4335 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4336 folding of vector compares.
4337 (fold_build_vec_cmp): New helper function.
4338 (fold_compare_helper): New helper function.
4339 (builtin_function_type): Add compare builtins to the list of functions
4340 having unsigned arguments. Cosmetic updates to comment indentation.
4341 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
4342 the not+eq combination.
4344 2017-11-15 Bin Cheng <bin.cheng@arm.com>
4346 PR tree-optimization/82726
4347 PR tree-optimization/70754
4348 * tree-predcom.c (order_drefs_by_pos): New function.
4349 (combine_chains): Move code setting has_max_use_after to...
4350 (try_combine_chains): ...here. New parameter. Sort combined chains
4351 according to position information.
4352 (tree_predictive_commoning_loop): Update call to above function.
4353 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
4355 2017-11-15 Bin Cheng <bin.cheng@arm.com>
4357 PR tree-optimization/82726
4359 2017-01-23 Bin Cheng <bin.cheng@arm.com>
4361 PR tree-optimization/70754
4362 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
4363 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
4364 combined stmt before it if not NULL.
4365 (combine_chains): Process refs reversely and compute dominance point
4369 2017-02-23 Bin Cheng <bin.cheng@arm.com>
4371 PR tree-optimization/79663
4372 * tree-predcom.c (combine_chains): Process refs in reverse order
4373 only for ZERO length chains, and add explaining comment.
4375 2017-11-15 Tamar Christina <tamar.christina@arm.com>
4377 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
4378 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
4379 * doc/invoke.texi (ARM Options): Add armv8.3-a.
4381 2017-11-15 Tamar Christina <tamar.christina@arm.com>
4383 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
4385 2017-11-15 Martin Liska <mliska@suse.cz>
4387 * tree-cfg.c (pass_warn_function_return::execute):
4388 Compare warn_return_type for greater than zero.
4390 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
4394 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
4395 to return true on Xeon and not on Xeon Phi.
4396 (ix86_check_avx256_register): Changed to ...
4397 (ix86_check_avx_upper_register): ... this. Add extra check for
4398 VALID_AVX512F_REG_OR_XI_MODE.
4399 (ix86_avx_u128_mode_needed): Changed
4400 ix86_check_avx256_register to ix86_check_avx_upper_register.
4401 (ix86_check_avx256_stores): Changed to ...
4402 (ix86_check_avx_upper_stores): ... this. Changed
4403 ix86_check_avx256_register to ix86_check_avx_upper_register.
4404 (ix86_avx_u128_mode_after): Changed
4405 avx_reg256_found to avx_upper_reg_found. Changed
4406 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
4407 (ix86_avx_u128_mode_entry): Changed
4408 ix86_check_avx256_register to ix86_check_avx_upper_register.
4409 (ix86_avx_u128_mode_exit): Ditto.
4410 * config/i386/i386.h: (host_detect_local_cpu): New define.
4412 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
4414 * config/arm/xgene1.md (xgene1): Split into automatons
4415 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
4416 (xgene1_f_load): Adjust reservations and/or types.
4417 (xgene1_f_store): Likewise.
4418 (xgene1_load_pair): Likewise.
4419 (xgene1_store_pair): Likewise.
4420 (xgene1_fp_load1): Likewise.
4421 (xgene1_load1): Likewise.
4422 (xgene1_store1): Likewise.
4423 (xgene1_move): Likewise.
4424 (xgene1_alu): Likewise.
4425 (xgene1_simd): Likewise.
4426 (xgene1_bfm): Likewise.
4427 (xgene1_neon_load1): Likewise.
4428 (xgene1_neon_store1): Likewise.
4429 (xgene1_neon_logic): Likewise.
4430 (xgene1_neon_st1): Likewise.
4431 (xgene1_neon_ld1r): Likewise.
4432 (xgene1_alu_cond): Added.
4433 (xgene1_shift_reg): Likwise.
4434 (xgene1_bfx): Likewise.
4435 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
4437 2017-11-15 Jakub Jelinek <jakub@redhat.com>
4440 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
4442 (can_widen_mult_without_libcall): New function.
4443 (expand_mul_overflow): If only checking unsigned mul overflow,
4444 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
4445 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
4446 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
4447 (expand_DIVMOD): Formatting fix.
4448 * expmed.h (expand_mult): Add NO_LIBCALL argument.
4449 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
4450 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
4452 PR tree-optimization/82977
4453 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
4454 constructed temporary to strlen_to_stridx.put.
4456 2017-11-15 Martin Liska <mliska@suse.cz>
4458 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
4459 * configure: Regenerate.
4461 2017-11-15 Martin Liska <mliska@suse.cz>
4464 * config/sh/sh-mem.cc: Use proper probability for
4467 2017-11-14 Jeff Law <law@redhat.com>
4469 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
4470 the red zone for stack_clash_protection_final_dynamic_probe targets
4471 when the total dynamic stack size is zero bytes.
4473 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
4474 blocks is post order.
4476 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
4478 * dumpfile.h (TDF_COMPARE_DEBUG): New.
4479 * final.c (rest_of_clean_state): Set it for the
4480 -fcompare-debug dump.
4481 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
4482 class when TDF_COMPARE_DEBUG is set.
4484 * dwarf2out.c (gen_producer_string): Discard
4487 2017-11-15 Joseph Myers <joseph@codesourcery.com>
4490 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
4491 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
4492 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
4493 (__TGMATH_REAL_2_3): Remove macros.
4494 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
4495 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
4497 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
4499 (remquo): Define using __TGMATH_REAL_3.
4501 2017-11-14 Jeff Law <law@redhat.com>
4503 * vr-values.c: New file with contents extracted from tree-vrp.c.
4504 * Makefile.in (OBJS): Add vr-values.o
4505 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
4506 (set_value_range, set_and_canonicalize_value_range): Likewise.
4507 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
4508 (value_range_constant_singleton, symbolic_range_p): Likewise.
4509 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
4510 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
4511 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
4512 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
4513 (find_case_label_range, find_case_label_index): Likewise.
4514 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
4515 (range_int_cst_singleton_p, value_inside_range): Likewise.
4516 (get_single_symbol): Likewise.
4517 (switch_update): Move structure definition here.
4518 (to_remove_edges, to_update_switch_stmts): Provide externs.
4519 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
4520 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
4521 (vrp_val_is_min, set_value_range): Likewise.
4522 (set_and_canonicalize_value_range, copy_value_range): Likewise.
4523 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
4524 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
4525 (range_is_nonnull, range_int_cst_p): Likewwise.
4526 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
4527 (get_single_symbol, operand_less_p): Likewise
4528 (compare_values_warnv, compare_values): Likewise.
4529 (value_inside_range, value_range_constant_singleton): Likewise.
4530 (zero_nonzero_bitgs_from_vr): Likewise.
4531 (extract_range_from_binary_expr_1): Likewise.
4532 (overflow_comparison_p): Likewise.
4533 (to_remove_edges, to_update_switch_stmts): Likewise.
4534 (find_case_label-index, find_case_label_range): Likewise.
4535 (switch_update, set_value_range_to_nonnegative): Remove.
4536 (set_value_range_to_truthvalue): Likewise.
4537 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
4538 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
4539 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
4540 (find_case_label_ranges, test_for_singularity): Likewise.
4541 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
4542 (x_vr_values): Move to its remaining use site.
4544 2017-11-10 Jeff Law <law@redhat.com>
4546 * vr-values.h (VR_INITIALIZER): Move #define here.
4547 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
4548 * Makefile.in (OBJS): Add tree-evrp.o
4549 * tree-vrp.h (assert_info): Move structure definition here.
4550 (set_value_range_to_varying): Prototype.
4551 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
4552 (infer_value_range, register_edge_assert_for): Likewise.
4553 (stmt_interesting_for_vrp): Likewise.
4554 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
4555 (set_value_range_to_varying): No longer static.
4556 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
4557 (infer_value_range, register_edge_assert_for): Likewise.
4559 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
4561 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
4562 generate the XXBRD instruction.
4564 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
4565 (rs6000_builtin_type_compatible): Treat _Float128 and long double
4566 as being compatible if -mabi=ieeelongdouble.
4567 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
4568 to setup float128 built-ins with hardware support.
4569 (BU_FLOAT128_HW_2): Likewise.
4570 (BU_FLOAT128_HW_3): Likewise.
4571 (BU_FLOAT128_HW_VSX_1): Likewise.
4572 (BU_FLOAT128_HW_VSX_2): Likewise.
4573 (scalar_extract_expq): Change float128 built-in functions to
4574 accommodate having both KFmode and TFmode functions. Use the
4575 KFmode variant as the default.
4576 (scalar_extract_sigq): Likewise.
4577 (scalar_test_neg_qp): Likewise.
4578 (scalar_insert_exp_q): Likewise.
4579 (scalar_insert_exp_qp): Likewise.
4580 (scalar_test_data_class_qp): Likewise.
4581 (sqrtf128_round_to_odd): Delete processing the round to odd
4582 built-in functions as special built-in functions, and define them
4583 as float128 built-ins. Use the KFmode variant as the default.
4584 (truncf128_round_to_odd): Likewise.
4585 (addf128_round_to_odd): Likewise.
4586 (subf128_round_to_odd): Likewise.
4587 (mulf128_round_to_odd): Likewise.
4588 (divf128_round_to_odd): Likewise.
4589 (fmaf128_round_to_odd): Likewise.
4590 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
4591 support for KFmode and TFmode xststdcqp calls.
4592 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
4593 point, switch the built-in handlers for the get/set float128
4594 exponent, get float128 mantissa, float128 test built-ins, and the
4595 float128 round to odd built-in functions. Eliminate creating the
4596 float128 round to odd built-in functions as special built-ins.
4597 (rs6000_init_builtins): Eliminate special creation of the float128
4598 round to odd built-in functions.
4599 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
4600 function insns to support both TFmode and KFmode variants.
4601 (xsxsigqp_<mode>): Likewise.
4602 (xsiexpqpf_<mode>): Likewise.
4603 (xsiexpqp_<mode>): Likewise.
4604 (xststdcqp_<mode>): Likewise.
4605 (xststdcnegqp_<mode>): Likewise.
4606 (xststdcqp_<mode>): Likewise.
4608 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4610 * tree-ssa-threadupdate.c (compute_path_counts): Remove
4611 unused path_in_freq_ptr parameter.
4612 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
4614 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4616 * ipa-inline.c (edge_badness): Dump sreal frequency.
4617 (compute_inlined_call_time): Match natural implementaiton ...
4618 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
4619 forgotten division by CGRAPH_FREQ_BASE.
4621 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4623 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
4624 Solaris 11. Update comment.
4625 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
4627 * config/sol2.h (STARTFILE_SPEC): Likewise.
4628 * configure: Regenerate.
4630 2017-11-14 Carl Love <cel@us.ibm.com>
4632 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
4633 le_ and be_ prefixes to swap* variables. Remove
4634 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
4637 2017-11-14 Jason Merrill <jason@redhat.com>
4639 Support GTY((cache)) on hash_map.
4640 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
4641 (ggc_cache_remove): Override it instead of ggc_mx.
4642 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
4643 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
4644 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
4645 (simple_cache_map_traits): Override maybe_mx.
4646 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
4647 (hash_map): Friend gt_cleare_cache.
4648 (gt_cleare_cache): New.
4649 * tree.h (tree_cache_traits): New hash_map traits class.
4650 (tree_cache_map): New typedef.
4652 2017-11-14 Richard Biener <rguenther@suse.de>
4654 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
4655 paramter and handling.
4656 (cleanup_control_flow_bb): Likewise.
4657 (cleanup_control_flow_pre): New helper performing a DFS walk
4658 to call cleanup_control_flow_bb in PRE order.
4659 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
4660 via cleanup_control_flow_pre.
4662 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
4664 * config/aarch64/aarch64-simd.md
4665 (aarch64_simd_bsl<mode>_internal): Remove DImode.
4666 (*aarch64_simd_bsl<mode>_alt): Likewise.
4667 (aarch64_simd_bsldi_internal): New.
4668 (aarch64_simd_bsldi_alt): Likewise.
4670 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4672 * tracer.c (better_p): Do not compare frequencies.
4673 * reg-stack.c (better_edge): Likewise.
4674 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
4677 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4679 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
4680 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
4681 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
4682 * lto-streamer-in.c (input_function): Use update_max_bb_count.
4683 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
4684 * predict.c (maybe_hot_frequency_p): Inline to ...
4685 (maybe_hot_count_p): ... here; rewrite to counts.
4686 (counts_to_freqs): Rename to ...
4687 (update_max_bb_count): ... this one.
4688 (expensive_function_p): Use counts.
4689 (estimate_bb_frequencies): Update.
4690 (rebuild_frequencies): Update.
4691 * predict.h (counts_to_freqs): Rename to ...
4692 (update_max_bb_count): ... this one.
4693 * profile.c (compute_branch_probabilities): Add debug info
4694 * tree-inline.c (expand_call_inline): Update debug info.
4695 (optimize_inline_calls): Use update_max_bb_count..
4696 (tree_function_versioning): Use update_max_bb_count..
4697 * value-prof.c (gimple_value_profile_transformations):
4698 Do not use update_max_bb_count.
4700 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4702 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
4703 always use frequencies.
4705 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4707 * bb-reorder.c: Remove frequencies from comments.
4708 (better_edge_p): Use profile counts.
4709 (find_traces): Dump profile counts.
4710 (rotate_loop): Use profile counts.
4711 (find_traces_1_round): Likewise.
4712 (connect_better_edge_p): Use counts instead of probabilities for
4714 (copy_bb_p): Drop early check for non-0 frequency.
4715 (sanitize_hot_paths): Update comments.
4717 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4719 * ipa-split.c (struct split_point): Add count.
4720 (consider_split): Do not compute incoming frequency; compute incoming
4721 count and store it to split_point.
4722 (split_function): Set count of the call to split part correctly.
4724 2017-11-13 Carl Love <cel@us.ibm.com>
4726 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
4728 2017-11-13 Tom Tromey <tom@tromey.com>
4730 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
4732 2017-11-13 Carl Love <cel@us.ibm.com>
4734 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
4735 Add support for builtins:
4736 unsigned int vec_first_{,miss}_match_{,or_eos}index,
4737 vector {un,}signed {char,int,short},
4738 vector {un,}signed {char,int,short}) arguments.
4739 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
4740 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
4741 Add BU_P9V_AV_2 expansions for the builtins.
4742 * config/rs6000/altivec.h (vec_first_match_index,
4743 vec_first_mismatch_index, vec_first_match_or_eos_index,
4744 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
4745 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
4746 new extern declaration.
4747 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
4748 * config/rs6000/vsx.md (first_match_index_<mode>,
4749 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
4750 first_mismatch_or_eos_index_<mode>): Add define expand.
4751 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
4752 * doc/extend.texi: Update the built-in documenation file for the new
4755 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
4757 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
4758 into the min/max operations for _Float<N> and _Float<N>X types.
4760 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4763 * dwarf2out.c (do_eh_frame): New static variable.
4764 (dwarf2out_begin_prologue): Set it.
4765 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
4767 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4769 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
4771 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
4773 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4776 * debug.h (dwarf2out_do_eh_frame): Declare.
4777 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
4778 (dwarf2out_do_frame): Use it.
4779 (dwarf2out_do_cfi_asm): Likewise.
4780 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
4781 (dwarf2out_assembly_start): Likewise.
4782 (dwarf2out_begin_prologue): Fix comment.
4783 * toplev.c (compile_file): Always call dwarf2out_frame_finish
4784 if the target needs either debug or unwind DWARF2 info.
4785 * lto-opts.c (lto_write_options): Do not save -fexceptions,
4786 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
4787 -ftrapping-math, -ftrapv and -fwrapv.
4789 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4791 * cgraph.c (cgraph_edge::sreal_frequency): New function.
4792 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
4793 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
4794 (estimate_edge_size_and_time): Likewise.
4795 (ipa_merge_fn_summary_after_inlining): Likewise.
4796 * ipa-inline.c (cgraph_freq_base_rec): Remove.
4797 (compute_uninlined_call_time): Use sreal_frequency.
4798 (compute_inlined_call_time): Likewise.
4799 (ipa_inline): Do not initialize cgraph_freq_base_rec.
4800 * profile-count.c: Include sreal.h.
4801 (profile_count::to_sreal_scale): New.
4802 * profile-count.h: Forward declare sreal.
4803 (profile_count::to_sreal_scale): Declare.
4805 2017-11-13 Nathan Sidwell <nathan@acm.org>
4807 * diagnostic.c (maybe_line_and_column): New.
4808 (diagnostic_get_location_text): Use it.
4809 (diagnostic_report_current_module): Likewise.
4810 (test_diagnostic_get_location_text): Add tests.
4812 2017-11-13 Luis Machado <luis.machado@linaro.org>
4814 * doc/md.texi (Specifying processor pipeline description): Fix
4815 incorrect latency for the div instruction example.
4817 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4819 PR tree-optimization/78821
4820 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
4821 that bit_not_p is the same.
4822 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
4823 (split_group): Count precisely bit_not_p bits in each statement.
4824 (invert_op): New function.
4825 (imm_store_chain_info::output_merged_store): Use invert_op to
4826 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
4827 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
4829 2017-11-13 Martin Liska <mliska@suse.cz>
4831 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
4832 (struct source_info): Likewise.
4833 (add_branch_counts): Likewise.
4834 (add_line_counts): Likewise.
4835 (function_summary): Likewise.
4836 (output_intermediate_line): Likewise.
4837 (generate_results): Likewise.
4839 2017-11-13 Martin Liska <mliska@suse.cz>
4841 * gcov.c (struct block_info): Remove typedef for block_t.
4842 (struct line_info): Likewise.
4843 (line_info::has_block): Likewise.
4844 (EXIT_BLOCK): Likewise.
4845 (unblock): Likewise.
4846 (circuit): Likewise.
4847 (get_cycles_count): Likewise.
4848 (process_file): Likewise.
4849 (read_graph_file): Likewise.
4850 (solve_flow_graph): Likewise.
4851 (find_exception_blocks): Likewise.
4852 (add_line_counts): Likewise.
4853 (accumulate_line_info): Likewise.
4854 (output_line_details): Likewise.
4856 2017-11-13 Martin Liska <mliska@suse.cz>
4858 * gcov.c (struct arc_info): Remove typedef for arc_t.
4859 (struct line_info): Likewise.
4860 (add_branch_counts): Likewise.
4861 (output_branch_count): Likewise.
4862 (function_info::~function_info): Likewise.
4863 (circuit): Likewise.
4864 (output_intermediate_line): Likewise.
4865 (read_graph_file): Likewise.
4866 (solve_flow_graph): Likewise.
4867 (find_exception_blocks): Likewise.
4868 (add_line_counts): Likewise.
4869 (accumulate_line_info): Likewise.
4870 (output_line_details): Likewise.
4871 (output_function_details): Likewise.
4873 2017-11-13 Martin Liska <mliska@suse.cz>
4875 * gcov.c (struct function_info): Remove typedef for function_t.
4876 (struct source_info): Likewise.
4877 (source_info::get_functions_at_location): Likewise.
4878 (solve_flow_graph): Likewise.
4879 (find_exception_blocks): Likewise.
4880 (add_line_counts): Likewise.
4881 (output_intermediate_file): Likewise.
4882 (process_file): Likewise.
4883 (generate_results): Likewise.
4884 (release_structures): Likewise.
4885 (read_graph_file): Likewise.
4886 (read_count_file): Likewise.
4887 (accumulate_line_counts): Likewise.
4888 (output_lines): Likewise.
4890 2017-11-13 Martin Liska <mliska@suse.cz>
4892 * gcov.c (function_info::function_info): Remove num_counts
4893 and add vector<gcov_type>.
4894 (function_info::~function_info): Use the vector.
4895 (process_file): Likewise.
4896 (read_graph_file): Likewise.
4897 (read_count_file): Likewise.
4898 (solve_flow_graph): Likewise.
4900 2017-11-13 Martin Liska <mliska@suse.cz>
4902 * gcov.c (function_info::is_artificial): New function.
4903 (process_file): Erase all artificial early.
4904 (generate_results): Skip as all artificial are already
4907 2017-11-13 Martin Liska <mliska@suse.cz>
4909 * gcov.c (read_graph_file): Store to global vector of functions.
4910 (read_count_file): Iterate the vector.
4911 (process_file): Likewise.
4912 (generate_results): Likewise.
4913 (release_structures): Likewise.
4915 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4917 PR tree-optimization/82954
4918 * gimple-ssa-store-merging.c
4919 (imm_store_chain_info::coalesce_immediate_stores): If
4920 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
4922 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
4924 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
4925 Upddate call to ENDIAN_LANE_N.
4926 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
4927 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
4928 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
4929 and use aarch64_endian_lane_rtx.
4930 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
4932 2017-11-12 Tom de Vries <tom@codesourcery.com>
4934 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
4936 2017-11-12 Tom de Vries <tom@codesourcery.com>
4938 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
4940 2017-11-12 Tom de Vries <tom@codesourcery.com>
4942 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
4944 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4945 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4946 * defaults.h (ASM_OUTPUT_LABELREF): Same.
4948 2017-11-11 Martin Sebor <msebor@redhat.com>
4951 * doc/extend.texi (attribute nonstring): Remove spurious argument.
4954 * prefic.c (translate_name): Replace strncpy with memcpy to
4955 avoid -Wstringop-truncation.
4957 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4959 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
4961 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4963 * predict.c (maybe_hot_frequency_p): Do not use cfun.
4965 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4967 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
4970 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
4972 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
4973 enable generating XXBRH if the value is in a vector register.
4974 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
4975 value is in a vector register.
4976 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
4977 register to register bswap64's instead of doing the GPR sequence
4978 used on previous machines.
4979 (bswapdi2_xxbrd): New insn.
4980 (bswapdi2_reg): Disallow on ISA 3.0.
4981 (register to register bswap64 splitter): Do not split the insn on
4982 ISA 3.0 systems that use XXBRD.
4984 2017-11-10 Martin Sebor <msebor@redhat.com>
4987 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
4989 (find_subframework_file): Same.
4991 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4993 * auto-profile.c (afdo_indirect_call): Drop frequency.
4994 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
4995 (cgraph_node::create_edge): Drop frequency argument.
4996 (cgraph_node::create_indirect_edge): Drop frequency argument.
4997 (cgraph_edge::make_speculative): Drop frequency arguments.
4998 (cgraph_edge::resolve_speculation): Do not update frequencies
4999 (cgraph_edge::dump_edge_flags): Do not dump frequency.
5000 (cgraph_node::dump): Check consistency in IPA mode.
5001 (cgraph_edge::maybe_hot_p): Use IPA counter.
5002 (cgraph_edge::verify_count_and_frequency): Rename to ...
5003 (cgraph_edge::verify_count): ... this one; drop frequency checking.
5004 (cgraph_node::verify_node): Update.
5005 * cgraph.h (struct cgraph_edge): Drop frequency.
5006 (cgraph_edge::frequency): New function.
5007 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
5009 (cgraph_edge::rebuild_edges): Likewise.
5010 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
5011 (duplicate_thunk_for_node): Do not pass frequency.
5012 (cgraph_node::create_clone): Scale only counts.
5013 (cgraph_node::create_virtual_clone): Do not pass frequency.
5014 (cgraph_node::create_edge_including_clones): Do not pass frequency.
5015 (cgraph_node::create_version_clone): Do not pass frequency.
5016 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
5017 (cgraph_node::expand_thunk): Do not pass frequency.
5018 (cgraph_node::create_wrapper): Do not pass frequency.
5019 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
5021 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
5022 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
5023 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
5024 (gather_caller_stats): Use frequency function.
5025 (ipcp_cloning_candidate_p): Use frequency function.
5026 (ipcp_propagate_stage): Use frequency function.
5027 (get_info_about_necessary_edges): Use frequency function.
5028 (update_profiling_info): Update only IPA profile.
5029 (update_specialized_profile): Use frequency functoin.
5030 (perhaps_add_new_callers): Update only IPA profile.
5031 * ipa-devirt.c (ipa_devirt): Use IPA profile.
5032 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
5033 (dump_ipa_call_summary): Use frequency function.
5034 (estimate_edge_size_and_time): Use frequency function.
5035 (ipa_merge_fn_summary_after_inlining): Use frequency function.
5036 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
5037 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
5038 (update_noncloned_counts): ... ths one; scale counts only.
5039 (clone_inlined_nodes): Do not scale frequency.
5040 (inline_call): Do not pass frequency.
5041 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
5042 (compute_inlined_call_time): Use IPA profile.
5043 (want_inline_small_function_p): Use IPA profile.
5044 (want_inline_self_recursive_call_p): Use IPA profile.
5045 (edge_badness): Use IPA profile.
5046 (lookup_recursive_calls): Use IPA profile.
5047 (recursive_inlining): Do not pass frequency.
5048 (resolve_noninline_speculation): Do not update frequency.
5049 (inline_small_functions): Collect max of IPA profile.
5050 (dump_overall_stats): Dump IPA porfile.
5051 (dump_inline_stats): Dump IPA porfile.
5052 (ipa_inline): Collect IPA stats.
5053 * ipa-inline.h (clone_inlined_nodes): Update prototype.
5054 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
5055 (ipa_propagate_frequency): Use frequency function.
5056 (ipa_profile): Cleanup.
5057 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
5058 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
5059 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
5060 (input_node): Do not stream frequency.
5061 (input_edge): Do not stream frequency.
5062 (merge_profile_summaries): Scale only IPA profiles.
5063 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
5064 * predict.c (drop_profile): Do not recompute frequency.
5065 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
5066 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
5067 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
5068 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
5069 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
5070 * tree-inline.c (copy_bb): Do not scale frequency.
5071 (expand_call_inline): Do not scale frequency.
5072 (tree_function_versioning): Do not scale frequency.
5073 * ubsan.c (ubsan_create_edge): Do not pass frequency.
5075 2017-11-10 Julia Koval <julia.koval@intel.com>
5077 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
5078 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
5079 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
5080 (_mm256_maskz_gf2p8affine_epi64_epi8)
5081 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
5082 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
5083 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
5084 (__builtin_ia32_vgf2p8affineqb_v32qi)
5085 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
5086 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
5088 2017-11-10 Tamar Christina <tamar.christina@arm.com>
5092 (arm_option_override): Refactor.
5093 (arm_option_reconfigure_globals): New.
5094 (arm_options_perform_arch_sanity_checks): New.
5095 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
5097 (arm_options_perform_arch_sanity_checks): Likewise
5099 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
5101 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
5102 (power9-qpmul): New.
5103 * rs6000/rs6000.md ("type" attr): Add qmul.
5104 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
5105 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
5106 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
5108 2017-11-10 Martin Sebor <msebor@redhat.com>
5111 * builtins.c (compute_objsize): Handle arrays that
5112 compute_builtin_object_size likes to fail for. Make extern.
5113 * builtins.h (compute_objsize): Declare.
5114 (check_strncpy_sizes): New function.
5115 (expand_builtin_strncpy): Call check_strncpy_sizes.
5116 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
5117 -Wstringop-truncation.
5118 (gimple_fold_builtin_strncat): Same.
5119 * gimple.c (gimple_build_call_from_tree): Set call location.
5120 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
5121 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
5122 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
5123 (handle_builtin_strlen): Use strlen_to_stridx.
5124 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
5126 Use strlen_to_stridx.
5127 (pass_strlen::execute): Release strlen_to_stridx.
5128 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
5129 (-Wstringop-truncation): Document new option.
5131 2017-11-10 Martin Liska <mliska@suse.cz>
5133 PR gcov-profile/82702
5134 * gcov.c (main): Handle intermediate files in a different
5136 (get_gcov_intermediate_filename): New function.
5137 (output_gcov_file): Remove support of intermediate files.
5138 (generate_results): Allocate intermediate file.
5139 (release_structures): Clean-up properly fn_end.
5140 (output_intermediate_file): Start iterating with line 1.
5142 2017-11-10 Jakub Jelinek <jakub@redhat.com>
5144 PR tree-optimization/82929
5145 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
5146 ops_swapped_p non-static data member.
5147 (store_immediate_info::store_immediate_info): Clear it.
5148 (imm_store_chain_info::coalesce_immediate_stores): If swapping
5149 ops set ops_swapped_p.
5150 (count_multiple_uses): Handle ops_swapped_p.
5152 2017-11-10 Martin Liska <mliska@suse.cz>
5154 * coverage.c (coverage_init): Stream information about
5155 support of has_unexecuted_blocks.
5156 * doc/gcov.texi: Document that.
5157 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
5158 * gcov.c (read_graph_file): Likewise.
5159 (output_line_beginning): Fix a small issue with
5162 2017-11-10 Bin Cheng <bin.cheng@arm.com>
5164 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
5165 reference of trivial component.
5167 2017-11-10 Jakub Jelinek <jakub@redhat.com>
5170 * gimple-ssa-store-merging.c
5171 (pass_store_merging::terminate_all_aliasing_chains): For
5172 gimple_store_p stmts also call refs_output_dependent_p.
5174 PR rtl-optimization/82913
5175 * compare-elim.c (try_merge_compare): Punt if def_insn is not
5178 2017-11-09 Jeff Law <law@redhat.com>
5180 * vr-values.h: New file with vr_values class.
5181 * tree-vrp.c: Include vr-values.h
5182 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
5183 data objects into the vr_values class.
5184 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
5185 (get_value_range): Make it a member function within vr_values class.
5186 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
5187 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
5188 (op_with_constant_singleton_value_range): Likewise.
5189 (extract_range_for_var_from_comparison_expr): Likewise.
5190 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
5191 (extract_range_from_binary_expr): Likewise.
5192 (extract_range_from_unary_expr): Likewise.
5193 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
5194 (check_for_binary_op_overflow, extract_range_basic): Likewise.
5195 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
5196 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
5197 (compare_name_with_value, compare_names): Likewise.
5198 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
5199 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
5200 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
5201 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
5202 (extract_range_from_phi_node): Likewise.
5203 (simplify_truth_ops_using_ranges): Likewise.
5204 (simplify_div_or_mod_using_ranges): Likewise.
5205 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
5206 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
5207 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
5208 (simplify_float_conversion_using_ranges): Likewise.
5209 (simplify_internal_call_using_ranges): Likewise.
5210 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
5211 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
5212 poitner via x_vr_values for calls into gimple folder.
5213 (vrp_initialize_lattice): Make this the vr_values ctor.
5214 (vrp_free_lattice): Make this the vr_values dtor.
5215 (set_vr_value): New function.
5216 (class vrp_prop): Add vr_values data member. Add various member
5217 functions as well as member functions that delegate to vr_values.
5218 (check_array_ref): Make a member function within vrp_prop class.
5219 (search_for_addr_array, vrp_initialize): Likewise.
5220 (vrp_finalize): Likewise. Revamp to avoid direct access to
5221 vr_value, values_propagated, etc.
5222 (check_array_bounds): Extract vrp_prop class instance pointer from
5223 walk info structure. Use it to call member functions.
5224 (check_all_array_refs): Make a member function within vrp_prop class.
5225 Smuggle class instance pointer via walk info structure.
5226 (x_vr_values): New local static.
5227 (vrp_valueize): Use x_vr_values to get class instance.
5228 (vr_valueize_1): Likewise.
5229 (class vrp_folder): Add vr_values data member. Add various member
5230 functions as well as member functions that delegate to vr_values.
5231 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
5232 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
5233 class instance from vr_values. Use it to call member functions.
5234 (vrp_dom_walker): Add vr_values data member.
5235 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
5236 instance via x_vr_values.
5237 (identify_jump_threads): Accept vr_values as argument. Store
5238 it into the walker structure.
5239 (evrp_dom_walker): Add vr_values class data member. Add various
5241 (evrp_dom_walker::try_find_new_range): Use vr_values data
5242 member to access the memory allocator.
5243 (evrp_dom_walker::before_dom_children): Store vr_values class
5244 instance into the vrp_folder class.
5245 (evrp_dom_walker::push_value_range): Rework to avoid direct
5246 access to num_vr_values and vr_value.
5247 (evrp_dom_walker::pop_value_range): Likewise.
5248 (execute_early_vrp): Remove call to vrp_initialize_lattice.
5249 Use vr_values to get to dump_all_value_ranges member function.
5250 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
5251 and simplify_cond_using_ranges_2 via vrp_prop class instance.
5252 Pass vr_values class instance down to identify_jump_threads.
5253 Remove call to vrp_free_lattice.
5254 (debug_all_value_ranges): Remove.
5256 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
5257 (vrp_folder): Likewise.
5259 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
5260 Get it from the existing bitmap instead.
5261 (vrp_intersect_ranges_1): Likewise.
5263 2017-11-09 Jakub Jelinek <jakub@redhat.com>
5265 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
5267 (store_immediate_info::store_immediate_info): Add bitnotp argument,
5268 set bit_not_p to it.
5269 (imm_store_chain_info::coalesce_immediate_stores): Break group
5270 if bit_not_p is different.
5271 (count_multiple_uses, split_group,
5272 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
5273 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
5274 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
5275 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
5276 to store_immediate_info ctor.
5278 2017-11-09 Jim Wilson <jimw@sifive.com>
5280 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
5281 (scan_prog_file): Likewise.
5283 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
5285 * bb-reorder.c (max_entry_frequency): Remove.
5286 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
5287 connect_traces, push_to_next_round_p): Remove prototypes.
5288 (find_traces_1_round): Use counts only.
5289 (push_to_next_round_p): Likewise.
5290 (find_traces): Likewise.
5291 (rotate_loop): Likewise.
5292 (find_traces_1_round): Likewise.
5293 (connect_traces): Likewise.
5294 (edge_order): Likewise.
5296 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
5298 * config/arm/arm.c (output_return_instruction): Add comments to
5299 indicate requirement for cmse_nonsecure_entry return to account
5300 for the size of clearing instruction output here.
5301 (thumb_exit): Likewise.
5302 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
5303 return in hardfloat mode.
5305 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
5307 * config/rs6000/rs6000.c (machine_function): Add a bool,
5308 "toc_is_wrapped_separately".
5309 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
5310 if it wasn't explicitly set or unset, we are optimizing for speed, and
5311 doing separate shrink-wrapping.
5312 (rs6000_get_separate_components): Enable the TOC component if
5313 saving the TOC register in the prologue.
5314 (rs6000_components_for_bb): Handle the TOC component.
5315 (rs6000_emit_prologue_components): Store the TOC register where needed.
5316 (rs6000_set_handled_components): Mark TOC as handled, if handled.
5317 (rs6000_emit_prologue): Don't save the TOC if that is already done.
5319 2017-11-09 Martin Jambor <mjambor@suse.cz>
5321 * ipa-param-manipulation.c: New file.
5322 * ipa-param-manipulation.h: Likewise.
5323 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
5324 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
5325 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
5326 (ipa_parm_adjustment): Likewise.
5327 (ipa_parm_adjustment_vec): Likewise.
5328 (ipa_get_vector_of_formal_parms): Moved declaration to
5329 ipa-param-manipulation.h.
5330 (ipa_get_vector_of_formal_parm_types): Likewise.
5331 (ipa_modify_formal_parameters): Likewise.
5332 (ipa_modify_call_arguments): Likewise.
5333 (ipa_combine_adjustments): Likewise.
5334 (ipa_dump_param_adjustments): Likewise.
5335 (ipa_modify_expr): Likewise.
5336 (ipa_get_adjustment_candidate): Likewise.
5337 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
5338 ipa-param-manipulation.c.
5339 (ipa_get_vector_of_formal_parm_types): Likewise.
5340 (ipa_modify_formal_parameters): Likewise.
5341 (ipa_modify_call_arguments): Likewise.
5342 (ipa_modify_expr): Likewise.
5343 (get_ssa_base_param): Likewise.
5344 (ipa_get_adjustment_candidate): Likewise.
5345 (index_in_adjustments_multiple_times_p): Likewise.
5346 (ipa_combine_adjustments): Likewise.
5347 (ipa_dump_param_adjustments): Likewise.
5348 * tree-sra.c: Also include ipa-param-manipulation.h
5349 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
5352 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5353 Alan Hayward <alan.hayward@arm.com>
5354 David Sherwood <david.sherwood@arm.com>
5356 * doc/sourcebuild.texi (vect_masked_store): Document.
5358 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5359 Alan Hayward <alan.hayward@arm.com>
5360 David Sherwood <david.sherwood@arm.com>
5362 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
5364 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5365 Alan Hayward <alan.hayward@arm.com>
5366 David Sherwood <david.sherwood@arm.com>
5368 * doc/sourcebuild.texi (vect_variable_length): Document.
5370 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5371 Alan Hayward <alan.hayward@arm.com>
5372 David Sherwood <david.sherwood@arm.com>
5374 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
5376 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5377 Alan Hayward <alan.hayward@arm.com>
5378 David Sherwood <david.sherwood@arm.com>
5380 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
5382 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5383 Alan Hayward <alan.hayward@arm.com>
5384 David Sherwood <david.sherwood@arm.com>
5386 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
5387 previously undocumented selectors.
5388 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
5390 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5392 * doc/rtl.texi (const_vector): Say that elements can be
5393 const_wide_ints too.
5394 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
5395 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
5396 (gen_vec_duplicate): Use it instead of CONSTANT_P.
5397 * optabs.c (expand_vector_broadcast): Likewise.
5399 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5400 Alan Hayward <alan.hayward@arm.com>
5401 David Sherwood <david.sherwood@arm.com>
5403 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
5404 scaled index even if the unscaled address was invalid.
5405 Don't increase the complexity of using a scale in that case.
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/rtl.texi: Rewrite the subreg rules so that they partition
5412 the inner register into REGMODE_NATURAL_SIZE bytes rather than
5413 UNITS_PER_WORD bytes.
5414 * emit-rtl.c (validate_subreg): Divide subregs into blocks
5415 based on REGMODE_NATURAL_SIZE of the inner mode.
5416 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
5417 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
5418 * expmed.c (lowpart_bit_field_p): Divide the value up into
5419 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
5420 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
5421 whether something is likely to occupy more than one register.
5423 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
5426 * ipa-inline-transform.c (update_noncloned_frequencies): Use
5427 profile_count::adjust_for_ipa_scaling.
5428 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
5429 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
5431 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
5433 2017-11-09 Jakub Jelinek <jakub@redhat.com>
5435 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
5436 (split_group): Add total_orig and total_new arguments, estimate the
5437 number of statements related to the store group without store merging
5438 and with store merging.
5439 (imm_store_chain_info::output_merged_store): Adjust split_group
5440 callers, punt if estimated number of statements with store merging
5441 is not smaller than estimated number of statements without it.
5443 (handled_load): Remove has_single_use checks.
5444 (pass_store_merging::process_store): Likewise.
5446 2017-11-09 Richard Biener <rguenther@suse.de>
5448 PR tree-optimization/82902
5449 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
5451 2017-11-09 Martin Liska <mliska@suse.cz>
5454 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
5457 2017-11-09 Martin Liska <mliska@suse.cz>
5459 PR tree-optimization/82669
5460 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
5462 2017-11-09 Martin Liska <mliska@suse.cz>
5464 PR gcov-profile/48463
5465 * coverage.c (coverage_begin_function): Output also end locus
5466 of a function and information whether the function is
5468 * gcov-dump.c (tag_function): Parse and print the information.
5469 * gcov.c (INCLUDE_MAP): Add include.
5470 (INCLUDE_SET): Likewise.
5471 (struct line_info): Move earlier in the source file because
5472 of vector<line_info> in function_info structure.
5473 (line_info::line_info): Likewise.
5474 (line_info::has_block): Likewise.
5475 (struct source_info): Add new member index.
5476 (source_info::get_functions_at_location): New function.
5477 (function_info::group_line_p): New function.
5478 (output_intermediate_line): New function.
5479 (output_intermediate_file): Use the mentioned function.
5480 (struct function_start): New.
5481 (struct function_start_pair_hash): Likewise.
5482 (process_file): Add code that identifies group functions.
5483 Assign lines either to global or function scope.
5484 (generate_results): Skip artificial functions.
5485 (find_source): Assign index for each source file.
5486 (read_graph_file): Read new flag artificial and end_line.
5487 (add_line_counts): Assign it either to global of function scope.
5488 (accumulate_line_counts): Isolate core of the function to
5489 accumulate_line_info and call it for both function and global
5491 (accumulate_line_info): New function.
5492 (output_line_beginning): Fix GNU coding style.
5493 (print_source_line): New function.
5494 (output_line_details): Likewise.
5495 (output_function_details): Likewise.
5496 (output_lines): Iterate both source (global) scope and function
5498 (struct function_line_start_cmp): New class.
5499 * doc/gcov.texi: Reflect changes in documentation.
5501 2017-11-09 Jakub Jelinek <jakub@redhat.com>
5504 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
5505 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
5506 and similarly for not instead of neg.
5508 2017-11-08 Andi Kleen <ak@linux.intel.com>
5510 * config/i386/i386.opt: Add -mforce-indirect-call.
5511 * config/i386/predicates.md: Check for flag_force_indirect_call.
5512 * doc/invoke.texi: Document -mforce-indirect-call
5514 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
5516 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
5518 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
5519 config/riscv/riscv.c (predict.h): New include.
5520 (riscv_slow_unaligned_access_p): No longer static.
5521 (riscv_block_move_straight): Add require.
5522 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
5524 2017-11-08 Jakub Jelinek <jakub@redhat.com>
5527 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
5528 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
5529 nonimmediate_operand predicate for operand 1 instead of
5532 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5534 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
5536 * config/aarch64/constraints.md (Uml): New constraint.
5537 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
5540 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5542 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
5543 of two vec_duplicates into a vec_concat.
5545 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5547 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
5548 Simplify vec_merge of vec_duplicate and vec_concat.
5549 * config/aarch64/constraints.md (Utq): New constraint.
5550 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
5553 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5555 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
5556 Simplify vec_merge of vec_duplicate and const_vector.
5557 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
5559 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
5560 mode iterator. Update predicate on operand 1 to
5561 handle non-const_vec constants. Delete constraints.
5562 (*aarch64_combinez_be<mode>): Likewise for operand 2.
5564 2017-11-08 Jakub Jelinek <jakub@redhat.com>
5566 PR tree-optimization/78821
5567 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
5569 (store_operand_info::store_operand_info): Initialize it to false.
5570 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
5571 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
5572 store in the group, and if chain_info is non-NULL, to ignore altogether
5574 (compatible_load_p): Fail if bit_not_p does not match.
5575 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
5576 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
5577 (pass_store_merging::process_store): Adjust
5578 terminate_all_aliasing_chains calls to pass NULL in all current spots,
5579 call terminate_all_aliasing_chains newly when adding a store into
5580 a chain with non-NULL chain_info.
5582 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
5584 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
5586 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
5588 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5590 (aarch64_layout_frame): Initialise emit_frame_chain.
5591 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
5592 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
5594 2017-11-08 Martin Liska <mliska@suse.cz>
5596 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
5597 of gimple_call_internal_p.
5599 2017-11-07 Tom de Vries <tom@codesourcery.com>
5601 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
5603 2017-11-07 Tom de Vries <tom@codesourcery.com>
5605 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5608 2017-11-08 Martin Liska <mliska@suse.cz>
5611 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
5613 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
5615 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
5617 2017-11-07 Jakub Jelinek <jakub@redhat.com>
5620 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
5621 (*cmp<mode>_ccz_1): New insn with $k alternative.
5624 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
5625 RTX_COMM_COMPARE as commutative as well.
5626 (ix86_binary_operator_ok): Formatting fix.
5627 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
5628 *<code><mode>3<mask_name><round_saeonly_name>,
5629 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
5630 *<s>mul<mode>3_highpart<mask_name>,
5631 *vec_widen_umult_even_v16si<mask_name>,
5632 *vec_widen_umult_even_v8si<mask_name>,
5633 *vec_widen_umult_even_v4si<mask_name>,
5634 *vec_widen_smult_even_v16si<mask_name>,
5635 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
5636 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
5637 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
5638 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
5639 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
5640 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
5641 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
5642 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
5643 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
5644 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
5645 ix86_binary_operator_ok. Formatting fixes.
5646 (*<plusminus_insn><mode>3<mask_name><round_name>,
5647 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
5650 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
5652 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
5653 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
5654 eq and ne if TARGET_ISEL.
5655 (cmp): New code_iterator.
5656 (UNS, UNSU_, UNSIK): New code_attrs.
5657 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
5658 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
5660 ("eq<mode>3"): ... this.
5661 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
5663 ("ne<mode>3"): ... this.
5665 2017-11-07 Julia Koval <julia.koval@intel.com>
5668 * common/config/i386/i386-common.c
5669 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
5670 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
5671 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
5672 * config/i386/i386.opt: Ditto.
5673 * config/i386/i386.c (ix86_target_string): Ditto.
5674 (ix86_option_override_internal): Ditto.
5675 (ix86_init_mpx_builtins): Move MPX to args2.
5676 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
5677 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
5678 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
5679 __builtin_ia32_vgf2p8affineinvqb_v32qi,
5680 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
5681 __builtin_ia32_vgf2p8affineinvqb_v16qi,
5682 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
5684 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
5687 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
5688 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
5689 (zero-extendsidi peephole2): Remove peephole.
5691 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
5694 * stor-layout.c: Include attribs.h.
5695 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
5696 explicit lookup of "aligned" attribute.
5698 2017-11-07 Andrew Waterman <andrew@sifive.com>
5700 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
5701 (riscv_expand_block_move): Likewise.
5702 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
5704 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
5705 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
5706 * config/riscv/riscv.c (riscv_block_move_straight): New function.
5707 (riscv_adjust_block_mem): Likewise.
5708 (riscv_block_move_loop): Likewise.
5709 (riscv_expand_block_move): Likewise.
5710 * config/riscv/riscv.md (movmemsi): New pattern.
5712 2017-11-07 Michael Clark <michaeljclark@mac.com>
5714 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
5715 (MUSL_DYNAMIC_LINKER): Likewise.
5717 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
5719 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
5720 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
5723 2017-11-07 Richard Biener <rguenther@suse.de>
5725 * match.pd: Fix build.
5727 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
5728 Jackson Woodruff <jackson.woodruff@arm.com>
5730 PR tree-optimization/71026
5731 * match.pd: Canonicalize negate in division.
5733 2017-11-07 Sudakshina Das <sudi.das@arm.com>
5736 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
5738 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5740 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
5741 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
5743 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5745 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
5746 non-scalar integral types.
5747 * match.pd (negate_expr_p): Handle MINUS_EXPR.
5748 (-(A-B), -(~A)): New transformations.
5750 2017-11-07 Tom de Vries <tom@codesourcery.com>
5752 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
5753 semicolon after "do {} while (0)".
5754 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5755 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5756 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5757 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5758 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5760 2017-11-07 Tom de Vries <tom@codesourcery.com>
5762 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
5763 after "do {} while (0)".
5764 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5765 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5766 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5767 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5768 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5770 2017-11-07 Tom de Vries <tom@codesourcery.com>
5773 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
5775 (arm_rtx_costs_internal): Add missing semicolon after
5776 HANDLE_NARROW_SHIFT_ARITH call.
5778 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5780 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
5781 disable isel if it was not set explicitly.
5783 2017-11-06 James Bowman <james.bowman@ftdichip.com>
5785 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
5786 (add_type_attribute) likewise.
5788 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
5790 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
5791 of struct ix86_frame.
5792 (ix86_initial_elimination_offset): Likewise.
5793 (ix86_expand_split_stack_prologue): Likewise.
5795 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
5797 * tree-vrp.h (enum value_range_type): Update stale comment.
5799 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5800 Alan Hayward <alan.hayward@arm.com>
5801 David Sherwood <david.sherwood@arm.com>
5803 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
5804 (aarch64_expand_vec_perm_const): Take the number of units too.
5805 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
5806 (aarch64_expand_vec_perm_const): Likewise.
5807 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
5808 (vec_perm<mode>): Update accordingly.
5810 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5811 Alan Hayward <alan.hayward@arm.com>
5812 David Sherwood <david.sherwood@arm.com>
5814 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
5815 Take the number of units too.
5816 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
5817 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
5818 but check for a vector mode before rather than after the call.
5819 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
5820 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
5821 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
5822 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
5823 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
5824 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
5825 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
5826 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
5827 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
5828 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
5829 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
5830 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
5831 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
5832 (aarch64_sqdmull2_n<mode>): Update accordingly.
5834 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5835 Alan Hayward <alan.hayward@arm.com>
5836 David Sherwood <david.sherwood@arm.com>
5838 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
5839 the number of units too.
5840 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
5841 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
5842 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
5843 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
5844 (vec_store_lanesxi<mode>): Update accordingly.
5846 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5847 Alan Hayward <alan.hayward@arm.com>
5848 David Sherwood <david.sherwood@arm.com>
5850 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
5851 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
5852 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
5853 of units rather than the mode.
5854 * config/aarch64/iterators.md (nunits): New mode attribute.
5855 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
5856 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
5857 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
5858 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
5859 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5860 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
5861 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
5862 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
5863 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
5864 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
5865 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5866 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
5867 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
5868 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
5869 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
5870 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
5871 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
5872 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
5873 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
5874 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
5875 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
5876 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
5877 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
5878 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
5879 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
5880 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
5881 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
5882 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
5883 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
5884 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
5885 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
5886 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
5887 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
5888 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
5889 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
5890 (aarch64_simd_vec_setv2di): Likewise.
5892 2017-11-06 Carl Love <cel@us.ibm.com>
5894 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
5896 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
5897 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
5899 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
5901 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
5902 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
5903 Add power 8 macro expansions.
5904 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
5905 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
5906 power 8 instructions. (VSX_XXBR): Add iterator.
5908 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
5910 * config/arm/arm.md (predicable_short_it): Change default to "no",
5911 improve documentation, remove uses that are identical to the default.
5912 (enabled_for_depr_it): Rename to enabled_for_short_it.
5913 * gcc/config/arm/arm-fixed.md (predicable_short_it):
5914 Remove default uses.
5915 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
5916 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
5917 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
5918 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
5920 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
5923 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
5924 float128 helper macros, which are no longer used after deleting
5925 the old 'q' built-in functions, and moving the round to odd
5926 built-in functions to being special built-in functions.
5927 (BU_FLOAT128_2): Likewise.
5928 (BU_FLOAT128_1_HW): Likewise.
5929 (BU_FLOAT128_2_HW): Likewise.
5930 (BU_FLOAT128_3_HW): Likewise.
5931 (FABSQ): Delete old 'q' built-in functions.
5932 (COPYSIGNQ): Likewise.
5933 (SQRTF128_ODD): Move round to odd built-in functions to be
5934 special built-in functions, so that we can handle
5935 -mabi=ieeelongdouble.
5936 (TRUNCF128_ODD): Likewise.
5937 (ADDF128_ODD): Likewise.
5938 (SUBF128_ODD): Likewise.
5939 (MULF128_ODD): Likewise.
5940 (DIVF128_ODD): Likewise.
5941 (FMAF128_ODD): Likewise.
5942 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
5943 built-in names to 'f128'.
5944 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
5945 old 'q' built-in functions, as the machine independent code for
5946 'f128' built-in functions handles this.
5947 (rs6000_expand_builtin): Add expansion for float128 round to odd
5948 functions, keying off on -mabi=ieeelongdouble of whether to use
5949 the KFmode or TFmode variant.
5950 (rs6000_init_builtins): Initialize the _Float128 round to odd
5952 * doc/extend.texi (PowerPC Built-in Functions): Document the old
5953 _Float128 'q' built-in functions are now mapped into the new
5954 'f128' built-in functions.
5956 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
5958 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
5959 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
5961 2017-11-06 David Malcolm <dmalcolm@redhat.com>
5964 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
5965 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
5966 * toplev.c: Include "ipa-fnsummary.h".
5967 (toplev::finalize): Call ipa_fnsummary_c_finalize.
5969 2017-11-06 Jakub Jelinek <jakub@redhat.com>
5971 PR tree-optimization/82838
5972 * gimple-ssa-store-merging.c
5973 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
5974 on a separate gimple_seq which is then appended to seq.
5976 2017-11-06 Jeff Law <law@redhat.com>
5979 * config/i386/i386.c (PROBE_INTERVAL): Remove.
5980 (get_probe_interval): New functions.
5981 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
5982 (ix86_adjust_stack_and_probe): Likewise.
5983 (output_adjust_stack_and_probe): Likewise.
5984 (ix86_emit_probe_stack_range): Likewise.
5985 (ix86_expand_prologue): Likewise.
5987 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5989 PR tree-optimization/82816
5990 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
5991 if the modes of the two types are the same.
5992 (convert_plusminus_to_widen): Likewise.
5994 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5996 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
5998 (usadv16qi): New define_expand.
5999 (usadv8hi): New define_expand.
6001 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
6004 * ipa-inline-transform.c (update_noncloned_frequencies): Always
6006 (inline_transform): Likewise.
6007 * predict.c (counts_to_freqs): Remove useless conditional.
6008 * profile-count.h (profile_count::apply_scale): Move sanity check.
6009 * tree-inline.c (copy_bb): Always scale.
6010 (copy_cfg_body): Likewise.
6012 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
6015 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
6017 (*cmp_ite0): Add enabled_for_depr_it attribute.
6018 (*cmp_ite1): Likewise.
6020 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
6022 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
6025 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6027 * tree-vrp.c (vrp_int_const_binop): Return true on success and
6028 return the value by pointer.
6029 (extract_range_from_multiplicative_op_1): Update accordingly.
6030 Return as soon as an operation fails.
6032 2017-11-05 Tom de Vries <tom@codesourcery.com>
6035 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
6036 (DEF_SANITIZER_BUILTIN): ... here.
6037 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
6038 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
6040 2017-11-05 Tom de Vries <tom@codesourcery.com>
6043 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
6045 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
6046 ASM_OUTPUT_BEFORE_CASE_LABEL call.
6047 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
6049 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
6050 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
6051 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
6053 2017-11-05 Tom de Vries <tom@codesourcery.com>
6056 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
6059 2017-11-04 Michael Clark <michaeljclark@mac.com>
6061 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
6062 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
6064 (*addsi3_extended): Likewise.
6065 (*addsi3_extended2): Likewise.
6066 (<optab>si3): Likewise.
6067 (<optab>di3): Likewise.
6068 (<optab><mode>3): Likewise.
6069 (<*optabe>si3_internal): Likewise.
6070 (zero_extendqi<SUPERQI:mode>2): Likewise.
6071 (*add<mode>hi3): Likewise.
6072 (*xor<mode>hi3): Likewise.
6073 (<optab>di3): Likewise.
6074 (*<optab>si3_extend): Likewise.
6075 (*sge<u>_<X:mode><GPR:mode>): Likewise.
6076 (*slt<u>_<X:mode><GPR:mode>): Likewise.
6077 (*sle<u>_<X:mode><GPR:mode>): Likewise.
6079 2017-11-04 Andrew Waterman <andrew@sifive.com>
6081 * config/riscv/riscv.c (riscv_option_override): Conditionally set
6082 TARGET_STRICT_ALIGN based upon -mtune argument.
6084 2017-11-04 Andrew Waterman <andrew@sifive.com>
6086 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
6088 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
6090 * config/i386/i386.c (choose_basereg): Use optional scratch
6091 register and add assertion.
6092 (x86_emit_outlined_ms2sysv_save): Use scratch register when
6093 needed, and don't allocate stack.
6094 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
6095 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
6096 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
6098 2017-11-03 Jeff Law <law@redhat.com>
6100 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
6101 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
6102 to probe at the start of a noreturn function.
6104 2017-11-03 Jakub Jelinek <jakub@redhat.com>
6106 PR tree-optimization/78821
6107 * gimple-ssa-store-merging.c: Update the file comment.
6108 (MAX_STORE_ALIAS_CHECKS): Define.
6109 (struct store_operand_info): New type.
6110 (store_operand_info::store_operand_info): New constructor.
6111 (struct store_immediate_info): Add rhs_code and ops data members.
6112 (store_immediate_info::store_immediate_info): Add rhscode, op0r
6113 and op1r arguments to the ctor, initialize corresponding data members.
6114 (struct merged_store_group): Add load_align_base and load_align
6116 (merged_store_group::merged_store_group): Initialize them.
6117 (merged_store_group::do_merge): Update them.
6118 (merged_store_group::apply_stores): Pick the constant for
6119 encode_tree_to_bitpos from one of the two operands, or skip
6120 encode_tree_to_bitpos if neither operand is a constant.
6121 (class pass_store_merging): Add process_store method decl. Remove
6122 bool argument from terminate_all_aliasing_chains method decl.
6123 (pass_store_merging::terminate_all_aliasing_chains): Remove
6124 var_offset_p argument and corresponding handling.
6125 (stmts_may_clobber_ref_p): New function.
6126 (compatible_load_p): New function.
6127 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
6128 if there is overlap and rhs_code is not INTEGER_CST. For
6129 non-overlapping stores terminate group if rhs is not mergeable.
6130 (get_alias_type_for_stmts): Change first argument from
6131 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
6132 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
6133 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
6135 (get_location_for_stmts): Change first argument from
6136 auto_vec<gimple *> & to vec<gimple *> &.
6137 (struct split_store): Remove orig_stmts data member, add orig_stores.
6138 (split_store::split_store): Create orig_stores rather than orig_stmts.
6139 (find_constituent_stmts): Renamed to ...
6140 (find_constituent_stores): ... this. Change second argument from
6141 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
6142 to info structures rather than the statements.
6143 (split_group): Rename ALLOW_UNALIGNED argument to
6144 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
6145 it. Adjust find_constituent_stores caller.
6146 (imm_store_chain_info::output_merged_store): Handle rhs_code other
6147 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
6148 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
6149 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
6150 (mem_valid_for_store_merging): New function.
6151 (handled_load): New function.
6152 (pass_store_merging::process_store): New method.
6153 (pass_store_merging::execute): Use process_store method. Adjust
6154 terminate_all_aliasing_chains caller.
6156 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6158 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
6159 Return true for more constants, symbols and label references.
6160 (aarch64_valid_floating_const): Remove unused function.
6162 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6165 * config/aarch64/aarch64.c (aarch64_layout_frame):
6166 Undo forcing of LR at bottom of frame.
6168 2017-11-03 Jeff Law <law@redhat.com>
6171 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
6172 for int_registers_saved.
6174 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
6175 extracted from tree-ssa-dom.c.
6176 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
6177 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
6178 (record_equivalences_from_incoming_edge): Add additional argument
6179 to single_pred_edge_ignoring_loop_edges call.
6180 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
6181 (uncprop_dom_walker::before_dom_children): Add additional argument
6182 to single_pred_edge_ignoring_loop_edges call.
6183 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
6184 single_pred_edge_ignoring_loop_edges rather than open coding.
6185 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
6187 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
6189 * match.pd (-(-A)): Rewrite.
6191 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
6193 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
6194 (rs6000_emit_int_cmove): New declaration.
6195 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
6196 (rs6000_emit_sISEL): Delete.
6197 (rs6000_emit_int_cmove): Make non-static.
6198 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
6199 instead of rs6000_emit_sISEL.
6201 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
6203 * asan.c (create_cond_insert_point): Maintain profile.
6204 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
6206 * basic-block.h (struct basic_block_def): Remove frequency.
6207 (EDGE_FREQUENCY): Use to_frequency
6208 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
6210 (find_traces): Update to use to_frequency.
6211 (find_traces_1_round): Likewise; use only IPA counts.
6212 (bb_to_key): Likewise.
6213 (connect_traces): Use IPA counts only.
6214 (copy_bb_p): Update to use to_frequency.
6215 (fix_up_crossing_landing_pad): Likewise.
6216 (sanitize_hot_paths): Likewise.
6217 * bt-load.c (basic_block_freq): Likewise.
6218 * cfg.c (init_flow): Set count_max to uninitialized.
6219 (check_bb_profile): Remove frequencies; check counts.
6220 (dump_bb_info): Do not dump frequencies.
6221 (update_bb_profile_for_threading): Update counts only.
6222 (scale_bbs_frequencies_int): Likewise.
6223 (MAX_SAFE_MULTIPLIER): Remove.
6224 (scale_bbs_frequencies_gcov_type): Update counts only.
6225 (scale_bbs_frequencies_profile_count): Update counts only.
6226 (scale_bbs_frequencies): Update counts only.
6227 * cfg.h (struct control_flow_graph): Add count-max.
6228 (update_bb_profile_for_threading): Update prototype.
6229 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
6230 (find_many_sub_basic_blocks): Likewise.
6231 * cfgcleanup.c (try_forward_edges): Likewise.
6232 (try_crossjump_to_edge): Likewise.
6233 * cfgexpand.c (expand_gimple_cond): Likewise.
6234 (expand_gimple_tailcall): Likewise.
6235 (construct_init_block): Likewise.
6236 (construct_exit_block): Likewise.
6237 * cfghooks.c (verify_flow_info): Check consistency of counts.
6238 (dump_bb_for_graph): Do not dump frequencies.
6239 (split_block_1): Do not update frequencies.
6240 (split_edge): Do not update frequencies.
6241 (make_forwarder_block): Do not update frequencies.
6242 (duplicate_block): Do not update frequencies.
6243 (account_profile_record): Do not update frequencies.
6244 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
6245 for global heuristics.
6246 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
6247 (expected_loop_iterations_unbounded): Use counts only.
6248 * cfgloopmanip.c (scale_loop_profile): Simplify.
6249 (create_empty_loop_on_edge): Simplify
6251 (duplicate_loop_to_header_edge): Simplify
6252 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
6253 (update_br_prob_note): Take care of removing note when profile
6255 (relink_block_chain): Do not dump frequency.
6256 (rtl_account_profile_record): Use to_frequency.
6257 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
6258 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
6259 (cgraph_update_edges_for_call_stmt_node): Likewise.
6260 (cgraph_edge::verify_count_and_frequency): Update.
6261 (cgraph_node::verify_node): Temporarily disable frequency verification.
6262 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
6263 to_cgraph_frequency.
6264 (cgraph_edge::rebuild_edges): Convert to ipa counts.
6265 * cgraphunit.c (init_lowered_empty_function): Do not initialize
6267 (cgraph_node::expand_thunk): Update profile.
6268 * except.c (dw2_build_landing_pads): Do not update frequency.
6269 * final.c (compute_alignments): Use to_frequency.
6270 (dump_basic_block_info): Do not dump frequency.
6271 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
6272 (dump_gimple_bb_header): Do not dump frequency.
6273 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
6275 * gimple-streamer-in.c (input_bb): Do not stream frequency.
6276 * gimple-streamer-out.c (output_bb): Do not stream frequency.
6277 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
6278 (init_before_recovery): Do not update frequency.
6279 (sched_create_recovery_edges): Do not update frequency.
6280 * hsa-gen.c (convert_switch_statements): Do not update frequency.
6281 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
6282 (ipa_cp_c_finalize): Set max_count to uninitialized.
6283 * ipa-fnsummary.c (get_minimal_bb): Use counts.
6284 (param_change_prob): Use counts.
6285 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
6287 * ipa-split.c (consider_split): Use to_frequency.
6288 (split_function): Use to_frequency.
6289 * ira-build.c (loop_compare_func): Likewise.
6290 (mark_loops_for_removal): Likewise.
6291 (mark_all_loops_for_removal): Likewise.
6292 * loop-doloop.c (doloop_modify): Do not update frequency.
6293 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
6295 * lto-streamer-in.c (input_function): Update count_max.
6296 * omp-expand.c (expand_omp_taskreg): Update count_max.
6297 * omp-simd-clone.c (simd_clone_adjust): Update profile.
6298 * predict.c (maybe_hot_frequency_p): Use to_frequency.
6299 (maybe_hot_count_p): Use ipa counts only.
6300 (maybe_hot_bb_p): Simplify.
6301 (maybe_hot_edge_p): Simplify.
6302 (probably_never_executed): Do not take frequency argument.
6303 (probably_never_executed_bb_p): Do not pass frequency.
6304 (probably_never_executed_edge_p): Likewise.
6305 (combine_predictions_for_bb): Check that profile is nonzero.
6306 (propagate_freq): Do not set frequency.
6307 (drop_profile): Simplify.
6308 (counts_to_freqs): Simplify.
6309 (expensive_function_p): Use to_frequency.
6310 (propagate_unlikely_bbs_forward): Simplify.
6311 (determine_unlikely_bbs): Simplify.
6312 (estimate_bb_frequencies): Add hack to silence graphite issues.
6313 (compute_function_frequency): Use ipa counts.
6314 (pass_profile::execute): Update.
6315 (rebuild_frequencies): Use counts only.
6316 (force_edge_cold): Use counts only.
6317 * profile-count.c (profile_count::dump): Dump new count types.
6318 (profile_count::differs_from_p): Check compatiblity.
6319 (profile_count::to_frequency): New function.
6320 (profile_count::to_cgraph_frequency): New function.
6321 * profile-count.h (struct function): Declare.
6322 (enum profile_quality): Add profile_guessed_local and
6323 profile_guessed_global0.
6324 (class profile_proability): Decrease number of bits to 29;
6325 update from_reg_br_prob_note and to_reg_br_prob_note.
6326 (class profile_count: Update comment; decrease number of bits
6327 to 61. Check compatibility.
6328 (profile_count::compatible_p): New private member function.
6329 (profile_count::ipa_p): New member function.
6330 (profile_count::operator<): Handle global zero correctly.
6331 (profile_count::operator>): Handle global zero correctly.
6332 (profile_count::operator<=): Handle global zero correctly.
6333 (profile_count::operator>=): Handle global zero correctly.
6334 (profile_count::nonzero_p): New member function.
6335 (profile_count::force_nonzero): New member function.
6336 (profile_count::max): New member function.
6337 (profile_count::apply_scale): Handle IPA scalling.
6338 (profile_count::guessed_local): New member function.
6339 (profile_count::global0): New member function.
6340 (profile_count::ipa): New member function.
6341 (profile_count::to_frequency): Declare.
6342 (profile_count::to_cgraph_frequency): Declare.
6343 * profile.c (OVERLAP_BASE): Delete.
6344 (compute_frequency_overlap): Delete.
6345 (compute_branch_probabilities): Do not use compute_frequency_overlap.
6346 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
6347 * sched-ebb.c (rank): Use counts only.
6348 * shrink-wrap.c (handle_simple_exit): Use counts only.
6349 (try_shrink_wrapping): Use counts only.
6350 (place_prologue_for_one_component): Use counts only.
6351 * tracer.c (find_best_predecessor): Use to_frequency.
6352 (find_trace): Use to_frequency.
6353 (tail_duplicate): Use to_frequency.
6354 * trans-mem.c (expand_transaction): Do not update frequency.
6355 * tree-call-cdce.c: Do not update frequency.
6356 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
6357 (gimple_merge_blocks): Likewise.
6358 (gimple_split_edge): Likewise.
6359 (gimple_duplicate_sese_region): Likewise.
6360 (gimple_duplicate_sese_tail): Likewise.
6361 (move_sese_region_to_fn): Likewise.
6362 (gimple_account_profile_record): Likewise.
6363 (insert_cond_bb): Likewise.
6364 * tree-complex.c (expand_complex_div_wide): Likewise.
6365 * tree-eh.c (lower_resx): Update profile.
6366 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
6368 (initialize_cfun): Do not initialize frequencies
6369 (freqs_to_counts): Delete.
6370 (copy_cfg_body): Ignore count parameter.
6371 (copy_body): Update.
6372 (expand_call_inline): Update count_max.
6373 (optimize_inline_calls): Update count_max.
6374 (tree_function_versioning): Update count_max.
6375 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
6376 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
6378 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
6379 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
6380 (try_peel_loop): Likewise.
6381 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
6383 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
6384 (tree_transform_and_unroll_loop): Do not use frequencies
6385 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
6386 Use reliable prediction only.
6387 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
6388 * tree-ssa-sink.c (select_best_block): Use to_frequency.
6389 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
6390 probability scaling.
6391 * tree-ssa-threadupdate.c (create_block_for_threading): Do
6392 not update frequency
6393 (any_remaining_duplicated_blocks): Likewise.
6394 (update_profile): Likewise.
6395 (estimated_freqs_path): Delete.
6396 (freqs_to_counts_path): Delete.
6397 (clear_counts_path): Delete.
6398 (ssa_fix_duplicate_block_edges): Likewise.
6399 (duplicate_thread_path): Likewise.
6400 * tree-switch-conversion.c (gen_inbound_check): Use counts.
6401 * tree-tailcall.c (decrease_profile): Do not update frequency.
6402 (eliminate_tail_call): Likewise.
6403 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
6404 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
6405 (optimize_mask_stores): Likewise.
6406 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
6407 * ubsan.c (ubsan_expand_null_ifn): Update profile.
6408 (ubsan_expand_ptr_ifn): Update profile.
6409 * value-prof.c (gimple_ic): Simplify.
6410 * value-prof.h (gimple_ic): Update prototype.
6411 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
6412 * ipa-inline.c (compute_uninlined_call_time): Be sure that
6414 (want_inline_self_recursive_call_p): Likewise.
6415 (resolve_noninline_speculation): Only cummulate defined counts.
6416 (inline_small_functions): Use nonzero_p.
6417 (ipa_inline): Do not access freed node.
6419 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6421 * config/aarch64/aarch64.c (aarch64_override_options_internal):
6422 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
6424 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
6426 * config/riscv/riscv.c (riscv_legitimize_move): Handle
6427 non-legitimate address.
6429 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
6431 * config/rs6000/rs6000.md (*lt0_disi): Delete.
6432 (*lt0_<mode>di, *lt0_<mode>si): New.
6434 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
6436 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
6437 TARGET_PAIRED_FLOAT.
6439 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
6440 Jim Wilson <jim.wilson@linaro.org>
6442 * config/aarch64/aarch64-cores.def (saphira): New CPU.
6443 * config/aarch64/aarch64-tune.md: Regenerated.
6444 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
6445 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
6447 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
6449 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
6450 (arc_expand_prologue): Restore blink for millicode.
6451 * config/arc/linux.h (LINK_EH_SPEC): Defined.
6453 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
6456 * config/i386/i386.c (ix86_vector_duplicate_value): Use
6457 gen_vec_duplicate after forcing the scalar into a register.
6459 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
6461 * combine (try_combine): Print the insns input to try_combine to the
6464 2017-11-02 Steve Ellcey <sellcey@cavium.com>
6467 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
6468 Remove second argument from aarch64_process_target_attr call.
6469 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
6471 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
6473 (aarch64_handle_attr_arch): Remove second argument.
6474 (aarch64_handle_attr_cpu): Ditto.
6475 (aarch64_handle_attr_tune): Ditto.
6476 (aarch64_handle_attr_isa_flags): Ditto.
6477 (aarch64_process_one_target_attr): Ditto.
6478 (aarch64_process_target_attr): Ditto.
6479 (aarch64_option_valid_attribute_p): Remove second argument.
6480 on aarch64_process_target_attr call.
6482 2017-11-02 David Malcolm <dmalcolm@redhat.com>
6484 * diagnostic.c: Include "selftest-diagnostic.h".
6485 (selftest::assert_location_text): New function.
6486 (selftest::test_diagnostic_get_location_text): New function.
6487 (selftest::diagnostic_c_tests): Call it.
6489 2017-11-02 David Malcolm <dmalcolm@redhat.com>
6491 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
6492 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
6493 (class selftest::test_diagnostic_context): Move to...
6494 * selftest-diagnostic.c: New file.
6495 * selftest-diagnostic.h: New file.
6497 2017-11-02 James Bowman <james.bowman@ftdichip.com>
6499 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
6500 offset range for FT32B.
6501 * config/ft32/ft32.h: option "mcompress" enables relaxation.
6502 * config/ft32/ft32.md: Add TARGET_NOPM.
6503 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
6504 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
6506 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
6508 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
6510 2017-11-02 Jeff Law <law@redhat.com>
6512 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
6513 virtual keyword on FINAL OVERRIDE members.
6515 * tree-ssa-propagate.h (ssa_propagation_engine): Group
6516 virtuals together. Add virtual destructor.
6517 (substitute_and_fold_engine): Similarly.
6519 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
6521 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
6523 2017-11-02 Richard Biener <rguenther@suse.de>
6525 PR tree-optimization/82795
6526 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
6528 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6530 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
6532 (gcc_GAS_CHECK_FEATURE): Remove.
6533 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
6534 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
6535 * configure: Regenerate.
6537 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
6539 * config/arc/arc.c (hwloop_optimize): Account for empty
6542 2017-11-02 Richard Biener <rguenther@suse.de>
6545 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
6546 Truncate ARRAY_REF index and element size.
6548 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
6550 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
6552 2017-11-01 Jeff Law <law@redhat.com>
6554 * tree-ssa-ccp.c (ccp_folder): New class derived from
6555 substitute_and_fold_engine.
6556 (ccp_folder::get_value): New member function.
6557 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
6558 (ccp_fold_stmt): Remove prototype.
6559 (ccp_finalize): Call substitute_and_fold from the ccp_class.
6560 * tree-ssa-copy.c (copy_folder): New class derived from
6561 substitute_and_fold_engine.
6562 (copy_folder::get_value): Renamed from get_value.
6563 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
6564 * tree-vrp.c (vrp_folder): New class derived from
6565 substitute_and_fold_engine.
6566 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
6567 (vrp_folder::get_value): New member function.
6568 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
6569 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
6570 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
6571 provide a class interface to folder/substitute routines.
6572 (ssa_prop_fold_stmt_fn): Remove typedef.
6573 (ssa_prop_get_value_fn): Likewise.
6574 (subsitute_and_fold): Remove prototype.
6575 (replace_uses_in): Likewise.
6576 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
6577 Renamed from replace_uses_in. Call the virtual member function
6578 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
6579 (substitute_and_fold_dom_walker): Remove initialization of
6580 data member entries for calbacks. Add substitute_and_fold_engine
6581 member and initialize it.
6582 (substitute_and_fold_dom_walker::before_dom_children0: Use the
6583 member functions for get_value, replace_phi_args_in c
6584 replace_uses_in, and fold_stmt calls.
6585 (substitute_and_fold_engine::substitute_and_fold): Renamed from
6586 substitute_and_fold. Remove assert. Update ctor call.
6588 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
6589 (ssa_prop_visit_phi_fn): Likewise.
6590 (class ssa_propagation_engine): New class to provide an interface
6592 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
6594 (ssa_prop_visit_phi): Likewise.
6595 (ssa_propagation_engine::simulate_stmt): Moved into class.
6596 Call visit_phi/visit_stmt from the class rather than via
6597 file scoped static variables.
6598 (ssa_propagation_engine::simulate_block): Moved into class.
6599 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
6600 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
6601 set file scoped statics for the visit_stmt/visit_phi callbacks.
6602 * tree-complex.c (complex_propagate): New class derived from
6603 ssa_propagation_engine.
6604 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
6605 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
6606 (tree_lower_complex): Call ssa_propagate via the complex_propagate
6608 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
6609 ssa_propagation_engine.
6610 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
6611 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
6612 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
6613 * tree-ssa-copy.c (copy_prop): New class derived from
6614 ssa_propagation_engine.
6615 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
6616 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
6617 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
6618 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
6619 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
6620 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
6621 (execute_vrp): Call ssa_propagate from the vrp_prop class.
6623 2017-11-01 Jakub Jelinek <jakub@redhat.com>
6625 PR rtl-optimization/82778
6626 PR rtl-optimization/82597
6627 * compare-elim.c (struct comparison): Add in_a_setter field.
6628 (find_comparison_dom_walker::before_dom_children): Remove killed
6629 bitmap and df_simulate_find_defs call, instead walk the defs.
6630 Compute last_setter and initialize in_a_setter. Merge definitions
6631 with first initialization for a few variables.
6632 (try_validate_parallel): Use insn_invalid_p instead of
6633 recog_memoized. Return insn rather than just the pattern.
6634 (try_merge_compare): Fix up comment. Don't uselessly test if
6635 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
6637 (execute_compare_elim_after_reload): Remove df_chain_add_problem
6640 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6641 Alan Hayward <alan.hayward@arm.com>
6642 David Sherwood <david.sherwood@arm.com>
6644 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
6645 aarch64_hard_regno_nregs to get the number of registers
6648 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6649 Alan Hayward <alan.hayward@arm.com>
6650 David Sherwood <david.sherwood@arm.com>
6652 * config/aarch64/constraints.md (Upl): Rename to...
6654 * config/aarch64/aarch64.md
6655 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
6658 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6659 Alan Hayward <alan.hayward@arm.com>
6660 David Sherwood <david.sherwood@arm.com>
6662 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
6663 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
6666 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6667 Alan Hayward <alan.hayward@arm.com>
6668 David Sherwood <david.sherwood@arm.com>
6670 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
6671 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
6672 (aarch64_evpc_dup): Generate rtl direcly, rather than using
6674 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
6675 of a single element.
6676 * config/aarch64/iterators.md: Add a comment above the permute
6677 unspecs to say that they are generated directly by
6678 aarch64_expand_vec_perm_const.
6679 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
6681 2017-11-01 Nathan Sidwell <nathan@acm.org>
6683 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
6685 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
6687 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
6688 and medany code models, and describe what they do.
6690 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6692 Revert accidental duplicate:
6694 * combine.c (can_change_dest_mode): Reject changes in
6695 REGMODE_NATURAL_SIZE.
6697 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
6699 PR rtl-optimization/64682
6700 PR rtl-optimization/69567
6701 PR rtl-optimization/69737
6702 PR rtl-optimization/82683
6703 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
6704 register mentioned in the note, drop the note, unless it came from I3,
6705 in which case it should go to I3 again.
6707 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6709 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
6710 and return false if not.
6711 (clear_bytes_written_by, live_bytes_read): Update accordingly.
6713 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6715 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
6716 range is known to be empty.
6718 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6719 Alan Hayward <alan.hayward@arm.com>
6720 David Sherwood <david.sherwood@arm.com>
6722 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
6723 and CONST_VECTOR_NUNITS instead of computing the number of units from
6724 the byte sizes of the vector and element.
6725 (simplify_binary_operation_1): Likewise.
6726 (simplify_const_binary_operation): Likewise.
6727 (simplify_ternary_operation): Likewise.
6729 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6730 Alan Hayward <alan.hayward@arm.com>
6731 David Sherwood <david.sherwood@arm.com>
6733 * var-tracking.c (INT_MEM_OFFSET): Replace with...
6734 (int_mem_offset): ...this new function.
6735 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
6736 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
6737 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
6740 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6741 Alan Hayward <alan.hayward@arm.com>
6742 David Sherwood <david.sherwood@arm.com>
6744 * lower-subreg.c (interesting_mode_p): New function.
6745 (compute_costs, find_decomposable_subregs, decompose_register)
6746 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
6747 (resolve_clobber, dump_choices): Use it.
6749 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6750 Alan Hayward <alan.hayward@arm.com>
6751 David Sherwood <david.sherwood@arm.com>
6753 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
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 * alias.c (find_base_value, find_base_term): Only process integer
6760 truncations. Check the precision rather than the size.
6762 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6763 Alan Hayward <alan.hayward@arm.com>
6764 David Sherwood <david.sherwood@arm.com>
6766 * machmode.h (is_narrower_int_mode): New function
6767 * optabs.c (expand_float, expand_fix): Use it.
6768 * dwarf2out.c (rotate_loc_descriptor): Likewise.
6770 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6771 Alan Hayward <alan.hayward@arm.com>
6772 David Sherwood <david.sherwood@arm.com>
6774 * rtl.h (narrower_subreg_mode): New function.
6775 * ira-color.c (update_costs_from_allocno): Use it.
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 * optabs-query.h (convert_optab_p): New function, split out from...
6782 (convert_optab_handler): ...here.
6783 (widening_optab_handler): Delete.
6784 (find_widening_optab_handler): Remove permit_non_widening parameter.
6785 (find_widening_optab_handler_and_mode): Likewise. Provide an
6786 override that operates on mode class wrappers.
6787 * optabs-query.c (widening_optab_handler): Delete.
6788 (find_widening_optab_handler_and_mode): Remove permit_non_widening
6789 parameter. Assert that the two modes are the same class and that
6790 the "from" mode is narrower than the "to" mode. Use
6791 convert_optab_handler instead of widening_optab_handler.
6792 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
6793 instead of widening_optab_handler.
6794 * expr.c (expand_expr_real_2): Update calls to
6795 find_widening_optab_handler.
6796 * optabs.c (expand_widen_pattern_expr): Likewise.
6797 (expand_binop_directly): Take the insn_code as a parameter.
6798 (expand_binop): Only call find_widening_optab_handler for
6799 conversion optabs; use optab_handler otherwise. Update calls
6800 to find_widening_optab_handler and expand_binop_directly.
6801 Use convert_optab_handler instead of widening_optab_handler.
6802 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
6803 find_widening_optab_handler and use scalar_mode rather than
6805 (convert_plusminus_to_widen): 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 * machmode.h (fixed_size_mode): New class.
6812 * rtl.h (get_pool_mode): Return fixed_size_mode.
6813 * gengtype.c (main): Add fixed_size_mode.
6814 * target.def (get_raw_result_mode): Return a fixed_size_mode.
6815 (get_raw_arg_mode): Likewise.
6816 * doc/tm.texi: Regenerate.
6817 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
6818 * targhooks.c (default_get_reg_raw_mode): Likewise.
6819 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
6820 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
6821 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
6822 (msp430_get_raw_result_mode): Likewise.
6823 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
6824 * dbxout.c (dbxout_parms): Require fixed-size modes.
6825 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
6826 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
6827 * omp-low.c (lower_oacc_reductions): Likewise.
6828 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
6829 (simplify_subreg): Update accordingly.
6830 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
6831 (force_const_mem): Update accordingly. Return NULL_RTX for modes
6832 that aren't fixed-size.
6833 (get_pool_mode): Return a fixed_size_mode.
6834 (output_constant_pool_2): Take a fixed_size_mode.
6836 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6837 Alan Hayward <alan.hayward@arm.com>
6838 David Sherwood <david.sherwood@arm.com>
6840 * doc/rtl.texi (vec_series): Document.
6841 (const): Say that the operand can be a vec_series.
6842 * rtl.def (VEC_SERIES): New rtx code.
6843 * rtl.h (const_vec_series_p_1): Declare.
6844 (const_vec_series_p): New function.
6845 * emit-rtl.h (gen_const_vec_series): Declare.
6846 (gen_vec_series): Likewise.
6847 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
6848 (gen_vec_series): Likewise.
6849 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
6850 * simplify-rtx.c (simplify_unary_operation): Handle negations
6852 (simplify_binary_operation_series): New function.
6853 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
6854 (test_vector_ops_series): New function.
6855 (test_vector_ops): Call it.
6856 * config/powerpcspe/altivec.md (altivec_lvsl): Use
6857 gen_const_vec_series.
6858 (altivec_lvsr): Likewise.
6859 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
6861 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6862 Alan Hayward <alan.hayward@arm.com>
6863 David Sherwood <david.sherwood@arm.com>
6865 * doc/rtl.texi (const): Update description of address constants.
6866 Say that vector constants are allowed too.
6867 * common.md (E, F): Use CONSTANT_P instead of checking for
6869 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
6870 checking for CONST_VECTOR.
6871 * expmed.c (make_tree): Use build_vector_from_val for a CONST
6873 * expr.c (expand_expr_real_2): Check for vector modes instead
6874 of checking for CONST_VECTOR.
6875 * rtl.h (const_vec_p): New function.
6876 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
6877 (unwrap_const_vec_duplicate): Handle them here too.
6879 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6880 David Malcolm <dmalcolm@redhat.com>
6881 Alan Hayward <alan.hayward@arm.com>
6882 David Sherwood <david.sherwood@arm.com>
6884 * rtl.h (vec_duplicate_p): New function.
6885 * selftest-rtl.c (assert_rtx_eq_at): New function.
6886 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
6887 (assert_rtx_eq_at): Declare.
6888 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
6889 * selftest-run-tests.c (selftest::run_tests): Call it.
6890 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
6891 (simplify_unary_operation_1): Recursively handle vector duplicates.
6892 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
6894 (simplify_subreg): Handle subregs of vector duplicates.
6895 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
6896 (selftest::simplify_rtx_c_tests): New functions.
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 * emit-rtl.h (gen_const_vec_duplicate): Declare.
6903 (gen_vec_duplicate): Likewise.
6904 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
6906 (gen_const_vector): ...here.
6907 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
6908 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
6909 whose elements are all equal.
6910 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
6911 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
6912 (simplify_relational_operation): Likewise.
6913 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
6915 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
6916 (aarch64_expand_vector_init): Likewise.
6917 * config/arm/arm.c (neon_vdup_constant): Likewise.
6918 (neon_expand_vector_init): Likewise.
6919 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
6920 (arm_block_set_unaligned_vect): Likewise.
6921 (arm_block_set_aligned_vect): Likewise.
6922 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
6923 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
6924 (expand_vec_perm_even_odd_pack): Likewise.
6925 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
6926 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
6927 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
6928 gen_const_vec_duplicate.
6929 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
6930 * config/mips/mips.c (mips_gen_const_int_vector): Use
6931 gen_const_vec_duplicate.
6932 (mips_expand_vector_init): Use CONST0_RTX.
6933 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
6934 (define_split): Use gen_const_vec_duplicate.
6935 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
6936 (define_split): Use gen_const_vec_duplicate.
6937 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
6938 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
6939 * config/spu/spu.c (spu_const): Likewise.
6941 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6942 Alan Hayward <alan.hayward@arm.com>
6943 David Sherwood <david.sherwood@arm.com>
6945 * combine.c (can_change_dest_mode): Reject changes in
6946 REGMODE_NATURAL_SIZE.
6948 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
6950 * configure.ac (--enable-libssp): New.
6951 (gcc_cv_libc_provides_ssp): Check for explicit setting before
6952 trying to determine target-specific default. Adjust indentation.
6953 * configure: Regenerated.
6954 * doc/install.texi (Configuration): Expand --disable-libssp
6957 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
6959 config/i386/i386.c (ix86_expand_epilogue): Correct stack
6962 2017-10-31 Martin Jambor <mjambor@suse.cz>
6965 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
6967 2017-10-31 David Malcolm <dmalcolm@redhat.com>
6969 * auto-profile.c (autofdo_source_profile::read): Use
6970 UNKNOWN_LOCATION rather than 0.
6971 * diagnostic-core.h (warning_at_rich_loc): Rename to...
6972 (warning_at): ...this overload.
6973 (warning_at_rich_loc_n): Rename to...
6974 (warning_n): ...this overload.
6975 (error_at_rich_loc): Rename to...
6976 (error_at): ...this overload.
6977 (pedwarn_at_rich_loc): Rename to...
6978 (pedwarn): ...this overload.
6979 (permerror_at_rich_loc): Rename to...
6980 (permerror): ...this overload.
6981 (inform_at_rich_loc): Rename to...
6982 (inform): ...this overload.
6983 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
6984 (diagnostic_n_impl_richloc): Rename to...
6985 (diagnostic_n_impl): ...this rich_location *-based decl.
6986 (inform_at_rich_loc): Rename to...
6987 (inform): ...this, and add an assertion.
6988 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
6989 (warning_at_rich_loc): Rename to...
6990 (warning_at): ...this, and add an assertion.
6991 (warning_at_rich_loc_n): Rename to...
6992 (warning_n): ...this, and add an assertion.
6993 (warning_n): Update location_t-based implementation for removal of
6994 location_t-based diagnostic_n_impl.
6995 (pedwarn_at_rich_loc): Rename to...
6996 (pedwarn): ...this, and add an assertion.
6997 (permerror_at_rich_loc): Rename to...
6998 (permerror): ...this, and add an assertion.
6999 (error_n): Update for removal of location_t-based diagnostic_n_impl.
7000 (error_at_rich_loc): Rename to...
7001 (error_at): ...this, and add an assertion.
7002 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
7003 (driver::do_spec_on_infiles): Likewise.
7004 * substring-locations.c (format_warning_va): Update for renaming
7005 of inform_at_rich_loc.
7007 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
7009 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
7010 _Float<N>X built-in functions so that the variant without the
7011 "__builtin_" prefix is only enabled for the GNU C and Objective C
7012 languages when they are in non-strict ANSI/ISO mode.
7013 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
7014 * target.def (floatn_builtin_p): Add a target hook to control
7015 whether _Float<N> and _Float<N>X built-in functions without the
7016 "__builtin_" prefix are enabled, and return true for C and
7017 Objective C in the default hook. Include langhooks.h in
7019 * targhooks.h (default_floatn_builtin_p): Likewise.
7020 * targhooks.c (default_floatn_builtin_p): Likewise.
7021 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
7022 floatn_builtin_p target hook.
7023 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
7025 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
7026 Eric Botcazou <ebotcazou@adacore.com>
7028 PR rtl-optimization/81803
7029 * lra-constraints.c (curr_insn_transform): Also reload the whole
7030 register for a strict subreg no wider than a word if this is for
7031 a WORD_REGISTER_OPERATIONS target.
7033 2017-10-31 Jason Merrill <jason@redhat.com>
7035 * gdbinit.in: Skip over inlines from timevar.h.
7037 2017-10-31 Martin Liska <mliska@suse.cz>
7039 * doc/gcov.texi: Document new option.
7040 * gcov.c (print_usage): Likewise print it.
7041 (process_args): Support the argument.
7042 (format_count): New function.
7043 (format_gcov): Use the function.
7045 2017-10-31 Martin Liska <mliska@suse.cz>
7047 * gcov.c (struct name_map): do not use typedef.
7048 Define operator== and operator<.
7049 (name_search): Remove.
7050 (name_sort): Remove.
7051 (main): Do not allocate names.
7052 (process_file): Add vertical space.
7053 (generate_results): Use std::find.
7054 (release_structures): Do not release memory.
7055 (find_source): Use std::find.
7057 2017-10-31 Martin Liska <mliska@suse.cz>
7059 * gcov.c (struct line_info): Remove it's typedef.
7060 (line_info::line_info): Add proper ctor.
7061 (line_info::has_block): Do not use a typedef.
7062 (struct source_info): Do not use typedef.
7063 (circuit): Likewise.
7064 (get_cycles_count): Likewise.
7065 (output_intermediate_file): Iterate via vector iterator.
7066 (add_line_counts): Use std::vector methods.
7067 (accumulate_line_counts): Likewise.
7068 (output_lines): Likewise.
7070 2017-10-31 Martin Liska <mliska@suse.cz>
7072 * gcov.c (struct source_info): Remove typedef.
7073 (source_info::source_info): Add proper ctor.
7074 (accumulate_line_counts): Use struct, not it's typedef.
7075 (output_gcov_file): Likewise.
7076 (output_lines): Likewise.
7077 (main): Do not allocate an array.
7078 (output_intermediate_file): Use size of vector container.
7079 (process_file): Resize the vector.
7080 (generate_results): Do not preallocate, use newly added vector
7082 (release_structures): Do not release sources.
7083 (find_source): Use vector methods.
7084 (add_line_counts): Do not use typedef.
7086 2017-10-31 Martin Liska <mliska@suse.cz>
7088 * doc/gcov.texi: Document that.
7089 * gcov.c (add_line_counts): Mark lines with a non-executed
7091 (output_line_beginning): Handle such lines.
7092 (output_lines): Pass new argument.
7093 (output_intermediate_file): Print it in intermediate format.
7095 2017-10-31 Martin Liska <mliska@suse.cz>
7097 * color-macros.h: New file.
7098 * diagnostic-color.c: Factor out color related to macros to
7100 * doc/gcov.texi: Document -k option.
7101 * gcov.c (INCLUDE_STRING): Include string.h.
7102 (print_usage): Add -k option.
7103 (process_args): Parse it.
7104 (pad_count_string): New function.
7105 (output_line_beginning): Likewise.
7106 (DEFAULT_LINE_START): New macro.
7107 (output_lines): Support color output.
7109 2017-10-31 Martin Liska <mliska@suse.cz>
7111 PR gcov-profile/82633
7112 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
7113 their interaction with GCOV infrastructure.
7114 * configure.ac: Add -fkeep-{inline,static}-functions to
7116 * configure: Regenerate.
7118 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
7121 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
7123 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
7126 * config/rs6000/rs6000.md (allocate_stack): Force update interval
7127 into a register if it does not fit into an immediate offset field.
7129 2017-10-31 Olivier Hainque <hainque@adacore.com>
7131 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
7133 2017-10-31 Julia Koval <julia.koval@intel.com>
7135 * config.gcc: Add gfniintrin.h.
7136 * config/i386/gfniintrin.h: New.
7137 * config/i386/i386-builtin-types.def
7138 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7139 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7140 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7141 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7142 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7143 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
7144 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
7145 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
7146 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
7147 V64QI_FTYPE_V64QI_V64QI_INT): New types.
7148 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
7149 * config/i386/immintrin.h: Include gfniintrin.h.
7150 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
7152 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
7154 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
7156 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
7158 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
7159 (arm_ashldi3_1bit): Remove pattern.
7160 (ashrdi3): Remove shift by 1 expansion.
7161 (arm_ashrdi3_1bit): Remove pattern.
7162 (lshrdi3): Remove shift by 1 expansion.
7163 (arm_lshrdi3_1bit): Remove pattern.
7164 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
7165 cost of ashldi3 by 1.
7166 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
7167 (<shift>di3_neon): Likewise.
7169 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7171 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
7172 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
7173 and (*aarch64_simd_mov<VQ:mode>).
7174 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
7175 pattern alternative.
7176 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
7177 attributes to match pattern alternative.
7179 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
7181 * config.gcc (powerpc*-*-*): Add emmintrin.h.
7182 * config/rs6000/emmintrin.h: New file.
7183 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
7185 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
7187 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
7188 * (movdi_vfp_cortexa8): Remove pattern.
7190 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7192 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
7194 (Specific, alpha*-dec-osf5.1): Remove.
7195 (Specific, mips-sgi-irix5): Remove.
7196 (Specific, mips-sgi-irix6): Remove.
7198 2017-10-30 Jakub Jelinek <jakub@redhat.com>
7201 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
7202 arguments to clear_bit_region_be.
7204 2017-10-30 Jim Wilson <wilson@tuliptree.org>
7206 * gimplify.c: Include memmodel.h.
7208 2017-10-30 Martin Jambor <mjambor@suse.cz>
7210 * omp-grid.c (grid_attempt_target_gridification): Also insert a
7211 condition whether loop should be executed at all.
7213 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
7215 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7216 gimple folding of vec_madd() intrinsics.
7217 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
7218 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
7219 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
7221 2017-10-30 Richard Biener <rguenther@suse.de>
7223 PR tree-optimization/82762
7225 2017-10-23 Richard Biener <rguenther@suse.de>
7227 PR tree-optimization/82129
7229 2017-08-01 Richard Biener <rguenther@suse.de>
7231 PR tree-optimization/81181
7232 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7233 (compute_antic): ... end of iteration here.
7235 2017-10-30 Joseph Myers <joseph@codesourcery.com>
7237 * doc/invoke.texi (C Dialect Options): Document -std=c17,
7238 -std=iso9899:2017 and -std=gnu17.
7239 * doc/standards.texi (C Language): Document C17 support.
7240 * doc/cpp.texi (Overview): Mention -std=c17.
7241 (Standard Predefined Macros): Document C11 and C17 values of
7242 __STDC_VERSION__. Do not refer to C99 support as incomplete.
7243 * doc/extend.texi (Inline): Do not list individual options for
7244 standards newer than C99.
7245 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
7247 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
7250 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
7252 * asan.c (asan_finish_file): Align asan globals array by shadow
7255 2017-10-30 Jakub Jelinek <jakub@redhat.com>
7258 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
7259 (struct store_immediate_info): Add bitregion_start and bitregion_end
7261 (store_immediate_info::store_immediate_info): Add brs and bre
7262 arguments and initialize bitregion_{start,end} from those.
7263 (struct merged_store_group): Add bitregion_start, bitregion_end,
7264 align_base and mask fields. Drop unnecessary struct keyword from
7265 struct store_immediate_info. Add do_merge method.
7266 (clear_bit_region_be): Use memset instead of loop storing zeros.
7267 (merged_store_group::do_merge): New method.
7268 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
7269 stores as long as the surrounding bitregions have no gaps.
7270 (merged_store_group::merge_overlapping): Use do_merge.
7271 (merged_store_group::apply_stores): Test that bitregion_{start,end}
7272 is byte aligned, rather than requiring that start and width are
7273 byte aligned. Drop unnecessary struct keyword from
7274 struct store_immediate_info. Allocate and populate also mask array.
7275 Make start of the arrays relative to bitregion_start rather than
7276 start and size them according to bitregion_{end,start} difference.
7277 (struct imm_store_chain_info): Drop unnecessary struct keyword from
7278 struct store_immediate_info.
7279 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
7280 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
7281 struct keyword from struct store_immediate_info.
7282 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
7283 between stores as long as the surrounding bitregions have no gaps.
7285 (struct split_store): Add orig non-static data member.
7286 (split_store::split_store): Initialize orig to false.
7287 (find_constituent_stmts): Return store_immediate_info *, non-NULL
7288 if there is exactly a single original stmt. Change stmts argument
7289 to pointer from reference, if NULL, don't push anything to it. Add
7290 first argument, use it to optimize skipping over orig stmts that
7291 are known to be before bitpos already. Simplify.
7292 (split_group): Return unsigned int count how many stores are or
7293 would be needed rather than a bool. Add allow_unaligned argument.
7294 Change split_stores argument from reference to pointer, if NULL,
7295 only do a dry run computing how many stores would be produced.
7296 Rewritten algorithm to use both alignment and misalign if
7297 !allow_unaligned and handle bitfield stores with gaps.
7298 (imm_store_chain_info::output_merged_store): Set start_byte_pos
7299 from bitregion_start instead of start. Compute allow_unaligned
7300 here, if true, do 2 split_group dry runs to compute which one
7301 produces fewer stores and prefer aligned if equal. Punt if
7302 new count is bigger or equal than original before emitting any
7303 statements, rather than during that. Remove no longer needed
7304 new_ssa_names tracking. Replace num_stmts with
7305 split_stores.length (). Use 32-bit stack allocated entries
7306 in split_stores auto_vec. Try to reuse original store lhs/rhs1
7307 if possible. Handle bitfields with gaps.
7308 (pass_store_merging::execute): Ignore bitsize == 0 stores.
7309 Compute bitregion_{start,end} for the stores and construct
7310 store_immediate_info with that. Formatting fixes.
7312 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
7315 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
7316 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
7318 2017-10-29 Jim Wilson <wilson@tuliptree.org>
7320 * gimplify.c: Include tm_p.h.
7322 * common.opt (gcoff): Re-add as ignored option.
7323 (gcoff1, gcoff2, gcoff3): Likewise.
7325 * Makefile.in (OBJS): Delete sdbout.o.
7326 (GTFILES): Delete $(srcdir)/sdbout.c.
7327 * debug.h: Delete sdb_debug_hooks.
7328 * final.c: Delete sdbout.h include.
7329 (final_scan_insn): Delete SDB_DEBUG check.
7330 (rest_of_clean_state): Likewise.
7331 * output.h: Delete sdb_begin_function_line.
7334 * toplev.c: Delete sdbout.h include.
7335 (process_options): Delete SDB_DEBUG check.
7336 * tree-core.h (tree_type_common): Delete pointer field of
7338 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
7339 TYPE_SYMTAB_POINTER.
7340 * tree.h (TYPE_SYMTAB_POINTER): Delete.
7341 (TYPE_SYMTAB_IS_POINTER): Delete.
7342 (TYPE_SYMTAB_IS_DIE): Renumber.
7343 * xcoffout.c: Refer to former sdbout.c file.
7344 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
7346 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
7347 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
7348 (Debugging Options): Delete -gcoff.
7349 (-gstabs): Delete SDB reference.
7351 (-gcoff@var{level}): Delete.
7352 * doc/passes.texi (Debugging information output): Delete SDB and
7353 sdbout.c references.
7354 * doc/tm.texi: Regenerate.
7355 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
7356 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
7358 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
7359 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
7360 (SDB_DEBUGGING_INFO): Delete.
7361 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
7362 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
7363 * target.def (output_source_filename): Delete COFF reference.
7365 * common.opt (gcoff): Delete.
7366 (gxcoff+): Update Negative chain.
7367 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
7369 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
7370 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
7371 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
7373 (expand_function_start): Change sdb reference to past tense.
7374 (expand_function_end): Change sdb reference to past tense.
7375 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
7376 * opts.c (debug_type_names): Delete coff entry.
7377 (common_handle_option): Delete OPT_gcoff case.
7378 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
7380 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
7381 * config/cris/cris.h: Delete SDB reference in comment.
7382 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
7383 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
7384 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
7385 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
7387 (ix86_expand_prologue): Likewise.
7388 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
7389 * config/ia64/ia64.h: Likewise.
7390 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
7391 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
7393 * config/mmix/mmix.h: Likewise.
7394 * config/nds32/nds32.c: Likewise.
7395 * config/stormy/storym16.h: Likewise.
7396 * config/visium/visium.h: Likewise.
7397 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
7399 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
7401 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
7402 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
7403 FRAME_POINTER_REGNUM point at high end of local var area.
7405 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
7407 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
7408 Move comment around. Do not reset best_edge for a copiable
7409 destination if the copy would cause a partition change.
7410 (better_edge_p): Remove redundant check.
7412 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
7414 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
7416 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
7418 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
7419 for math functions that have _Float<N> and _Float<N>X variants.
7420 (mathfn_built_in_2): Add support for math functions that have
7421 _Float<N> and _Float<N>X variants.
7422 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
7423 (expand_builtin_mathfn_ternary): Add support for fma with
7424 _Float<N> and _Float<N>X variants.
7425 (expand_builtin): Likewise.
7426 (fold_builtin_3): Likewise.
7427 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
7428 create math function _Float<N> and _Float<N>X variants as external
7430 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
7431 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
7432 the __builtin_ prefix and if not strict ansi, without the prefix.
7433 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
7434 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
7435 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
7436 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
7437 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
7438 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
7439 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
7440 function signatures for fma _Float<N> and _Float<N>X variants.
7441 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
7442 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
7443 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
7444 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
7445 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
7446 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
7447 * gencfn-macros.c (print_case_cfn): Add support for math functions
7448 that have _Float<N> and _Float<N>X variants.
7449 (print_define_operator_list): Likewise.
7450 (fltfn_suffixes): Likewise.
7452 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
7453 for math functions that have _Float<N> and _Float<N>X variants.
7454 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
7455 and _Float<N>X variants.
7456 (COPYSIGN): Likewise.
7459 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
7460 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
7462 (integer_valued_read_call_p): Likewise.
7463 * fold-const-call.c (fold_const_call_ss): Likewise.
7464 (fold_const_call_sss): Add support for copysign, fmin, and fmax
7465 _Float<N> and _Float<N>X variants.
7466 (fold_const_call_ssss): Add support for fma _Float<N> and
7467 _Float<N>X variants.
7468 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
7469 support for copysign and fma _Float<N> and _Float<N>X variants.
7470 (backprop::process_builtin_call_use): Likewise.
7471 * tree-call-cdce.c (can_test_argument_range); Add support for
7472 sqrt _Float<N> and _Float<N>X variants.
7473 (edom_only_function): Likewise.
7474 (get_no_error_domain): Likewise.
7475 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
7476 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
7477 copysign _Float<N> and _Float<N>X variants.
7478 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
7479 handled by machine independent code.
7480 (FMAF128): Likewise.
7481 * doc/cpp.texi (Common Predefined Macros): Document defining
7482 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
7483 fma _Float<N> and _Float<N>X variants.
7485 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
7488 * config/i386/i386-modes.def (CCFPU): Remove definition.
7489 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
7490 (ix86_cc_modes_compatible): Ditto.
7491 (ix86_expand_carry_flag_compare): Ditto.
7492 (ix86_expand_int_movcc): Ditto.
7493 (ix86_expand_int_addcc): Ditto.
7494 (ix86_reverse_condition): Ditto.
7495 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
7496 Return true/false for unordered/ordered fp comparisons.
7497 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
7498 (ix86_prepare_fp_compare_args): Update for rename.
7499 (ix86_expand_fp_compare): Update for rename. Generate unordered
7500 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
7501 (ix86_expand_sse_compare_and_jump): Ditto.
7502 * config/i386/predicates.md (fcmov_comparison_operator):
7503 Remove CCFPU mode handling.
7504 (ix86_comparison_operator): Ditto.
7505 (ix86_carry_flag_operator): Ditto.
7506 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
7507 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
7508 (*cmpu<mode>_cc_i387): Ditto.
7509 (FPCMP): Remove mode iterator.
7510 (unord): Remove mode attribute.
7511 (unord_subst): New define_subst transformation
7512 (unord): New define_subst attribute.
7514 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
7515 (*cmpi<unord>xf_i387): Ditto.
7516 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
7517 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
7518 using unord_subst transformation.
7519 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
7520 (round_saeonly): Also handle CCFP mode.
7521 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
7522 Remove UNSPEC_SAHF unspec handling.
7524 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
7526 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
7528 2017-10-27 Jeff Law <law@redhat.com>
7530 * gimple-ssa-sprintf.c: Include domwalk.h.
7531 (class sprintf_dom_walker): New class, derived from dom_walker.
7532 (sprintf_dom_walker::before_dom_children): New function.
7533 (struct call_info): Moved into sprintf_dom_walker class
7534 (compute_formath_length, handle_gimple_call): Likewise.
7535 (sprintf_length::execute): Call the dominator walker rather
7536 than walking the statements.
7538 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
7539 gimple statement locations.
7540 (check_array_bounds): Corresponding changes. Get the statement's
7541 location directly from wi->stmt.
7543 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
7546 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
7548 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
7550 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
7551 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
7553 2017-10-27 Jakub Jelinek <jakub@redhat.com>
7556 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
7557 * config/i386/i386.c (maybe_get_pool_constant): Removed.
7558 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
7559 maybe_get_pool_constant.
7560 * config/i386/predicates.md (zero_extended_scalar_load_operand):
7563 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7565 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
7566 2.26 caveat. Update gas and gld versions.
7567 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
7570 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7572 * cgraph.h (set_malloc_flag): Declare.
7573 * cgraph.c (set_malloc_flag_1): New function.
7574 (set_malloc_flag): Likewise.
7575 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
7576 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
7578 (read_ipa_call_summary): Add support for reading is_return_callee.
7579 (write_ipa_call_summary): Stream is_return_callee.
7580 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
7581 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
7582 ipa-prop.h, ipa-fnsummary.h.
7583 (pure_const_names): Change to static.
7584 (malloc_state_e): Define.
7585 (malloc_state_names): Define.
7586 (funct_state_d): Add field malloc_state.
7587 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
7588 (check_retval_uses): New function.
7589 (malloc_candidate_p): Likewise.
7590 (analyze_function): Add support for malloc attribute.
7591 (pure_const_write_summary): Stream malloc_state.
7592 (pure_const_read_summary): Add support for reading malloc_state.
7593 (dump_malloc_lattice): New function.
7594 (propagate_malloc): New function.
7595 (warn_function_malloc): New function.
7596 (ipa_pure_const::execute): Call propagate_malloc and
7597 ipa_free_fn_summary.
7598 (pass_local_pure_const::execute): Add support for malloc attribute.
7599 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
7600 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
7602 2017-10-27 Martin Liska <mliska@suse.cz>
7604 PR gcov-profile/82457
7605 * doc/invoke.texi: Document that one needs a non-strict ISO mode
7606 for fork-like functions to be properly instrumented.
7608 2017-10-27 Richard Biener <rguenther@suse.de>
7611 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
7612 info when we redirected EH.
7614 2017-10-26 Michael Collison <michael.collison@arm.com>
7616 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
7618 (<optab>_trunchf<GPI:mode>2: New pattern.
7619 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
7620 * config/aarch64/iterators.md (wv): New mode attribute.
7621 (vf, VF): New mode attributes.
7622 (vgp, VGP): New mode attributes.
7623 (s): Update attribute with SImode and DImode prefixes.
7625 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7627 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
7628 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
7629 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
7630 (nios2_option_overide): Initialize it. Don't allow R0-relative
7631 addressing with PIC.
7632 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
7633 (nios2_symbolic_constant_p): Likewise.
7634 (nios2_legitimate_address_p): Likewise.
7635 (nios2_r0rel_section_name_p): New.
7636 (nios2_symbol_ref_in_r0rel_data_p): New.
7637 (nios2_emit_move_sequence): Handle r0rel_constant_p.
7638 (r0rel_constant_p): New.
7639 (nios2_print_operand_address): Handle r0rel_constant_p.
7640 (nios2_cdx_narrow_form_p): Likewise.
7641 * config/nios2/nios2.opt (mr0rel-sec=): New option.
7642 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
7643 (Nios II Options): Document -mr0rel-sec.
7645 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7647 * config/nios2/nios2.c: Include xregex.h.
7648 (nios2_gprel_sec_regex): New.
7649 (nios2_option_overide): Initialize it. Don't allow GP-relative
7650 addressing with PIC.
7651 (nios2_small_section_name_p): Check for regex match.
7652 * config/nios2/nios2.opt (mgprel-sec=): New option.
7653 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
7654 (Nios II Options): Document -mgprel-sec.
7656 2017-10-26 Jim Wilson <wilson@tuliptree.org>
7658 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
7660 2017-10-26 Tom de Vries <tom@codesourcery.com>
7662 PR tree-optimization/82707
7663 * gimple.c (gimple_copy): Fix unsharing of
7664 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
7666 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
7668 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
7669 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
7670 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
7671 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
7672 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
7673 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
7674 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
7675 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
7676 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
7677 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
7678 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
7679 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
7680 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
7681 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
7682 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
7683 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
7684 _mm512_mask_cmpunord_ps_mask): New intrinsics.
7686 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7688 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
7690 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
7691 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
7692 warning to rs6000.c. Remove the Undocumented flag, since it has
7694 (-mabi=ibmlongdouble): Likewise.
7695 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
7696 already set, set the default format for long double.
7697 (rs6000_debug_reg_global): Print whether long double is IBM or
7699 (rs6000_option_override_internal): Rework setting long double
7700 format. Only warn if the user is changing the long double default
7701 and they did not use -Wno-psabi.
7702 * doc/invoke.texi (PowerPC options): Update the documentation for
7703 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
7705 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7706 Alan Hayward <alan.hayward@arm.com>
7707 David Sherwood <david.sherwood@arm.com>
7709 * rtl.h (wider_subreg_mode): New function.
7710 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
7711 rather than an unsigned int *.
7712 * ira-color.c (regno_max_ref_width): Replace with...
7713 (regno_max_ref_mode): ...this new variable.
7714 (coalesced_pseudo_reg_slot_compare): Update accordingly.
7715 Use wider_subreg_mode.
7716 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
7717 rather than an unsigned int *.
7718 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
7719 (process_alt_operands): Likewise.
7720 (invariant_p): Likewise.
7721 * lra-spills.c (assign_mem_slot): Likewise.
7722 (add_pseudo_to_slot): Likewise.
7723 * lra.c (collect_non_operand_hard_regs): Likewise.
7724 (add_regs_to_insn_regno_info): Likewise.
7725 * reload1.c (regno_max_ref_width): Replace with...
7726 (regno_max_ref_mode): ...this new variable.
7727 (reload): Update accordingly. Update call to
7728 ira_sort_regnos_for_alter_reg.
7729 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
7730 (init_eliminable_invariants): Update to use regno_max_ref_mode.
7731 (scan_paradoxical_subregs): Likewise.
7733 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7735 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
7736 (aarch64_frame): Add emit_frame_chain boolean.
7737 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7738 Move eh_return case to aarch64_layout_frame.
7739 (aarch64_layout_frame): Initialize emit_frame_chain.
7740 (aarch64_expand_prologue): Use emit_frame_chain.
7742 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7744 * config/aarch64/aarch64.c (aarch64_layout_frame):
7745 Ensure LR is always stored at the bottom of the callee-saves.
7746 Remove rarely used frame layout which saves callee-saves at top of
7747 frame, so the store of LR can be used as a valid probe in all cases.
7749 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7751 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
7752 Improve unaligned TImode/TFmode base/offset split.
7754 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7755 Alan Hayward <alan.hayward@arm.com>
7756 David Sherwood <david.sherwood@arm.com>
7758 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
7759 * combine.c (find_single_use_1): Likewise.
7760 (expand_field_assignment): Likewise.
7761 (move_deaths): Likewise.
7762 * lra-constraints.c (simplify_operand_subreg): Likewise.
7763 (curr_insn_transform): Likewise.
7764 * lra.c (collect_non_operand_hard_regs): Likewise.
7765 (add_regs_to_insn_regno_info): Likewise.
7766 * rtlanal.c (reg_referenced_p): Likewise.
7767 (covers_regno_no_parallel_p): Likewise.
7769 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7771 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
7772 Don't print any bits outside the precision of the value.
7773 * wide-int.cc (test_printing): Add some new tests.
7775 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7777 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
7778 supports -xbrace_comment option.
7779 * configure: Regenerate.
7780 * config.in: Regenerate.
7781 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
7782 (ASM_CPU_SPEC): Use it.
7784 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7786 * target.def (static_rtx_alignment): New hook.
7787 * targhooks.h (default_static_rtx_alignment): Declare.
7788 * targhooks.c (default_static_rtx_alignment): New function.
7789 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
7790 * doc/tm.texi: Regenerate.
7791 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
7792 instead of targetm.constant_alignment. Remove call to
7794 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7795 (cris_preferred_mininum_alignment): New function, split out from...
7796 (cris_constant_alignment): ...here.
7797 (cris_static_rtx_alignment): New function.
7798 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
7800 (ix86_constant_alignment): ...here.
7801 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7802 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7803 (mmix_static_rtx_alignment): New function.
7804 * config/spu/spu.c (spu_static_rtx_alignment): New function.
7805 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7807 2017-10-26 Tamar Christina <tamar.christina@arm.com>
7810 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
7811 Add flag_trapping_math and flag_fp_int_builtin_inexact.
7813 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
7815 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
7816 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
7817 mark as a sign-extending load.
7818 (local_pic_load_u): Define.
7820 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
7823 * fold-const.c (operand_equal_for_comparison_p): Also return true
7824 if ARG0 is a simple variant of ARG1 with narrower precision.
7825 (fold_ternary_loc): Always pass unstripped operands to the predicate.
7827 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
7829 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
7831 * i386.h (processor_costs): Add gather_static, gather_per_elt,
7832 scatter_static, scatter_per_elt.
7833 * x86-tune-costs.h: Add new cost entries.
7835 2017-10-25 Richard Biener <rguenther@suse.de>
7837 * tree-ssa-sccvn.h (vn_eliminate): Declare.
7838 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
7839 class pass_fre): Move to ...
7840 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
7841 class pass_fre): ... here and adjust for statistics.
7843 2017-10-25 Jakub Jelinek <jakub@redhat.com>
7846 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
7847 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
7848 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
7849 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
7852 2017-10-25 Richard Biener <rguenther@suse.de>
7854 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
7855 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
7856 eliminate_push_avail, eliminate_insert): Move inside...
7857 (class eliminate_dom_walker): ... this class in preparation
7859 (fini_eliminate): Remove by merging with ...
7860 (eliminate): ... this function. Adjust for class changes.
7861 (pass_pre::execute): Remove fini_eliminate call.
7862 (pass_fre::execute): Likewise.
7864 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7867 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
7868 (VPERMI2, VPERMI2I): New mode iterators.
7869 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
7870 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
7872 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
7873 mode iterator. Remove 3 old define_insn patterns.
7874 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
7875 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
7876 VPERMI2 mode iterator, remove the other two expanders.
7877 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
7878 to use VPERMI2 mode iterator, add another alternative for vpermi2*
7879 instructions, remove the other two patterns.
7880 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
7881 mode iterator, remove the other two patterns.
7882 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
7883 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
7884 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
7885 and adjust argument order accordingly.
7886 (ix86_expand_vec_perm): Adjust caller.
7887 (expand_vec_perm_1): Likewise.
7888 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
7889 (expand_vec_perm_vpermt2_vpshub2): ... this.
7890 (ix86_expand_vec_perm_const_1): Adjust caller.
7891 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
7894 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
7895 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
7896 (vec_shl_<mode>): Remove unused expander.
7897 (avx512bw_<shift_insn><mode>3): New define_insn.
7898 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
7899 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
7901 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
7904 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
7907 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7909 PR rtl-optimization/82396
7910 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
7911 (autopref_multipass_init): Simplify initialization.
7912 (autopref_rank_data): Simplify sort order.
7913 * gcc/sched-int.h (autopref_multipass_data_): Remove
7914 multi_mem_insn_p, min_offset and max_offset.
7916 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7919 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7920 Check special value of flag_omit_frame_pointer.
7921 (aarch64_can_eliminate): Likewise.
7922 (aarch64_override_options_after_change_1): Simplify handling of
7923 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
7925 2017-10-24 Richard Biener <rguenther@suse.de>
7927 PR tree-optimization/82697
7928 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
7929 zero for conditional load and unconditional store.
7931 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7933 * doc/install.texi: Document bootstrap-cet.
7935 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7938 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
7939 ENDBR instruction at function entrance if function is only
7942 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7945 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
7946 patterns to better describe from which operation the CF is computed.
7947 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
7948 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
7949 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
7950 is 0, use _0 suffixed expanders instead of emitting a comparison
7953 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
7955 * config/i386/i386.md(*movsf_internal, *movdf_internal):
7956 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
7958 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
7961 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
7962 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
7963 * loop-iv.c (iv_get_reaching_def): Likewise.
7964 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
7965 variable is promoted and the partition contains undefined values.
7967 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7969 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
7971 (nios2_address_cost): Define.
7972 (nios2_legitimize_address): Recognize (exp + constant) directly.
7973 (TARGET_ADDRESS_COST): Define.
7975 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7977 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
7978 (nios2_symbolic_memory_operand_p): Declare.
7979 (nios2_split_large_constant): Declare.
7980 (nios2_split_symbolic_memory_operand): Declare.
7981 * config/nios2/nios2.c: Adjust includes.
7982 (nios2_symbolic_constant_allowed): New.
7983 (nios2_symbolic_constant_p): New.
7984 (nios2_plus_symbolic_constant_p): New.
7985 (nios2_valid_addr_expr_p): Recognize addresses involving
7987 (nios2_legitimate_address_p): Likewise, also LO_SUM.
7988 (nios2_symbolic_memory_operand_p): New.
7989 (nios2_large_constant_p): New.
7990 (nios2_split_large_constant): New.
7991 (nios2_split_plus_large_constant): New.
7992 (nios2_split_symbolic_memory_operand): New.
7993 (nios2_legitimize_address): Code refactoring. Handle addresses
7994 involving symbolic constants.
7995 (nios2_emit_move_sequence): Likewise.
7996 (nios2_print_operand): Improve error output.
7997 (nios2_print_operand_address): Handle LO_SUM.
7998 (nios2_cdx_narrow_form_p): Likewise.
7999 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
8000 operands involving symbolic constants.
8001 (movhi_internal, movsi_internal): Likewise.
8002 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
8003 (extendhisi2, extendqi<mode>2): Likewise.
8005 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8007 * tree-pass.h (PROP_rtl_split_insns): Define.
8008 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
8010 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8012 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
8014 2017-10-23 Jakub Jelinek <jakub@redhat.com>
8017 * target.def (const_not_ok_for_debug_p): Default to
8018 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
8019 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
8020 * targhooks.c (default_const_not_ok_for_debug_p): New function.
8021 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
8022 which targetm.const_not_ok_for_debug_p returned true.
8023 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
8025 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
8027 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
8028 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
8029 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
8031 (ix86_const_not_ok_for_debug_p): New function.
8032 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
8033 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
8035 2017-10-23 David Malcolm <dmalcolm@redhat.com>
8038 * system.h: Conditionally include "unique-ptr.h" if
8039 INCLUDE_UNIQUE_PTR is defined.
8040 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
8041 of defining INCLUDE_UNIQUE_PTR before including "system.h".
8043 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
8045 * config/rl78/rl78.md: New define_expand "subdi3".
8047 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
8050 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
8051 DF_REF_INSN if DF_REF_INSN_INFO is false.
8053 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
8055 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
8056 xmm_move instead of sse_move.
8057 (sse_store_index): New function.
8058 (ix86_register_move_cost): Be more sensible about mismatch stall;
8059 model AVX moves correctly; make difference between sse->integer and
8061 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
8062 moves; make difference between SSE and AVX.
8063 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
8064 and zmm_move. Increase size of sse load and store tables;
8065 add unaligned load and store tables; add ssemmx_to_integer.
8066 * x86-tune-costs.h: Update all entries according to real
8067 move latencies from Agner Fog's manual and chip documentation.
8069 2017-10-23 Jakub Jelinek <jakub@redhat.com>
8072 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
8073 * config/i386/constraints.md (Wf): New constraint.
8074 * config/i386/i386.md (UNSPEC_SBB): New unspec.
8075 (cmp<dwi>_doubleword): Removed.
8076 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
8077 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
8078 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
8079 expand with cmp<dwi>_doubleword. For LTU and GEU use
8080 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
8082 * common.opt (gcolumn-info): Enable by default.
8083 * doc/invoke.texi (gcolumn-info): Document new default.
8085 2017-10-23 Richard Biener <rguenther@suse.de>
8087 PR tree-optimization/82672
8088 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
8089 Fold the stmt if we propagated into it.
8091 2017-10-23 Richard Biener <rguenther@suse.de>
8093 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
8094 (bitmap_remove_expr_from_set): ... this. All callers call this
8095 for non-constant values.
8096 (bitmap_set_subtract): Rename to...
8097 (bitmap_set_subtract_expressions): ... this. Adjust and
8099 (bitmap_set_contains_value): Remove superfluous check.
8100 (bitmap_set_replace_value): Inline into single caller ...
8101 (bitmap_value_replace_in_set): ... here and simplify.
8102 (dependent_clean): Merge into ...
8103 (clean): ... this using an overload. Adjust.
8104 (prune_clobbered_mems): Adjust.
8105 (compute_antic_aux): Likewise.
8106 (compute_partial_antic_aux): Likewise.
8108 2017-10-23 Richard Biener <rguenther@suse.de>
8110 PR tree-optimization/82129
8112 2017-08-01 Richard Biener <rguenther@suse.de>
8114 PR tree-optimization/81181
8115 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8116 (compute_antic): ... end of iteration here.
8118 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8120 * target.def (starting_frame_offset): New hook.
8121 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
8122 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
8123 * doc/tm.texi.in: Regenerate.
8124 * hooks.h (hook_hwi_void_0): Declare.
8125 * hooks.c (hook_hwi_void_0): New function.
8126 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
8127 STARTING_FRAME_OFFSET.
8128 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
8129 * reload1.c (reload): Likewise.
8130 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
8131 instead of STARTING_FRAME_OFFSET.
8132 * function.c (try_fit_stack_local): Likewise.
8133 (assign_stack_local_1): Likewise
8134 (instantiate_virtual_regs): Likewise.
8135 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
8136 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
8137 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
8138 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
8139 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
8140 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
8141 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
8142 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
8143 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
8144 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
8145 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
8146 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
8147 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
8148 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
8149 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
8150 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
8151 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
8152 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
8153 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
8154 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
8155 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
8156 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
8157 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
8158 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
8159 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
8160 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
8161 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
8162 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
8163 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
8164 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
8165 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
8166 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
8167 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
8168 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
8169 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
8170 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
8171 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
8172 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
8173 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
8174 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
8175 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
8176 * config/avr/avr.c (avr_starting_frame_offset): Make static and
8177 return a HOST_WIDE_INT.
8178 (avr_builtin_setjmp_frame_value): Use it instead of
8179 STARTING_FRAME_OFFSET.
8180 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8181 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
8182 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
8184 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8185 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
8186 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
8187 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8188 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
8189 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
8190 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8191 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
8192 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
8193 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8194 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
8195 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
8197 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8198 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
8199 * config/mips/mips.c (mips_compute_frame_info): Refer to
8200 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
8201 (mips_starting_frame_offset): New function.
8202 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8203 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
8204 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
8205 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
8206 and return a HOST_WIDE_INT.
8207 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8208 (mmix_initial_elimination_offset): Refer to
8209 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
8210 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
8211 * config/pa/pa.c (pa_starting_frame_offset): New function.
8212 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
8213 (pa_expand_prologue): Likewise.
8214 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8215 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
8216 !FRAME_GROWS_DOWNWARD handling to...
8217 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8218 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
8219 !FRAME_GROWS_DOWNWARD handling to...
8220 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8221 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
8222 !FRAME_GROWS_DOWNWARD handling to...
8223 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8224 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
8226 (rs6000_starting_frame_offset): New function.
8227 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
8228 !FRAME_GROWS_DOWNWARD handling to...
8229 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8230 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
8231 !FRAME_GROWS_DOWNWARD handling to...
8232 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8233 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
8234 !FRAME_GROWS_DOWNWARD handling to...
8235 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8236 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
8237 (rs6000_starting_frame_offset): New function.
8238 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
8239 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
8240 * config/vax/vax.c (vax_starting_frame_offset): New function.
8241 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
8242 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8243 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
8244 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
8245 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8246 * system.h (STARTING_FRAME_OFFSET): Poison.
8248 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8250 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
8251 SCALAR_TYPE_MODE instead of TYPE_MODE.
8253 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8254 Alan Hayward <alan.hayward@arm.com>
8255 David Sherwood <david.sherwood@arm.com>
8257 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
8259 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8260 Alan Hayward <alan.hayward@arm.com>
8261 David Sherwood <david.sherwood@arm.com>
8263 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
8265 2017-10-23 Richard Biener <rguenther@suse.de>
8267 PR tree-optimization/82129
8268 * tree-ssa-pre.c (bitmap_set_and): Remove.
8269 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
8270 canonicalizing expressions in the set to those with lowest
8271 ID rather than taking that from the first edge.
8273 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8275 * combine.c (rtx_equal_for_field_assignment_p): Use
8276 byte_lowpart_offset.
8278 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8279 Alan Hayward <alan.hayward@arm.com>
8280 David Sherwood <david.sherwood@arm.com>
8282 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
8283 to a SUBREG_PROMOTED_VAR.
8285 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8286 Alan Hayward <alan.hayward@arm.com>
8287 David Sherwood <david.sherwood@arm.com>
8289 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
8290 (expand_debug_source_expr): Likewise.
8291 * combine.c (combine_simplify_rtx): Likewise.
8292 * cse.c (fold_rtx): Likewise.
8293 * optabs.c (expand_float): Likewise.
8294 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8295 (simplify_binary_operation_1): Likewise.
8297 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8298 Alan Hayward <alan.hayward@arm.com>
8299 David Sherwood <david.sherwood@arm.com>
8301 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
8302 (record_promoted_value): Likewise.
8303 * expr.c (expand_expr_real_2): Likewise.
8304 * ree.c (update_reg_equal_equiv_notes): Likewise.
8305 (combine_set_extension): Likewise.
8306 * rtlanal.c (low_bitmask_len): Likewise.
8307 * simplify-rtx.c (neg_const_int): Likewise.
8308 (simplify_binary_operation_1): Likewise.
8310 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8311 Alan Hayward <alan.hayward@arm.com>
8312 David Sherwood <david.sherwood@arm.com>
8314 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
8315 * regcprop.c (maybe_mode_change): Likewise.
8316 * reload1.c (alter_reg): Likewise.
8318 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8320 * inchash.h (inchash::hash::add_wide_int): New function.
8321 * lto-streamer-out.c (hash_tree): Use it.
8323 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8325 * inchash.h (inchash::hash::add_wide_int): Rename to...
8326 (inchash::hash::add_hwi): ...this.
8327 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
8328 (polymorphic_call_target_hasher::hash): Likewise.
8329 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
8330 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
8331 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
8332 * lto-streamer-out.c (hash_tree): Likewise.
8333 * optc-save-gen.awk: Likewise.
8334 * tree.c (add_expr): Likewise.
8336 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
8339 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
8340 for ordered inequality comparisons even with TARGET_IEEE_FP.
8342 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
8345 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
8346 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
8347 Expand with cmp<dwi>_doubleword.
8349 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8351 * extend.texi: Add x86 specific to 'nocf_check' attribute.
8352 List CET intrinsics.
8353 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
8354 specific to -fcf-protection option.
8356 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8358 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
8359 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
8360 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
8361 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
8362 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
8363 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
8364 (extra_objs): Add cet.o for Linux/x86 targets.
8365 (tmake_file): Add i386/t-cet for Linux/x86 targets.
8366 * config/i386/cet.c: New file.
8367 * config/i386/cetintrin.h: Likewise.
8368 * config/i386/t-cet: Likewise.
8369 * config/i386/cpuid.h (bit_SHSTK): New.
8370 (bit_IBT): Likewise.
8371 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
8372 pass IBT and SHSTK bits.
8373 * config/i386/i386-builtin-types.def
8374 (VOID_FTYPE_UNSIGNED_PVOID): New.
8375 (VOID_FTYPE_UINT64_PVOID): Likewise.
8376 * config/i386/i386-builtin.def: Add CET intrinsics.
8377 * config/i386/i386-c.c (ix86_target_macros_internal): Add
8378 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
8379 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
8380 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
8382 * config/i386/i386.c (rest_of_insert_endbranch): New.
8383 (pass_data_insert_endbranch): Likewise.
8384 (pass_insert_endbranch): Likewise.
8385 (make_pass_insert_endbranch): Likewise.
8386 (ix86_notrack_prefixed_insn_p): Likewise.
8387 (ix86_target_string): Add -mibt, -mshstk flags.
8388 (ix86_option_override_internal): Add flag_cf_protection
8390 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
8391 (ix86_print_operand): Add 'notrack' prefix output.
8392 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
8393 (ix86_expand_builtin): Expand CET intrinsics.
8394 (x86_output_mi_thunk): Add 'endbranch' instruction.
8395 * config/i386/i386.h (TARGET_IBT): New.
8396 (TARGET_IBT_P): Likewise.
8397 (TARGET_SHSTK): Likewise.
8398 (TARGET_SHSTK_P): Likewise.
8399 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
8400 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
8401 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
8402 (builtin_setjmp_setup): New pattern.
8403 (builtin_longjmp): Likewise.
8404 (rdssp<mode>): Likewise.
8405 (incssp<mode>): Likewise.
8406 (saveprevssp): Likewise.
8407 (rstorssp): Likewise.
8408 (wrss<mode>): Likewise.
8409 (wruss<mode>): Likewise.
8410 (setssbsy): Likewise.
8411 (clrssbsy): Likewise.
8412 (nop_endbr): Likewise.
8413 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
8415 * config/i386/immintrin.h: Include <cetintrin.h>.
8416 * config/i386/linux-common.h
8417 (file_end_indicate_exec_stack_and_cet): New prototype.
8418 (TARGET_ASM_FILE_END): New.
8420 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
8422 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
8423 latencies instead of having separate table; make difference between
8424 integer and float costs.
8425 * i386.h (processor_costs): Remove scalar_stmt_cost,
8426 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
8427 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
8429 * x86-tune-costs.h: Remove entries which has been removed in
8430 procesor_costs from all tables; make cond_taken_branch_cost
8431 and cond_not_taken_branch_cost COST_N_INSNS based.
8433 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
8435 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
8437 2017-10-20 Jakub Jelinek <jakub@redhat.com>
8439 * config/i386/i386.md (isa): Remove fma_avx512f.
8440 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
8441 <avx512>_fmadd_<mode>_mask3<round_name>,
8442 <avx512>_fmsub_<mode>_mask<round_name>,
8443 <avx512>_fmsub_<mode>_mask3<round_name>,
8444 <avx512>_fnmadd_<mode>_mask<round_name>,
8445 <avx512>_fnmadd_<mode>_mask3<round_name>,
8446 <avx512>_fnmsub_<mode>_mask<round_name>,
8447 <avx512>_fnmsub_<mode>_mask3<round_name>,
8448 <avx512>_fmaddsub_<mode>_mask<round_name>,
8449 <avx512>_fmaddsub_<mode>_mask3<round_name>,
8450 <avx512>_fmsubadd_<mode>_mask<round_name>,
8451 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
8452 (*vec_widen_umult_even_v16si<mask_name>,
8453 *vec_widen_smult_even_v16si<mask_name>): Likewise.
8454 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
8456 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8458 * extend.texi: Add 'nocf_check' documentation.
8459 * gimple.texi: Add second parameter to
8460 gimple_build_call_from_tree.
8461 * invoke.texi: Add -fcf-protection documentation.
8462 * rtl.texi: Add REG_CALL_NOTRACK documenation.
8464 2017-10-20 Richard Biener <rguenther@suse.de>
8466 PR tree-optimization/82473
8467 * tree-vect-loop.c (vectorizable_reduction): Properly get at
8468 the largest input type.
8470 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8472 * c-attribs.c (handle_nocf_check_attribute): New function.
8473 (c_common_attribute_table): Add 'nocf_check' handling.
8474 * gimple-parser.c: Add second argument NULL to
8475 gimple_build_call_from_tree.
8476 * attrib.c (comp_type_attributes): Check nocf_check attribute.
8477 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
8479 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
8480 * common.opt: Add fcf-protection flag.
8481 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
8482 * flag-types.h: Add enum cf_protection_level.
8483 * gimple.c (gimple_build_call_from_tree): Add second parameter.
8484 Add 'nocf_check' attribute propagation to gimple call.
8485 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
8486 (gimple_build_call_from_tree): Update prototype.
8487 (gimple_call_nocf_check_p): New function.
8488 (gimple_call_set_nocf_check): Likewise.
8489 * gimplify.c: Add second argument to gimple_build_call_from_tree.
8490 * ipa-icf.c: Add nocf_check attribute in statement hash.
8491 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
8492 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
8493 * toplev.c (process_options): Add flag_cf_protection handling.
8495 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8497 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
8499 2017-10-20 Richard Biener <rguenther@suse.de>
8501 PR tree-optimization/82603
8502 * tree-if-conv.c (predicate_mem_writes): Make sure to only
8503 remove false predicated stores.
8505 2017-10-20 Richard Biener <rguenther@suse.de>
8507 * graphite-isl-ast-to-gimple.c
8508 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
8509 Remove return value and simplify, dump copied stmt after lhs
8511 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
8512 Reduce dump verbosity.
8513 (gsi_insert_earliest): Likewise.
8514 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
8515 * graphite.c (print_global_statistics): Adjust dumping.
8516 (print_graphite_scop_statistics): Likewise.
8517 (print_graphite_statistics): Do not dump loops here.
8518 (graphite_transform_loops): But here.
8520 2017-10-20 Nicolas Roche <roche@adacore.com>
8522 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
8523 * configure: Regenerate.
8525 2017-10-20 Jakub Jelinek <jakub@redhat.com>
8528 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
8529 functions when optimizing replace GIMPLE_RETURN stmts with
8530 calls to __builtin_unreachable ().
8533 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
8534 for -fsanitize=thread link of executables.
8535 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
8536 link of executables.
8539 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
8541 (<shift_insn><mode>3<mask_name>): Change the last of the 3
8542 define_insns for logical vector shifts to use VI248_AVX512BW
8543 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
8544 condition, useless isa and prefix attributes. Change the first
8545 2 of these define_insns to ...
8546 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
8547 define_insn for avx512vl.
8548 (<shift_insn><mode>3): ... and this, new define_insn without
8549 masking for non-avx512vl.
8552 * config/i386/sse.md (*andnot<mode>3,
8553 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
8554 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
8555 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
8556 not applied use empty suffix even for TARGET_AVX512VL.
8557 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
8558 is applied, supply evex,evex or evex,evex,evex instead of just
8561 2017-10-20 Julia Koval <julia.koval@intel.com>
8563 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
8564 (OPTION_MASK_ISA_GFNI_UNSET): New.
8565 (ix86_handle_option): Handle OPT_mgfni.
8566 * config/i386/cpuid.h (bit_GFNI): New.
8567 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
8568 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
8569 * config/i386/i386.c (ix86_target_string): Add -mgfni.
8570 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
8571 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
8572 * config/i386/i386.opt: Add mgfni.
8574 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
8576 * config/msp430/msp430.c (msp430_option_override): Disable
8577 -fdelete-null-pointer-checks.
8578 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
8580 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8582 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
8583 of x87 and SSE instructions.
8585 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8587 * asan.c (create_cond_insert_point): Do not update edge count.
8588 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
8589 (afdo_propagate_circuit): Likewise.
8590 (afdo_calculate_branch_prob): Likewise.
8591 (afdo_annotate_cfg): Likewise.
8592 * basic-block.h (struct edge_def): Remove count.
8593 (edge_def::count): New accessor.
8594 * bb-reorder.c (rotate_loop): Update.
8595 (find_traces_1_round): Update.
8596 (connect_traces): Update.
8597 (sanitize_hot_paths): Update.
8598 * cfg.c (unchecked_make_edge): Update.
8599 (make_single_succ_edge): Update.
8600 (check_bb_profile): Update.
8601 (dump_edge_info): Update.
8602 (update_bb_profile_for_threading): Update.
8603 (scale_bbs_frequencies_int): Update.
8604 (scale_bbs_frequencies_gcov_type): Update.
8605 (scale_bbs_frequencies_profile_count): Update.
8606 (scale_bbs_frequencies): Update.
8607 * cfganal.c (connect_infinite_loops_to_exit): Update.
8608 * cfgbuild.c (compute_outgoing_frequencies): Update.
8609 (find_many_sub_basic_blocks): Update.
8610 * cfgcleanup.c (try_forward_edges): Update.
8611 (try_crossjump_to_edge): Update
8612 * cfgexpand.c (expand_gimple_cond): Update
8613 (expand_gimple_tailcall): Update
8614 (construct_exit_block): Update
8615 * cfghooks.c (verify_flow_info): Update
8616 (redirect_edge_succ_nodup): Update
8617 (split_edge): Update
8618 (make_forwarder_block): Update
8619 (duplicate_block): Update
8620 (account_profile_record): Update
8621 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
8622 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
8623 * cfgloopmanip.c (scale_loop_profile): Update.
8625 (lv_adjust_loop_entry_edge): Update.
8626 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
8627 (force_nonfallthru_and_redirect): Update.
8628 (purge_dead_edges): Update.
8629 (rtl_flow_call_edges_add): Update.
8630 * cgraphunit.c (init_lowered_empty_function): Update.
8631 (cgraph_node::expand_thunk): Update.
8632 * gimple-pretty-print.c (dump_probability): Update.
8633 (dump_edge_probability): Update.
8634 * gimple-ssa-isolate-paths.c (isolate_path): Update.
8635 * haifa-sched.c (sched_create_recovery_edges): Update.
8636 * hsa-gen.c (convert_switch_statements): Update.
8637 * ifcvt.c (dead_or_predicable): Update.
8638 * ipa-inline-transform.c (inline_transform): Update.
8639 * ipa-split.c (split_function): Update.
8640 * ipa-utils.c (ipa_merge_profiles): Update.
8641 * loop-doloop.c (add_test): Update.
8642 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
8643 * lto-streamer-in.c (input_cfg): Update.
8644 (input_function): Update.
8645 * lto-streamer-out.c (output_cfg): Update.
8646 * modulo-sched.c (sms_schedule): Update.
8647 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
8648 * predict.c (maybe_hot_edge_p): Update.
8649 (unlikely_executed_edge_p): Update.
8650 (probably_never_executed_edge_p): Update.
8651 (dump_prediction): Update.
8652 (drop_profile): Update.
8653 (propagate_unlikely_bbs_forward): Update.
8654 (determine_unlikely_bbs): Update.
8655 (force_edge_cold): Update.
8656 * profile.c (compute_branch_probabilities): Update.
8657 * reg-stack.c (better_edge): Update.
8658 * shrink-wrap.c (handle_simple_exit): Update.
8659 * tracer.c (better_p): Update.
8660 * trans-mem.c (expand_transaction): Update.
8661 (split_bb_make_tm_edge): Update.
8662 * tree-call-cdce.c: Update.
8663 * tree-cfg.c (gimple_find_sub_bbs): Update.
8664 (gimple_split_edge): Update.
8665 (gimple_duplicate_sese_region): Update.
8666 (gimple_duplicate_sese_tail): Update.
8667 (gimple_flow_call_edges_add): Update.
8668 (insert_cond_bb): Update.
8669 (execute_fixup_cfg): Update.
8670 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
8671 * tree-complex.c (expand_complex_div_wide): Update.
8672 * tree-eh.c (lower_resx): Update.
8673 (unsplit_eh): Update.
8674 (cleanup_empty_eh_move_lp): Update.
8675 * tree-inline.c (copy_edges_for_bb): Update.
8676 (freqs_to_counts): Update.
8677 (copy_cfg_body): Update.
8678 * tree-ssa-dce.c (remove_dead_stmt): Update.
8679 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
8680 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
8681 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
8682 (unloop_loops): Update.
8683 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8684 * tree-ssa-loop-split.c (connect_loops): Update.
8685 (split_loop): Update.
8686 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
8687 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
8688 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
8689 * tree-ssa-reassoc.c (branch_fixup): Update.
8690 * tree-ssa-tail-merge.c (replace_block_by): Update.
8691 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
8692 (compute_path_counts): Update.
8693 (update_profile): Update.
8694 (recompute_probabilities): Update.
8695 (update_joiner_offpath_counts): Update.
8696 (estimated_freqs_path): Update.
8697 (freqs_to_counts_path): Update.
8698 (clear_counts_path): Update.
8699 (ssa_fix_duplicate_block_edges): Update.
8700 (duplicate_thread_path): Update.
8701 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
8702 (case_bit_test_cmp): Update.
8703 (collect_switch_conv_info): Update.
8704 (gen_inbound_check): Update.
8705 (do_jump_if_equal): Update.
8706 (emit_cmp_and_jump_insns): Update.
8707 * tree-tailcall.c (decrease_profile): Update.
8708 (eliminate_tail_call): Update.
8709 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
8710 (vect_do_peeling): Update.
8711 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8712 * ubsan.c (ubsan_expand_null_ifn): Update.
8713 (ubsan_expand_ptr_ifn): Update.
8714 * value-prof.c (gimple_divmod_fixed_value): Update.
8715 (gimple_mod_pow2): Update.
8716 (gimple_mod_subtract): Update.
8717 (gimple_ic): Update.
8718 (gimple_stringop_fixed_value): Update.
8720 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8723 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
8725 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
8727 PR rtl-optimization/82395
8728 * ira-color.c (allocno_priority_compare_func): Fix comparison step
8729 based on non_spilled_static_chain_regno_p.
8731 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8733 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
8734 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
8735 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
8737 2017-10-19 Martin Sebor <msebor@redhat.com>
8739 PR tree-optimization/82596
8740 * tree.c (array_at_struct_end_p): Handle STRING_CST.
8742 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8744 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
8745 (get_mem_refs_of_builtin_call): Likewise.
8746 * builtins.c (expand_builtin_apply): Adjust call to
8747 allocate_dynamic_stack_space.
8748 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
8749 the third argument to allocate_dynamic_stack_space, otherwise -1.
8750 (expand_builtin): Deal with all alloca variants.
8751 (is_inexpensive_builtin): Likewise.
8752 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
8753 * calls.c (special_function_p): Deal with all alloca variants.
8754 (initialize_argument_information): Adjust call to
8755 allocate_dynamic_stack_space.
8756 (expand_call): Likewise.
8757 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
8758 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
8759 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
8760 use it for the stack usage computation.
8761 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
8762 * function.c (gimplify_parameters): Call build_alloca_call_expr.
8763 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
8764 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
8765 (in_loop_p): Remove first argument and useless check.
8766 (pass_walloca::execute): Remove useless test and adjust call to above.
8767 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
8768 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
8769 (gimplify_call_expr): Deal with all alloca variants.
8770 * hsa-gen.c (gen_hsa_alloca): Likewise.
8771 (gen_hsa_insns_for_call): Likewise.
8772 * ipa-pure-const.c (special_builtin_state): Likewise.
8773 * tree-chkp.c (chkp_build_returned_bound): Likewise.
8774 * tree-object-size.c (alloc_object_size): Likewise.
8775 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
8776 (call_may_clobber_ref_p_1): Likewise.
8777 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
8778 (ccp_fold_stmt): Likewise.
8779 (optimize_stack_restore): Likewise.
8780 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
8781 (mark_all_reaching_defs_necessary_1): Likewise.
8782 (propagate_necessity): Likewise.
8783 (eliminate_unnecessary_stmts): Likewise.
8784 * tree.c (build_common_builtin_nodes): Build
8785 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
8786 (build_alloca_call_expr): New function.
8787 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
8788 (CASE_BUILT_IN_ALLOCA): Likewise.
8789 (build_alloca_call_expr): Declare.
8790 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
8792 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8795 * dwarf2out.c (new_die_raw): New static inline function.
8796 (new_die): Use it to create the DIE.
8797 (add_AT_external_die_ref): Likewise.
8798 (clone_die): Likewise.
8799 (clone_as_declaration): Likewise.
8800 (dwarf2out_vms_debug_main_pointer): Likewise.
8801 (base_type_die): Likewise. Remove early return for corner cases.
8802 Do not call add_pubtype on the DIE here.
8803 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
8804 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
8805 typedefs for base types with DW_AT_endianity. Make sure a DIE with
8806 native order exists for base types, attach the DIE manually and call
8807 add_pubtype on it. Do not equate a reverse order DIE to the type.
8809 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
8811 * config/arm/arm.c (align_ok_ldrd_strd): New function.
8812 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
8814 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
8816 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8818 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
8819 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
8820 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
8821 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
8822 * opts.c (sanitizer_opts): Add builtin.
8823 * ubsan.c (instrument_builtin): New function.
8824 (pass_ubsan::execute): Call it.
8825 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
8826 * doc/invoke.texi: Document -fsanitize=builtin.
8828 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
8829 builtins, store max (log2 (align), 0) into uchar field instead of
8830 align into uptr field.
8831 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
8832 store uchar 0 field instead of uptr 0 field.
8833 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
8834 instead of passing one address of struct with 2 locations pass
8835 two addresses of structs with 1 location each.
8836 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
8837 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
8838 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
8839 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
8840 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
8841 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
8842 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
8843 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
8845 2017-10-19 Martin Liska <mliska@suse.cz>
8848 * file-find.c (remove_prefix): Remove.
8849 * file-find.h (remove_prefix): Likewise.
8850 * gcc-ar.c: Remove smartness of lookup.
8852 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
8854 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
8855 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
8856 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
8858 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8861 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
8862 (setcc + and to xor + setcc): New peephole2.
8864 2017-10-19 Tom de Vries <tom@codesourcery.com>
8866 * doc/sourcebuild.texi (Test Directives, Variants of
8867 dg-require-support): Add dg-require-stack-size.
8869 2017-10-19 Martin Liska <mliska@suse.cz>
8872 * gimplify.c (gimplify_decl_expr): Do not instrument variables
8873 that have a large alignment.
8874 (gimplify_target_expr): Likewise.
8876 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
8878 PR rtl-optimization/82602
8879 * ira.c (rtx_moveable_p): Return false for volatile asm.
8881 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
8884 * config/i386/i386-modes.def (CCGZ): New CC mode.
8885 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
8886 * config/i386/predicates.md (ix86_comparison_operator):
8888 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
8889 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
8890 with double-word subtraction.
8891 (put_condition_code): Handle CCGZmode.
8893 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
8895 * wide-int.cc (debug (const wide_int &)): New.
8896 (debug (const wide_int *)): New.
8897 (debug (const widest_int &)): New.
8898 (debug (const widest_int *)): New.
8900 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
8903 * lra-constraints.c (curr_insn_transform): Use non-input operand
8904 instead of output one for matched reload.
8906 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8908 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
8909 (tree-ssa-loop-ivopts.h): New header file.
8910 (struct builtin_info): New fields.
8911 (classify_builtin_1): Compute and record base and offset parts for
8912 memset builtin partition by calling strip_offset.
8913 (offset_cmp, fuse_memset_builtins): New functions.
8914 (finalize_partitions): Fuse adjacent memset partitions by calling
8916 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
8917 Expose the interface.
8918 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
8920 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8922 PR tree-optimization/82574
8923 * tree-loop-distribution.c (find_single_drs): New parameter. Check
8924 that data reference must be executed exactly once per iteration
8925 against the outermost loop in nest.
8926 (classify_partition): Update call to above function.
8928 2017-10-18 Richard Biener <rguenther@suse.de>
8930 PR tree-optimization/82591
8931 * graphite.c (graphite_transform_loops): Move code gen message
8933 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8934 Here. Handle scop_to_isl_ast failing.
8935 (scop_to_isl_ast): Limit the number of ISL operations.
8937 2017-10-18 Richard Biener <rguenther@suse.de>
8939 * graphite-isl-ast-to-gimple.c
8940 (translate_isl_ast_to_gimple::set_rename): Simplify.
8941 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
8942 (graphite_copy_stmts_from_block): ... here.
8943 (copy_bb_and_scalar_dependences): Simplify.
8944 (add_parameters_to_ivs_params): Canonicalize.
8945 (generate_entry_out_of_ssa_copies): Simplify.
8946 * graphite-sese-to-poly.c (extract_affine_name): Simplify
8947 by passing in ISL dimension.
8948 (parameter_index_in_region_1): Rename to ...
8949 (parameter_index_in_region): ... this.
8950 (extract_affine): Adjust assert, pass down parameter index.
8951 (add_param_constraints): Use range-info when available.
8952 (build_scop_context): Adjust.
8953 * sese.c (new_sese_info): Adjust.
8954 (free_sese_info): Likewise.
8955 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
8956 Remove unused typedefs.
8957 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
8959 2017-10-18 Martin Liska <mliska@suse.cz>
8961 * combine.c (simplify_compare_const): Add gcc_fallthrough.
8963 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8965 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
8966 (s390_sched_init): Do not reset s390_sched_state if we entered the
8967 current basic block via a fallthru edge and all others are unlikely.
8969 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8971 * config/s390/s390.c (NUM_SIDES): New variable.
8972 (LONGRUNNING_THRESHOLD): New variable.
8973 (LATENCY_FACTOR): New variable.
8974 (s390_sched_score): Decrease score for long-running instructions on
8976 (s390_sched_variable_issue): Perform bookkeeping for long-running
8979 2017-10-18 Richard Biener <rguenther@suse.de>
8981 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
8982 Simplify with removal of the parameter rename map.
8983 (set_rename): Likewise.
8984 (should_copy_to_new_region): Likewise.
8985 (graphite_copy_stmts_from_block): Likewise.
8986 (copy_bb_and_scalar_dependences): Remove initialization of
8987 unused copied_bb_map.
8989 (copy_internal_parameters): Likewise.
8990 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
8991 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
8992 Use INTEGRAL_TYPE_P.
8993 (parameter_index_in_region_1): Rename to ...
8994 (assign_parameter_index_in_region): ... this. Assert we have
8995 a parameter we handle.
8996 (scan_tree_for_params): Adjust.
8997 * sese.h (parameter_rename_map_t): Remove.
8998 (struct sese_info_t): Remove unused parameter_rename_map and
8999 copied_bb_map members.
9000 * sese.c (new_sese_info): Adjust.
9001 (free_sese_info): Likewise.
9003 2017-10-18 Martin Liska <mliska@suse.cz>
9006 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
9007 on an abnormal edge.
9009 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
9011 * doc/invoke.texi (ffunction-sections and fdata-sections):
9014 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
9016 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
9017 the use statement can throw internally.
9019 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
9021 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
9022 any RTX present on the RHS of a SET.
9023 * compare-elim.c (try_eliminate_compare): Restore comment.
9025 2017-10-17 Jakub Jelinek <jakub@redhat.com>
9027 * langhooks.h (struct lang_hooks): Document that tree_size langhook
9028 may be also called on tcc_type nodes.
9029 * langhooks.c (lhd_tree_size): Likewise.
9031 2017-10-17 David Malcolm <dmalcolm@redhat.com>
9033 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
9034 format_warning_at_substring.
9035 (maybe_warn): Convert source_range * param to a location_t. Pass
9036 UNKNOWN_LOCATION rather than NULL to fmtwarn.
9037 (format_directive): Remove code to extract source_ranges and
9038 source_range * in favor of just a location_t.
9039 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
9041 * substring-locations.c (format_warning_va): Convert
9042 source_range * param to a location_t.
9043 (format_warning_at_substring): Likewise.
9044 * substring-locations.h (format_warning_va): Likewise.
9045 (format_warning_at_substring): Likewise.
9047 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
9049 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
9051 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
9052 and scatter/gather ops.
9054 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
9055 vec_gather_load and vec_scatter_store.
9056 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
9057 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
9058 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
9059 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
9060 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
9061 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
9063 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
9065 * reg-stack.c (compare_for_stack_reg): Add bool argument.
9066 Detect FTST instruction and handle its register pops. Only pop
9067 second operand if can_pop_second_op is true.
9068 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
9069 set can_pop_second_op to false in the compare_for_stack_reg call.
9071 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
9072 output_fp_compare for stack register operands.
9073 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
9074 instructions here. Do not emit stack register pops here. Assert
9075 that FCOMPP pops next to top stack register. Rewrite function.
9077 2017-10-17 Nathan Sidwell <nathan@acm.org>
9080 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
9081 use DECL_ASSEMBLER_NAME_RAW.
9084 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
9087 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
9088 Wilco Dijkstra <wilco.dijkstra@arm.com>
9090 * builtins.c (expand_builtin_update_setjmp_buf): Add a
9091 converstion to Pmode from the buf_addr.
9093 2017-10-17 Richard Biener <rguenther@suse.de>
9095 * graphite-dependences.c (scop_get_reads_and_writes): Change
9096 output parameters to references.
9098 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
9100 PR 71026/tree-optimization
9101 * fold-const.c (distribute_real_division): Removed.
9102 (fold_binary_loc): Remove calls to distribute_real_divison.
9104 2017-10-17 Richard Biener <rguenther@suse.de>
9106 * graphite-scop-detection.c
9107 (scop_detection::stmt_has_simple_data_refs_p): Always use
9108 the full nest as region.
9109 (try_generate_gimple_bb): Likewise.
9110 * sese.c (scalar_evolution_in_region): Simplify now that
9111 SCEV can handle instantiation in regions.
9112 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
9113 in the non-loop part of a function if requested.
9115 2017-10-17 Richard Biener <rguenther@suse.de>
9117 PR tree-optimization/82563
9118 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
9120 (graphite_regenerate_ast_isl): Call it.
9121 * graphite-scop-detection.c (build_scops): Remove entry edge split.
9123 2017-10-17 Jakub Jelinek <jakub@redhat.com>
9125 PR tree-optimization/82549
9126 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
9127 Formatting fixes. Instead of calling make_bit_field_ref with negative
9130 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
9132 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
9133 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
9134 _mm_maskz_reduce_ss): New.
9135 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
9136 __builtin_ia32_reducess_mask): Ditto..
9137 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
9138 * config/i386/sse.md (reduces<mode>): Renamed to ...
9139 (reduces<mode><mask_scalar_name>): ... this.
9140 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
9142 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
9143 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
9145 2017-10-16 David Malcolm <dmalcolm@redhat.com>
9147 * Makefile.in (OBJS): Add unique-ptr-tests.o.
9148 * selftest-run-tests.c (selftest::run_tests): Call
9149 selftest::unique_ptr_tests_cc_tests.
9150 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
9151 * unique-ptr-tests.cc: New file.
9153 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
9156 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9158 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9159 (make_hard_regno_born, make_hard_regno_dead): Update
9160 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
9162 2017-10-16 Jeff Law <law@redhat.com>
9164 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
9166 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
9168 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
9170 2017-10-16 Olivier Hainque <hainque@adacore.com>
9172 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
9173 with_cpu if we were configured for an e500v2 target cpu name.
9175 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
9177 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
9178 * doc/invoke.texi: Document +nodsp as a valid extension for
9181 2017-10-16 Martin Liska <mliska@suse.cz>
9183 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
9184 (test_set_range): Likewise.
9185 (test_range_functions): Rename to ...
9186 (test_bit_in_range): ... this.
9187 (sbitmap_c_tests): Add new test.
9189 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9191 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
9193 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
9194 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
9196 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9198 * config/aarch64/aarch64-builtins.c
9199 (aarch64_types_quadopu_lane_qualifiers): New.
9200 (TYPES_QUADOPU_LANE): New.
9201 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
9202 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
9203 (aarch64_<sur>dot_laneq<vsi2qi>): New.
9204 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
9205 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
9206 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
9207 (Vdottype, DOTPROD): New.
9208 (sur): Add SDOT and UDOT.
9210 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9212 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
9213 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
9214 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
9216 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
9217 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
9218 Enable TARGET_DOTPROD.
9219 (cortex-a75.cortex-a55): Likewise.
9220 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
9222 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9224 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
9225 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
9227 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
9229 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
9230 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
9231 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
9232 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
9233 * config/arm/types.md (neon_dot, neon_dot_q): New.
9234 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
9236 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9238 * config/arm/arm.h (TARGET_DOTPROD): New.
9239 * config/arm/arm.c (arm_arch_dotprod): New.
9240 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
9241 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
9242 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
9243 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
9244 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
9245 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
9246 * doc/invoke.texi (armv8.2-a): Document dotprod
9248 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
9250 * i386.c (ix86_vec_cost): New function.
9251 (ix86_rtx_costs): Handle vector operations better.
9252 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
9253 * x86-tune-costs.h: Add new costs to all tables.
9255 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
9257 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
9259 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
9260 divsd, sqrtss and sqrtsd
9261 * x86-tune-costs.h: Add new entries to all costs.
9262 (znver1_cost): Fix to match real instruction latencies.
9264 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9265 Michael Collison <michael.collison@arm.com>
9267 * compare-elim.c: Include emit-rtl.h.
9268 (can_merge_compare_into_arith): New function.
9269 (try_validate_parallel): Likewise.
9270 (try_merge_compare): Likewise.
9271 (try_eliminate_compare): Call the above when no previous clobber
9273 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
9276 2017-10-14 Jakub Jelinek <jakub@redhat.com>
9280 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
9281 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
9285 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
9286 to be any operand_equal_p operands. For & (B - 1) require
9287 B to be power of 2. Recognize
9288 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
9290 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
9293 * optabs.c (expand_memory_blockage): Fix call of
9294 targetm.have_memory_blockage.
9296 2017-10-14 Jakub Jelinek <jakub@redhat.com>
9299 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
9300 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
9301 objects to extra_objs instead of overwriting it.
9303 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
9305 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
9306 Use any_fp_register_operand as operand[3] predicate. Simplify
9307 equality test for operands[2] and operands[4] memory location.
9308 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
9309 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
9310 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
9311 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
9312 any_fp_register_operand as operand[1] predicate. Simplify
9313 equality test for operands[0] and operands[3] memory location.
9314 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
9315 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
9316 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
9318 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
9320 * target-insns.def: Add memory_blockage.
9321 * optabs.c (expand_memory_blockage): New function.
9322 (expand_asm_memory_barrier): Rename ...
9323 (expand_asm_memory_blockage): ... to this.
9324 (expand_mem_thread_fence): Call expand_memory_blockage
9325 instead of expand_asm_memory_barrier.
9326 (expand_mem_singnal_fence): Ditto.
9327 (expand_atomic_load): Ditto.
9328 (expand_atomic_store): Ditto.
9329 * doc/md.texi (Standard Pattern Names For Generation):
9330 Document memory_blockage instruction pattern.
9332 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
9334 * config/rl78/rl78.c (rl78_emit_libcall): New function.
9335 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
9336 * config/rl78/rl78.md: New define_expand "adddi3".
9338 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
9340 * cfghooks.c (verify_flow_info): Disable check that all probabilities
9343 2017-10-13 Jeff Law <law@redhat.com>
9345 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
9347 2017-10-13 Jakub Jelinek <jakub@redhat.com>
9350 * internal-fn.c (expand_mul_overflow): If both operands have
9351 the same highpart of -1 or 0 and the topmost bit of lowpart
9352 is different, overflow is if res <= 0 rather than res < 0.
9354 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
9356 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
9357 TARGET_P9_VECTOR code for unaligned_load case.
9359 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
9361 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
9363 2017-10-13 Nathan Sidwell <nathan@acm.org>
9365 * tree-core.h (tree_contains_struct): Make bool.
9366 * tree.c (tree_contains_struct): Likewise.
9367 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
9368 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
9369 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
9370 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
9371 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
9373 2017-10-13 Richard Biener <rguenther@suse.de>
9375 * graphite-isl-ast-to-gimple.c
9376 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
9377 parameters and dominance check.
9378 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
9379 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
9380 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
9381 Do not update SSA form here or do intermediate IL verification.
9382 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
9383 (graphite_initialize): Remove check on the number of loops in
9384 the function and inline into graphite_transform_loops.
9385 (graphite_finalize): Inline into graphite_transform_loops.
9386 (graphite_transform_loops): Perform SSA update and IL verification
9388 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
9390 2017-10-13 Richard Biener <rguenther@suse.de>
9392 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
9393 graphite_expression_type_precision): Avoid global constructor
9395 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
9396 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
9397 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
9398 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
9399 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
9400 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
9402 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
9405 * config/i386/i386.h (ix86_red_zone_size): New.
9406 * config/i386/i386.md (push peephole2s): Replace
9407 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
9409 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9410 Alan Hayward <alan.hayward@arm.com>
9411 David Sherwood <david.sherwood@arm.com>
9413 * combine.c (can_change_dest_mode): Reject changes in
9414 REGMODE_NATURAL_SIZE.
9416 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9417 Alan Hayward <alan.hayward@arm.com>
9418 David Sherwood <david.sherwood@arm.com>
9420 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
9421 (expand_debug_source_expr): Likewise.
9422 * combine.c (combine_simplify_rtx): Likewise.
9423 * cse.c (fold_rtx): Likewise.
9424 * fwprop.c (canonicalize_address): Likewise.
9425 * targhooks.c (default_shift_truncation_mask): Likewise.
9427 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9428 Alan Hayward <alan.hayward@arm.com>
9429 David Sherwood <david.sherwood@arm.com>
9431 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
9432 (widened_mode): Likewise.
9433 (expand_unop): Likewise.
9434 * ree.c (transform_ifelse): Likewise.
9435 (merge_def_and_ext): Likewise.
9436 (combine_reaching_defs): Likewise.
9437 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9439 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9440 Alan Hayward <alan.hayward@arm.com>
9441 David Sherwood <david.sherwood@arm.com>
9443 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
9444 * combine.c (gen_lowpart_for_combine): Likewise.
9445 * dwarf2out.c (rtl_for_decl_location): Likewise.
9446 * final.c (alter_subreg): Likewise.
9447 * rtlhooks.c (gen_lowpart_general): Likewise.
9448 (gen_lowpart_if_possible): Likewise.
9450 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9451 Alan Hayward <alan.hayward@arm.com>
9452 David Sherwood <david.sherwood@arm.com>
9454 * calls.c (expand_call): Use subreg_lowpart_offset.
9455 * cse.c (cse_insn): Likewise.
9456 * regcprop.c (copy_value): Likewise.
9457 (copyprop_hardreg_forward_1): Likewise.
9459 2017-10-13 Jakub Jelinek <jakub@redhat.com>
9462 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
9463 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
9464 =Q constraints to +Q and into insn condition add check
9465 that operands[0] and operands[1] are equal.
9466 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
9467 =Q constraints to +Q and into insn condition add check
9468 that operands[0] is equal to either operands[1] or operands[2].
9471 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
9472 instead of handling MINUS_EXPR twice (once for each argument),
9473 canonicalize operand order and handle just once, use rtype where
9474 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
9477 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
9478 any values of __C while still being pattern recognizable as a simple
9481 2017-10-13 Richard Biener <rguenther@suse.de>
9483 PR tree-optimization/82451
9485 2017-10-02 Richard Biener <rguenther@suse.de>
9487 PR tree-optimization/82355
9488 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
9489 a mapping for the enclosing loop but avoid generating one for
9491 (copy_bb_and_scalar_dependences): Remove premature codegen
9492 error on PHIs in blocks duplicated into multiple places.
9493 * graphite-scop-detection.c
9494 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
9495 in the region use it as loop and nest to analyze the DR in.
9496 (try_generate_gimple_bb): Likewise.
9497 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
9498 (add_loop_constraints): For blocks in a loop not in the region
9499 create a dimension with a single iteration.
9500 * sese.h (gbb_loop_at_index): Remove assert.
9502 * cfgloop.c (loop_preheader_edge): For the loop tree root
9503 return the single successor of the entry block.
9504 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
9505 Reset the SCEV hashtable and niters.
9506 * graphite-scop-detection.c
9507 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
9508 assert that we only have POLYNOMIAL_CHREC that vary in loops
9509 contained in the region.
9510 (scop_detection::graphite_can_represent_expr): Adjust.
9511 (scop_detection::stmt_has_simple_data_refs_p): For loops
9512 not in the region set loop to NULL. The nest is now the
9513 entry edge to the region.
9514 (try_generate_gimple_bb): Likewise.
9515 * sese.c (scalar_evolution_in_region): Adjust for
9516 instantiate_scev change.
9517 * tree-data-ref.h (graphite_find_data_references_in_stmt):
9518 Make nest parameter the edge into the region.
9519 (create_data_ref): Likewise.
9520 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
9521 entry edge into a region and adjust instantiate_scev calls.
9522 (create_data_ref): Likewise.
9523 (graphite_find_data_references_in_stmt): Likewise.
9524 (find_data_references_in_stmt): Pass the loop preheader edge
9525 from the nest argument.
9526 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
9527 parameter the edge into the region.
9528 (instantiate_parameters): Use the loop preheader edge as entry.
9529 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
9531 (get_instantiated_value_entry): Make instantiate_below parameter
9532 the edge into the region.
9533 (instantiate_scev_name): Likewise. Adjust dominance checks,
9534 when we cannot use loop-based instantiation instantiate by
9535 walking use-def chains.
9536 (instantiate_scev_poly): Adjust.
9537 (instantiate_scev_binary): Likewise.
9538 (instantiate_scev_convert): Likewise.
9539 (instantiate_scev_not): Likewise.
9540 (instantiate_array_ref): Remove.
9541 (instantiate_scev_3): Likewise.
9542 (instantiate_scev_2): Likewise.
9543 (instantiate_scev_1): Likewise.
9544 (instantiate_scev_r): Do not blindly handle N-operand trees.
9545 Do not instantiate array-refs. Handle all constants and invariants.
9546 (instantiate_scev): Make instantiate_below parameter
9547 the edge into the region.
9548 (resolve_mixers): Use the loop preheader edge for the region
9549 parameter to instantiate_scev_r.
9550 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
9552 2017-10-13 Richard Biener <rguenther@suse.de>
9554 PR tree-optimization/82525
9555 * graphite-isl-ast-to-gimple.c
9556 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
9558 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
9559 Fail code generation when we cannot represent the isl integer.
9560 (binary_op_to_tree): Elide modulo operations that are no-ops
9561 in the type we code generate. Remove now superfluous code
9564 2017-10-13 Richard Biener <rguenther@suse.de>
9566 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
9567 (scop_detection::harmful_loop_in_region): Remove premature
9568 IV type restriction.
9569 (scop_detection::graphite_can_represent_scev): We can handle
9570 pointer IVs just fine.
9572 2017-10-13 Alan Modra <amodra@gmail.com>
9574 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
9575 "Clobbers and Scratch Registers". Add paragraph on
9576 alternative to clobbers for scratch registers and OpenBLAS
9579 2017-10-13 Alan Modra <amodra@gmail.com>
9581 * doc/extend.texi (Clobbers): Correct vax example. Delete old
9582 example of a memory input for a string of known length. Move
9583 commentary out of table. Add a number of new examples
9584 covering array memory inputs.
9586 2017-10-12 Martin Liska <mliska@suse.cz>
9588 PR tree-optimization/82493
9589 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
9590 (test_range_functions): New function.
9591 (sbitmap_c_tests): Likewise.
9592 * selftest-run-tests.c (selftest::run_tests): Run new tests.
9593 * selftest.h (sbitmap_c_tests): New function.
9595 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
9597 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
9599 * config/rs6000/amo.h: Fix spacing issue.
9601 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9604 * config/i386/i386.md (*ashl<mode>3_mask_1,
9605 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
9606 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
9609 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
9611 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
9612 (profile_probability): Set max_probability
9613 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
9615 * profile-count.c (profile_probability::differs_from_p): Do not
9616 rely on max_probaiblity == 10000
9618 2017-10-12 Jeff Law <law@redhat.com>
9620 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
9623 2017-10-12 Martin Sebor <msebor@redhat.com>
9627 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
9628 (handle_alias_pairs): Call it.
9629 * common.opt (-Wattribute-alias): New option.
9630 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
9631 * doc/invoke.texi (-Wattribute-alias): Document.
9633 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
9636 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9638 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9640 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9641 (make_hard_regno_born, make_hard_regno_dead): Update
9642 bb_killed_pseudos and bb_gen_pseudos.
9644 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
9646 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
9648 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
9650 * config/alpha/alpha.c (alpha_split_conditional_move):
9651 Use std::swap instead of manually swapping.
9652 (alpha_stdarg_optimize_hook): Ditto.
9653 (alpha_canonicalize_comparison): Ditto.
9655 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9657 * tree-loop-distribution.c (struct builtin_info): New struct.
9658 (struct partition): Refactor fields into struct builtin_info.
9659 (partition_free): Free struct builtin_info.
9660 (build_size_arg_loc, build_addr_arg_loc): Delete.
9661 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
9662 information from struct builtin_info.
9663 (find_single_drs): New function refactored from classify_partition.
9664 Also moved builtin validity checks to this function.
9665 (compute_access_range, alloc_builtin): New functions.
9666 (classify_builtin_st, classify_builtin_ldst): New functions.
9667 (classify_partition): Refactor code into functions find_single_drs,
9668 classify_builtin_st and classify_builtin_ldst.
9669 (distribute_loop): Don't do runtime alias check when distributing
9671 (find_seed_stmts_for_distribution): New function.
9672 (pass_loop_distribution::execute): Refactor code finding seed
9673 stmts into above function. Support distribution for the innermost
9674 two-level loop nest. Adjust dump information.
9676 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9678 * tree-loop-distribution.c: Adjust the general comment.
9679 (NUM_PARTITION_THRESHOLD): New macro.
9680 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
9681 (classify_partition): Skip builtin pattern of loop nest's inner loop.
9682 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
9683 in call to build_partition_graph.
9684 (finalize_partitions): New parameter. Make loop distribution more
9685 conservative by fusing more partitions.
9686 (distribute_loop): Don't do runtime alias check in case of loop nest
9688 (find_seed_stmts_for_distribution): New function.
9689 (prepare_perfect_loop_nest): New function.
9690 (pass_loop_distribution::execute): Refactor code finding seed stmts
9691 and loop nest into above functions. Support loop nest distribution.
9692 Adjust dump information accordingly.
9694 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9696 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
9697 and set PTYPE_SEQUENTIAL for merged partition.
9699 2017-10-12 Richard Biener <rguenther@suse.de>
9701 PR tree-optimization/69728
9703 2017-09-19 Richard Biener <rguenther@suse.de>
9705 PR tree-optimization/69728
9706 * graphite-sese-to-poly.c (schedule_error): New global.
9707 (add_loop_schedule): Handle empty domain by failing the
9709 (build_original_schedule): Handle schedule_error.
9711 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
9712 domain by returning an unchanged schedule.
9714 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9716 * genrecog.c (validate_pattern): For VEC_SELECT verify that
9717 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
9719 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
9721 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
9724 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9727 * expr.c (store_field): Don't optimize away bitsize == 0 store
9728 from CALL_EXPR with addressable return type.
9730 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9732 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
9733 * config/rs6000/rs6000.md (sel): Delete mode attribute.
9734 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
9735 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
9736 TARGET_ISEL instead of TARGET_ISEL<sel>.
9738 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
9740 * config/rs6000/rs6000.c
9741 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
9743 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9745 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
9747 (reg_or_cint_operand): Fix comment.
9748 (reg_or_zero_operand): New predicate.
9749 * config/rs6000/rs6000-protos.h (output_isel): Delete.
9750 * config/rs6000/rs6000.c (output_isel): Delete.
9751 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
9752 instead of reg_or_cint_operand. Output instruction directly (not via
9754 (isel_unsigned_<mode>): Ditto.
9755 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
9756 gpc_reg_operand. Add an instruction alternative for this. Output
9757 instruction directly.
9758 (*isel_reversed_unsigned_<mode>): Ditto.
9760 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9762 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
9763 (TARGET_CANONICALIZE_COMPARISON): Define.
9765 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
9768 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
9769 Check whether the dest is REG before adding REG_EQUIV note.
9771 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9774 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9776 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9777 (make_hard_regno_born, make_hard_regno_dead): Update
9778 bb_killed_pseudos and bb_gen_pseudos.
9780 2017-10-11 Nathan Sidwell <nathan@acm.org>
9782 * incpath.h (enum incpath_kind): Name enum, prefix values.
9783 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
9784 * incpath.c (heads, tails): Use INC_MAX.
9785 (add_env_var_paths, add_standard_paths): Use incpath_kind.
9786 (merge_include_chains, split_quote_chain,
9787 register_include_chains): Update incpath_kind names.
9788 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
9789 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
9791 (add_framework_path, darwin_register_objc_includes): Likewise.
9792 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
9794 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9796 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
9797 Do not use float_operator operator predicate.
9798 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
9799 * config/i386/predicates.md (float_operator): Remove predicate.
9801 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9803 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
9804 (*jccxf_i387): Ditto.
9805 (*jcc<mode>_i387): Ditto.
9806 (*jccu<mode>_i387): Ditto.
9807 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
9808 (*jcc_*_i387 splitters): Remove.
9809 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
9810 * config/i386/i386.c (ix86_split_fp_branch): Remove.
9811 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
9814 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9816 * profile-count.h (slow_safe_scale_64bit): New function.
9817 (safe_scale_64bit): New inline.
9818 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
9819 * profile-count.c: Include wide-int.h
9820 (slow_safe_scale_64bit): New.
9822 2017-10-11 Nathan Sidwell <nathan@acm.org>
9824 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
9825 HAS_DECL_ASSEMBLER_NAME_P.
9826 * gimple-expr.c (gimple_decl_printable_name: Check
9827 HAS_DECL_ASSEMBLER_NAME_P too.
9828 * ipa-utils.h (type_in_anonymous_namespace_p): Check
9829 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
9830 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
9831 * passes.c (rest_of_decl_compilation): Check
9832 HAS_DECL_ASSEMBLER_NAME_P too.
9833 * recog.c (verify_changes): Likewise.
9834 * tree-pretty-print.c (dump_decl_name): Likewise.
9835 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
9837 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
9838 (SET_DECL_ASSEMBLER_NAME): Use it.
9839 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
9840 (COPY_DECL_ASSEMBLER_NAME): Likewise.
9841 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
9843 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9845 * config.gcc (i386, x86_64): Add extra objects.
9846 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
9847 (ix86_min_insn_size): Declare.
9848 (ix86_issue_rate): Declare.
9849 (ix86_adjust_cost): Declare.
9850 (ia32_multipass_dfa_lookahead): Declare.
9851 (ix86_macro_fusion_p): Declare.
9852 (ix86_macro_fusion_pair_p): Declare.
9853 (ix86_bd_has_dispatch): Declare.
9854 (ix86_bd_do_dispatch): Declare.
9855 (ix86_core2i7_init_hooks): Declare.
9856 (ix86_atom_sched_reorder): Declare.
9857 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
9858 (COSTS_N_BYTES): Move to x86-tune-costs.h.
9859 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
9860 (rip_relative_addr_p): Rename to ...
9861 (ix86_rip_relative_addr_p): ... this one; export.
9862 (memory_address_length): Update.
9863 (ix86_issue_rate): Move to x86-tune-sched.c.
9864 (ix86_flags_dependent): Move to x86-tune-sched.c.
9865 (ix86_agi_dependent): Move to x86-tune-sched.c.
9866 (exact_dependency_1): Move to x86-tune-sched.c.
9867 (exact_store_load_dependency): Move to x86-tune-sched.c.
9868 (ix86_adjust_cost): Move to x86-tune-sched.c.
9869 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
9870 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
9871 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
9872 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
9873 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
9874 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
9875 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
9876 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
9877 (min_insn_size): Rename to ...
9878 (ix86_min_insn_size): ... this one; export.
9879 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
9880 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
9881 (core2i7_first_cycle_multipass_backtrack): Move to
9882 x86-tune-sched-core.c.
9883 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
9884 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
9885 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
9886 (ix86_avoid_jump_mispredicts): Update.
9887 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
9888 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
9889 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
9890 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
9891 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
9892 (MAX_INSN): Move to ix86-tune-sched-bd.c.
9893 (MAX_IMM): Move to ix86-tune-sched-bd.c.
9894 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
9895 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
9896 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
9897 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
9898 (MAX_STORE): Move to ix86-tune-sched-bd.c.
9899 (BIG): Move to ix86-tune-sched-bd.c.
9900 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
9901 (enum insn_path): Move to ix86-tune-sched-bd.c.
9902 (get_mem_group): Move to ix86-tune-sched-bd.c.
9903 (is_cmp): Move to ix86-tune-sched-bd.c.
9904 (dispatch_violation): Move to ix86-tune-sched-bd.c.
9905 (is_branch): Move to ix86-tune-sched-bd.c.
9906 (is_prefetch): Move to ix86-tune-sched-bd.c.
9907 (init_window): Move to ix86-tune-sched-bd.c.
9908 (allocate_window): Move to ix86-tune-sched-bd.c.
9909 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
9910 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
9911 (process_end_window): Move to ix86-tune-sched-bd.c.
9912 (allocate_next_window): Move to ix86-tune-sched-bd.c.
9913 (find_constant): Move to ix86-tune-sched-bd.c.
9914 (get_num_immediates): Move to ix86-tune-sched-bd.c.
9915 (has_immediate): Move to ix86-tune-sched-bd.c.
9916 (get_insn_path): Move to ix86-tune-sched-bd.c.
9917 (get_insn_group): Move to ix86-tune-sched-bd.c.
9918 (count_num_restricted): Move to ix86-tune-sched-bd.c.
9919 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
9920 (add_insn_window): Move to ix86-tune-sched-bd.c.
9921 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
9922 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
9923 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
9924 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
9925 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
9926 (do_dispatch): Move to ix86-tune-sched-bd.c.
9927 (has_dispatch): Move to ix86-tune-sched-bd.c.
9928 * config/i386/t-i386: Add new object files.
9929 * config/i386/x86-tune-costs.h: New file.
9930 * config/i386/x86-tune-sched-atom.c: New file.
9931 * config/i386/x86-tune-sched-bd.c: New file.
9932 * config/i386/x86-tune-sched-core.c: New file.
9933 * config/i386/x86-tune-sched.c: New file.
9935 2017-10-11 Liu Hao <lh_mouse@126.com>
9937 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
9938 the generic version below instead.
9939 (should_colorize): Recognize Windows consoles as terminals
9941 * pretty-print.c [__MINGW32__] (write_all): New function.
9942 [__MINGW32__] (find_esc_head): Likewise.
9943 [__MINGW32__] (find_esc_terminator): Likewise.
9944 [__MINGW32__] (eat_esc_sequence): Likewise.
9945 [__MINGW32__] (mingw_ansi_fputs): New function that handles
9947 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
9950 2017-10-11 Richard Biener <rguenther@suse.de>
9952 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
9953 Properly call analyze_scalar_evolution with the loop of the stmt.
9955 2017-10-11 Richard Biener <rguenther@suse.de>
9957 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
9958 * tree-core.h (tree_base): Add chrec_var union member.
9959 * tree.h (CHREC_VAR): Remove.
9960 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
9961 * tree-chrec.h (build_polynomial_chrec): Adjust.
9962 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
9963 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
9965 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
9967 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
9968 * match.pd: ... here.
9969 ((T) X == (T) Y): Relax condition.
9971 2017-10-11 Bin Cheng <bin.cheng@arm.com>
9973 PR tree-optimization/82472
9974 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
9976 (break_alias_scc_partitions): Update postorder number.
9978 2017-10-11 Martin Liska <mliska@suse.cz>
9981 * opts.c (parse_no_sanitize_attribute): Do not use error_value
9983 * opts.h (parse_no_sanitize_attribute): Remove last argument.
9985 2017-10-11 Martin Liska <mliska@suse.cz>
9987 * print-rtl.c (print_insn): Move declaration of idbuf
9988 to same scope as name.
9990 2017-10-11 Martin Liska <mliska@suse.cz>
9995 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9998 2017-10-11 Martin Liska <mliska@suse.cz>
10001 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
10002 volatile arguments.
10004 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
10006 * config.gcc (default_gnu_indirect_function): Default to yes for
10007 arm*-*-linux* with glibc.
10009 2017-10-11 Richard Biener <rguenther@suse.de>
10011 * tree-scalar-evolution.c (get_scalar_evolution): Handle
10012 default-defs and types we do not want to analyze.
10013 (interpret_loop_phi): Replace unreachable code with an assert.
10014 (compute_scalar_evolution_in_loop): Remove and inline ...
10015 (analyze_scalar_evolution_1): ... here, replacing condition with
10016 what makes the intent clearer. Remove handling of cases
10017 get_scalar_evolution now handles.
10019 2017-10-10 Jim Wilson <wilson@tuliptree.org>
10021 PR rtl-optimization/81434
10022 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
10023 comment for main loop. In sched_group_found if, also add checks for
10024 pass and min_cost_group.
10026 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
10028 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
10029 (rs6000_insn_cost): New function.
10030 * config/rs6000/rs6000.md (cost): New attribute.
10032 2017-10-10 Jakub Jelinek <jakub@redhat.com>
10033 H.J. Lu <hongjiu.lu@intel.com>
10037 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
10038 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
10039 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
10040 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
10041 to OPTION_MASK_ISA_AVX512VL - builtins that have both
10042 OPTION_MASK_ISA_MMX and some other bit set require both
10043 mmx and the ISAs without the mmx bit.
10044 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
10045 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
10046 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
10047 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
10048 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
10049 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
10050 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
10051 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
10052 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
10053 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
10054 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
10055 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
10056 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
10057 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
10058 Add OPTION_MASK_ISA_MMX.
10060 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
10062 * config.gcc (armv7*-*-freebsd*): New target.
10063 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
10065 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
10067 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
10068 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
10071 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
10073 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
10074 a template parameter.
10075 (WIDE_INT_REF_FOR): Update accordingly.
10076 * tree.h (wi::int_traits <const_tree>): Delete.
10077 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
10078 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
10079 (wi::tree_to_wide_ref): New typedef.
10080 (wi::to_wide): New function.
10081 * calls.c (get_size_range): Use wi::to_wide when operating on
10082 trees as wide_ints.
10083 * cgraph.c (cgraph_node::create_thunk): Likewise.
10084 * config/i386/i386.c (ix86_data_alignment): Likewise.
10085 (ix86_local_alignment): Likewise.
10086 * dbxout.c (stabstr_O): Likewise.
10087 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
10088 * expr.c (const_vector_from_tree): Likewise.
10089 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
10090 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
10091 (fold_negate_expr_1, int_const_binop_1, const_binop)
10092 (fold_convert_const_int_from_real, optimize_bit_field_compare)
10093 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
10094 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
10095 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
10096 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
10097 (fold_not_const, round_up_loc): Likewise.
10098 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
10099 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
10100 (alloca_call_type): Likewise.
10101 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
10102 * godump.c (go_output_typedef): Likewise.
10103 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
10104 * internal-fn.c (get_min_precision): Likewise.
10105 * ipa-cp.c (ipcp_store_vr_results): Likewise.
10106 * ipa-polymorphic-call.c
10107 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
10108 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
10109 (ipa_modify_call_arguments): Likewise.
10110 * match.pd: Likewise.
10111 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
10112 * print-tree.c (print_node_brief, print_node): Likewise.
10113 * stmt.c (expand_case): Likewise.
10114 * stor-layout.c (layout_type): Likewise.
10115 * tree-affine.c (tree_to_aff_combination): Likewise.
10116 * tree-cfg.c (group_case_labels_stmt): Likewise.
10117 * tree-data-ref.c (dr_analyze_indices): Likewise.
10118 (prune_runtime_alias_test_list): Likewise.
10119 * tree-dump.c (dequeue_and_dump): Likewise.
10120 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
10121 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
10122 * tree-pretty-print.c (dump_generic_node): Likewise.
10123 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
10124 (simple_iv_with_niters): Likewise.
10125 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
10126 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
10127 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
10128 * tree-ssa-loop-niter.c (split_to_var_and_offset)
10129 (refine_value_range_using_guard, number_of_iterations_ne_max)
10130 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
10131 (get_cst_init_from_scev, record_nonwrapping_iv)
10132 (scev_var_range_cant_overflow): Likewise.
10133 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
10134 * tree-ssa-pre.c (compute_avail): Likewise.
10135 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
10136 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
10137 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
10138 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
10139 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
10140 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
10141 (dump_case_nodes, try_switch_expansion): Likewise.
10142 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
10143 (vect_do_peeling): Likewise.
10144 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
10145 * tree-vect-stmts.c (vectorizable_load): Likewise.
10146 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
10147 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
10148 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
10149 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
10150 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
10151 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
10152 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
10153 (evrp_dom_walker::before_dom_children): Likewise.
10154 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
10155 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
10156 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
10157 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
10158 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
10159 (get_range_pos_neg): Likewise.
10160 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
10161 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
10162 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
10163 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
10164 * config/avr/avr.c (avr_fold_builtin): Likewise.
10165 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
10166 * config/msp430/msp430.c (msp430_attr): Likewise.
10167 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
10168 * config/powerpcspe/powerpcspe-c.c
10169 (altivec_resolve_overloaded_builtin): Likewise.
10170 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
10171 (rs6000_expand_ternop_builtin): Likewise.
10172 * config/rs6000/rs6000-c.c
10173 (altivec_resolve_overloaded_builtin): Likewise.
10174 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
10175 (rs6000_expand_ternop_builtin): Likewise.
10176 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
10178 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10180 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
10181 when copying loop nest with only one inner loop.
10183 2017-10-10 Richard Biener <rguenther@suse.de>
10185 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
10186 blocks if SCEV is active.
10187 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
10189 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
10190 (scev_initialize): Assert we are not yet initialized.
10192 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10194 * tree-loop-distribution.c (generate_loops_for_partition): Remove
10195 inner loop's exit stmt by making it always exit the loop, otherwise
10196 we would generate an infinite empty loop.
10198 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10200 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
10201 renaming variables in new preheader if it's deleted.
10203 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10205 * tree-loop-distribution.c (struct partition): Remove unused field
10206 loops of the structure.
10207 (partition_alloc, partition_free): Ditto.
10208 (build_rdg_partition_for_vertex): Ditto.
10210 2017-10-09 Jeff Law <law@redhat.com>
10212 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
10213 return type to match prototype and documentation.
10215 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10217 * config/rs6000/rs6000.c (processor_costs): Move to ...
10218 * config/rs6000/rs6000.h: ... here.
10219 (rs6000_cost): Declare.
10221 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
10223 * except.c (setjmp_fn): New global variable.
10224 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
10225 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
10226 if DONT_USE_BUILTIN_SETJMP is defined.
10228 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10230 * target.def (insn_cost): New hook.
10231 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
10232 * doc/tm.texi: Regenerate.
10233 * rtlanal.c (insn_cost): Use the new hook.
10235 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10237 * combine.c (combine_validate_cost): Compute the new insn_cost,
10238 not just pattern_cost.
10239 (try_combine): Adjust comment.
10241 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10243 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
10245 * combine.c (uid_insn_cost): Adjust comment.
10246 (combine_validate_cost): Adjust comment. Use pattern_cost instead
10248 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
10249 * dse.c (find_shift_sequence): Ditto.
10250 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
10251 (bb_valid_for_noce_process_p): Use pattern_cost.
10252 * rtl.h (insn_rtx_cost): Delete.
10253 (pattern_cost): New prototype.
10254 (insn_cost): New prototype.
10255 * rtlanal.c (insn_rtx_cost): Rename to...
10256 (pattern_cost): ... this.
10259 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
10261 * config/i386/i386.md (*jcc_2): Remove insn pattern.
10262 (*jcc<mode>_0_r_i387): Ditto.
10263 (*jccxf_r_i387): Ditto.
10264 (*jcc<mode>_r_i387): Ditto.
10265 (*jccu<mode>_r_i387): Ditto.
10266 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
10267 (*jcc): Rename from *jcc_1.
10269 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10271 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
10272 deferred rescans after the lvx/stvx recombination pre-pass.
10274 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
10276 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
10277 memory operation instruction support.
10278 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
10279 (rs6000-ibm-aix[789]*): Likewise.
10280 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
10281 Document new functions.
10283 2017-10-09 Richard Biener <rguenther@suse.de>
10285 PR tree-optimization/82397
10286 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
10287 equality only for semantically equal trees.
10289 2017-10-09 Richard Biener <rguenther@suse.de>
10291 PR tree-optimization/82449
10292 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
10293 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
10294 allow constant addresses.
10295 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
10298 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10300 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
10303 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10306 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
10309 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10312 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
10314 2017-10-09 Jakub Jelinek <jakub@redhat.com>
10317 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
10318 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
10320 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
10322 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
10323 (WI_BINARY_PREDICATE_RESULT): Likewise.
10324 (wi::binary_traits::operator_result): New type.
10325 (wi::binary_traits::predicate_result): Likewise.
10326 (generic_wide_int::operator~, unary generic_wide_int::operator-)
10327 (generic_wide_int::operator==, generic_wide_int::operator!=)
10328 (generic_wide_int::operator&, generic_wide_int::and_not)
10329 (generic_wide_int::operator|, generic_wide_int::or_not)
10330 (generic_wide_int::operator^, generic_wide_int::operator+
10331 (binary generic_wide_int::operator-, generic_wide_int::operator*):
10333 (operator~, unary operator-, operator==, operator!=, operator&)
10334 (operator|, operator^, operator+, binary operator-, operator*): New
10336 * expr.c (get_inner_reference): Use wi::bit_and_not.
10337 * fold-const.c (fold_binary_loc): Likewise.
10338 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
10339 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
10340 (bit_value_binop): Likewise.
10341 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
10342 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
10343 (extract_range_from_binary_expr_1): Likewise.
10344 (masked_increment): Likewise.
10345 (simplify_bit_ops_using_ranges): Likewise.
10347 2017-10-09 Martin Jambor <mjambor@suse.cz>
10350 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
10351 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
10352 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
10354 (hsa_fixup_mov_insn_type): New function.
10355 (hsa_op_with_type::get_in_type): Use it.
10356 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
10357 immediates in an assert.
10358 (hsa_op_with_type::extend_int_to_32bit): New method.
10359 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
10360 types. Convert to dest type if necessary.
10361 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
10362 (reg_for_gimple_ssa): Pass false as min32int to
10363 hsa_type_for_scalar_tree_type.
10364 (gen_hsa_addr): Fixup type when creating addresable temporary.
10365 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
10366 (gen_hsa_unary_operation): Extend operands and convert to dest type if
10367 necessary. Call hsa_fixup_mov_insn_type.
10368 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
10369 extend operands and convert to dest type if necessary.
10370 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
10371 to dest type if necessary.
10372 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
10373 if conversion nt necessary and size matches.
10374 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
10375 to dest type if necessary.
10376 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
10377 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
10379 (gen_hsa_clrsb): Likewise.
10380 (gen_hsa_ffs): Likewise.
10381 (gen_hsa_divmod): Extend operands and convert to dest type if
10383 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
10385 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
10387 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
10388 Remove empty default arguments. Use a brace block as output
10390 (conditional return): Ditto.
10392 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
10393 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
10395 (group_ending_nop): Ditto.
10396 (doloop_end): Ditto.
10397 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
10398 (splitters for those): Ditto.
10400 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
10402 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
10403 a conditional jump (and the compare for it) so that pc_rtx is the
10405 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
10406 for the deleted and renamed ctr<mode>_internal[234] patterns.
10407 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
10408 Delete second conditional return pattern.
10409 (ctr<mode>_internal2): Delete this second bdnz pattern.
10410 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
10411 (ctr<mode>_internal4): Delete this second bdz pattern.
10413 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
10415 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
10416 (always_initialized_rtx_for_ssa_name_p): New predicate.
10417 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
10418 (finish_out_of_ssa): Free new field of SA.
10419 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
10420 * tree-ssa-coalesce.c: Include tree-ssa.h.
10421 (get_parm_default_def_partitions): Remove extern keyword.
10422 (get_undefined_value_partitions): New function.
10423 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
10424 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
10425 uninitialized bits.
10426 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
10428 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
10430 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
10432 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
10434 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
10435 for targets that preffer 128bit.
10437 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
10439 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
10441 2017-10-08 Olivier Hainque <hainque@adacore.com>
10443 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
10444 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
10446 (thumb_set_return_address): Likewise.
10448 2017-10-08 Olivier Hainque <hainque@adacore.com>
10450 * common/config/arm/arm-common.c (arm_except_unwind_info):
10451 Handle DWARF2_UNWIND_INFO.
10453 2017-10-07 Michael Collison <michael.collison@arm.com>
10455 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
10458 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
10460 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
10461 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
10462 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
10463 defined, force the creation of a new block for a dispatch label.
10465 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
10467 * invoke.texi (Wsuggest-attribute=cold): Document.
10468 * common.opt (Wsuggest-attribute=cold): New
10469 * ipa-pure-const.c (warn_function_cold): New function.
10470 * predict.c (compute_function_frequency): Use it.
10471 * predict.h (warn_function_cold): Declare.
10473 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
10475 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
10478 2017-10-06 Martin Liska <mliska@suse.cz>
10480 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
10481 keyword for member functions.
10482 (struct sanopt_tree_couple): New struct.
10483 (struct sanopt_tree_couple_hash): New function.
10484 (struct sanopt_ctx): Add new hash_map.
10485 (has_dominating_ubsan_ptr_check): New function.
10486 (record_ubsan_ptr_check_stmt): Likewise.
10487 (maybe_optimize_ubsan_ptr_ifn): Likewise.
10488 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
10489 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
10491 2017-10-06 Sudakshina Das <sudi.das@arm.com>
10494 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
10495 aarch64_simd_valid_immediate on CONST_VECTORs.
10496 (aarch64_reg_or_bic_imm): Likewise.
10498 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
10500 PR rtl-optimization/82396
10501 * haifa-sched.c (ready_sort_real): Disable qsort checking.
10503 2017-10-06 Sebastian Pop <sebpop@gmail.com>
10505 * graphite-dependences.c (scop_get_reads): Move code to...
10506 (scop_get_must_writes): Move code to...
10507 (scop_get_may_writes): Move code to...
10508 (scop_get_reads_and_writes): ... here.
10509 (scop_get_dependences): Call scop_get_reads_and_writes.
10511 2017-10-06 Jakub Jelinek <jakub@redhat.com>
10513 PR tree-optimization/82434
10514 * fold-const.h (can_native_encode_type_p,
10515 can_native_encode_string_p): Remove.
10516 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
10517 don't encode anything, just return what would be otherwise returned.
10518 (native_encode_fixed, native_encode_complex, native_encode_vector):
10520 (native_encode_string): Likewise. Inline by hand
10521 can_native_encode_string_p.
10522 (can_native_encode_type_p): Remove.
10523 (can_native_encode_string_p): Remove.
10524 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
10525 STRING_CSTs using can_native_encode_string_p, test all
10526 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
10527 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
10528 argument from native_encode_expr.
10529 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
10530 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
10533 2017-10-06 Richard Biener <rguenther@suse.de>
10535 PR tree-optimization/82397
10536 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
10537 operand_equal_p but rely on data_ref_compare_tree for detecting
10539 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
10540 to match up with dr_group_sort_cmp.
10542 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10545 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
10547 * config/s390/s390-builtin-types.def: Regenerate.
10549 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10552 * config/s390/s390-builtin-types.def: Regenerate.
10553 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
10554 Change flag from B_VXE to B_VX.
10555 (s390_vec_min_dbl): Remove B_VXE flag.
10557 2017-10-06 Richard Biener <rguenther@suse.de>
10559 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
10560 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
10561 translate_isl_ast_to_gimple::is_valid_rename,
10562 translate_isl_ast_to_gimple::get_rename,
10563 translate_isl_ast_to_gimple::get_def_bb_for_const,
10564 translate_isl_ast_to_gimple::get_new_name,
10565 translate_isl_ast_to_gimple::collect_all_ssa_names,
10566 translate_isl_ast_to_gimple::copy_loop_phi_args,
10567 translate_isl_ast_to_gimple::collect_all_ssa_names,
10568 translate_isl_ast_to_gimple::copy_loop_phi_args,
10569 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
10570 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
10571 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
10572 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
10573 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
10574 translate_isl_ast_to_gimple::copy_cond_phi_args,
10575 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
10576 translate_isl_ast_to_gimple::edge_for_new_close_phis,
10577 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
10578 translate_isl_ast_to_gimple::rename_uses,
10579 translate_isl_ast_to_gimple::rename_all_uses): Remove.
10580 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
10581 (set_rename_for_each_def): Likewise.
10582 (graphite_copy_stmts_from_block): Handle debug stmt resetting
10583 here. Handle rewriting SCEV analyzable uses here.
10584 (copy_bb_and_scalar_dependences): Generate code for PHI
10586 (graphite_regenerate_ast_isl): Adjust.
10587 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
10588 (add_write, add_read): New functions.
10589 (build_cross_bb_scalars_def): Use it and simplify.
10590 (build_cross_bb_scalars_use): Likewise.
10591 (graphite_find_cross_bb_scalar_vars): Inline into...
10592 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
10593 simulating out-of-SSA. Compute liveout and add dependencies.
10594 (build_scops): Force an empty entry block.
10595 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
10597 (sese_build_liveouts): Declare.
10598 (sese_trivially_empty_bb_p): Likewise.
10599 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
10600 compute liveout and debug_liveout.
10601 (sese_bad_liveouts_use): Remove.
10602 (sese_reset_debug_liveouts_bb): Likewise.
10603 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
10604 (sese_build_liveouts): Build liveout and debug_liveout and store
10606 (new_sese_info): Adjust.
10607 (free_sese_info): Likewise.
10608 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
10609 do not build liveout here.
10610 (move_sese_in_condition): Adjust region entry.
10611 (scev_analyzable_p): Match up with chrec_apply requirements.
10612 (sese_trivially_empty_bb_p): New.
10613 * tree-into-ssa.c (get_reaching_def): Properly support generating
10614 default-defs for incremental rewrite of anonymous names.
10616 2017-10-06 Richard Biener <rguenther@suse.de>
10618 * graphite-sese-to-poly.c (extract_affine): For casts increasing
10619 precision do not perform modulo reduction.
10621 2017-10-06 Richard Biener <rguenther@suse.de>
10623 PR tree-optimization/82436
10624 * tree-vect-slp.c (vect_supported_load_permutation_p): More
10625 conservatively choose the vectorization factor when checking
10626 whether we can perform the required load permutation.
10627 (vect_transform_slp_perm_load): Assert when we may not fail.
10629 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
10631 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
10632 message for incompatible -msdata=* and -mcall-* options.
10634 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10636 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
10637 rate for post-reload scheduling.
10639 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10641 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
10643 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10645 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
10646 to improve monte carlo in scimark.
10648 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10650 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
10651 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
10652 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
10653 pentium4_cost, nocona_cost): Set reassociation width to 1.
10654 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
10655 width to 2 for fp operations and 1 otherwise.
10656 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
10658 (atom_cost): Set reassociation width to 2.
10659 (slm_cost, generic_cost): Set fp reassociation width
10660 to 2 and 1 otherwise.
10661 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
10662 (core_cost): Set fp reassociation width to 4 and vector to 2.
10663 (ix86_reassociation_width): Rewrite using cost table; special case
10664 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
10665 and TARGET_AVX128_OPTIMAL.
10666 * config/i386/i386.h (processor_costs): Add
10667 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
10668 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
10669 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
10670 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
10671 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
10672 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
10674 2017-10-05 Nathan Sidwell <nathan@acm.org>
10676 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
10678 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10680 * config/arm/arm.c (arm_test_fpu_data): New.
10681 (arm_run_selftests): Call arm_test_fpu_data.
10683 2017-10-04 Nathan Sidwell <nathan@acm.org>
10685 * toplev.c (toplev::main): Remove excess parens on pretty_printer
10687 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
10689 2017-10-04 Sudakshina Das <sudi.das@arm.com>
10691 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
10692 check type for aarch64_simd_valid_immediate.
10693 (aarch64_output_simd_mov_immediate): Update prototype.
10694 (aarch64_simd_valid_immediate): Update prototype.
10695 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
10696 support for ORR-immediate.
10697 (and<mode>3): modified pattern to add support for BIC-immediate.
10698 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
10699 now checks for valid immediate for BIC and ORR based on new enum
10701 (aarch64_output_simd_mov_immediate): Function now used to output
10702 BIC/ORR imm as well based on new enum argument.
10703 * config/aarch64/constraints.md (Do): New vector immediate constraint.
10705 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
10706 (aarch64_reg_or_bic_imm): Likewise.
10708 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10710 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
10711 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
10713 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10717 PR rtl-optimization/82396
10718 * haifa-sched.c (autopref_multipass_init): Simplify
10720 (autopref_rank_data): Simplify sort order.
10721 * sched-int.h (autopref_multipass_data_): Remove
10722 multi_mem_insn_p, min_offset and max_offset.
10724 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10726 * doc/sourcebuild.texi: Document vect_peeling_profitable.
10728 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10730 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
10731 vect_doubleint_cvt.
10733 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10735 * doc/sourcebuild.texi: Document vect_long_mult.
10737 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
10739 PR tree-optimization/82413
10740 * fold-const.c (build_range_check): Use widest_int when comparing
10741 the maximum ETYPE value with HIGH.
10743 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10745 PR rtl-optimization/82396
10746 * haifa-sched.c (autopref_multipass_init): Simplify
10748 (autopref_rank_data): Simplify sort order.
10749 * sched-int.h (autopref_multipass_data_): Remove
10750 multi_mem_insn_p, min_offset and max_offset.
10752 2017-10-04 Jakub Jelinek <jakub@redhat.com>
10754 PR tree-optimization/82381
10755 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
10756 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
10759 PR tree-optimization/82374
10760 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
10761 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
10762 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
10763 current_function_decl to the new decl.
10765 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
10767 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
10768 helper macro for IEEE float128 hardware built-in functions.
10769 (SQRTF128_ODD): Add built-in functions with the round-to-odd
10771 (TRUNCF128_ODD): Likewise.
10772 (ADDF128_ODD): Likewise.
10773 (SUBF128_ODD): Likewise.
10774 (MULF128_ODD): Likewise.
10775 (DIVF128_ODD): Likewise.
10776 (FMAF128_ODD): Likewise.
10777 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
10778 UNSPEC_TRUNC_ROUND_TO_ODD.
10779 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
10780 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
10781 floating point round to odd instructions.
10782 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
10783 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
10784 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
10785 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
10786 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
10787 (trunc<mode>sf2_hw): Change the truncate with round to odd
10788 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
10789 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
10790 to odd hardware instructions.
10791 (sub<mode>3_odd): Likewise.
10792 (mul<mode>3_odd): Likewise.
10793 (div<mode>3_odd): Likewise.
10794 (sqrt<mode>2_odd): Likewise.
10795 (fma<mode>4_odd): Likewise.
10796 (fms<mode>4_odd): Likewise.
10797 (nfma<mode>4_odd): Likewise.
10798 (nfms<mode>4_odd): Likewise.
10799 (trunc<mode>df2_odd): Change the truncate with round to odd
10800 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
10802 * doc/extend.texi (PowerPC built-in functions): Update documentation
10803 for existing IEEE float128-bit built-in functions. Add built-in
10804 functions that generate the IEEE 128-bit floating point round to
10807 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
10809 PR rtl-optimization/77729
10810 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
10811 to (X&(C1&~C2))|C2 transformations.
10813 2017-10-03 Martin Jambor <mjambor@suse.cz>
10815 PR tree-optimization/82363
10816 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
10817 mismatch, mark lacc written regardless of racc.
10819 2017-10-03 Jakub Jelinek <jakub@redhat.com>
10821 PR tree-optimization/82381
10822 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
10823 stmt_to_insert nor wheather SSA_NAMEs are default defs.
10824 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
10825 fallthrough into reassoc_stmt_dominates_stmt_p.
10828 * combine.c (combine_instructions): Don't combine in unreachable
10831 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
10834 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
10835 function to not use the have_cpu variable. Do not set cpu_index,
10836 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
10837 or the default cpu.
10838 (rs6000_valid_attribute_p): Remove duplicate initializations of
10839 old_optimize and func_optimize.
10840 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
10841 (rs6000_activate_target_options): Make global.
10842 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
10845 2017-10-02 Jakub Jelinek <jakub@redhat.com>
10847 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
10848 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
10849 Set *poffset to 0 and *psize and *pmax_size to -1 if
10850 *poffset + *psize overflows in HOST_WIDE_INT.
10852 PR tree-optimization/82387
10853 PR tree-optimization/82388
10854 PR tree-optimization/82389
10855 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
10856 instead of live_bytes non-NULL.
10858 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
10861 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
10864 2017-10-02 Richard Biener <rguenther@suse.de>
10866 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
10867 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
10868 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
10870 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10872 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
10873 requested precision matches the type's.
10874 * calls.c (alloc_max_size): Calculate the new candidate size as
10875 a widest_int and use wi::to_widest when comparing it with the
10876 current candidate size.
10877 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
10878 zero rather than integer_zero_node.
10879 * match.pd: Check for a no-op conversion before using wi::add
10880 rather than after. Use tree_to_uhwi when summing small shift
10881 counts into an unsigned int.
10883 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10884 Alan Hayward <alan.hayward@arm.com>
10885 David Sherwood <david.sherwood@arm.com>
10888 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
10889 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
10890 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
10891 POINTER_AND_FP_REGS.
10893 2017-10-02 Richard Biener <rguenther@suse.de>
10895 PR tree-optimization/82355
10896 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
10897 a mapping for the enclosing loop but avoid generating one for
10898 the loop tree root.
10899 (copy_bb_and_scalar_dependences): Remove premature codegen
10900 error on PHIs in blocks duplicated into multiple places.
10901 * graphite-scop-detection.c
10902 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
10903 in the region use it as loop and nest to analyze the DR in.
10904 (try_generate_gimple_bb): Likewise.
10905 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
10906 (add_loop_constraints): For blocks in a loop not in the region
10907 create a dimension with a single iteration.
10908 * sese.h (gbb_loop_at_index): Remove assert.
10910 2017-10-01 Kevin Buettner <kevinb@redhat.com>
10912 * omp-expand.c (adjust_context_scope): New function.
10913 (expand_parallel_call): Call adjust_context_scope.
10915 2017-10-01 Jeff Law <law@redhat.com>
10917 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
10918 dom_opt_dom_walker class with direct access to private members.
10919 Add comments. Call test_for_singularity.
10920 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
10921 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
10922 m_dummy_cond anymore.
10923 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
10925 (pass_dominator:execute): Build the dummy_cond here and pass it
10926 to the dom_opt_dom_walker ctor.
10927 (test_for_singularity): New function.
10929 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
10930 Maya Rashish <coypu@sdf.org>
10932 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
10933 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
10934 (alpha*-*-netbsd*) Use nbsd_tm_file.
10935 (arm*-*-netbsdelf*) Likewise.
10936 (i[34567]86-*-netbsdelf*) Likewise.
10937 (x86_64-*-netbsd*) Likewise.
10938 (mips*-*-netbsd*) Likewise.
10939 (powerpc-*-netbsd*) Likewise.
10940 (sh*-*-netbsd*) Likewise.
10941 (sparc-*-netbsdelf*) Likewise.
10942 (sparc64-*-netbsd*) Likewise.
10943 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
10945 (vax-*-netbsdelf*) Likewise.
10946 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
10947 (UINT_FAST8_TYPE) Likewise.
10948 (INT_FAST16_TYPE) Check CHAR_FAST16.
10949 (UINT_FAST16_TYPE) Likewise.
10951 2017-09-30 Jakub Jelinek <jakub@redhat.com>
10954 * config/i386/i386.md
10955 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
10956 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
10957 *divmodsi4_zext_2): New define_insn_and_split.
10958 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
10959 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
10960 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
10961 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
10962 New define_insn_and_split.
10963 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
10964 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
10965 operands[1] having DImode when mode is SImode.
10967 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
10968 always SImode for DIV and MOD in REG_EQUAL notes.
10970 2017-09-29 Yury Gribov <tetra2005@gmail.com>
10972 PR middle-end/82319
10973 * match.pd: Fix handling of NaNs in pattern.
10975 2017-09-29 Jeff Law <law@redhat.com>
10977 * sbitmap.c (bitmap_bit_in_range_p): New function.
10978 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
10979 * tree-ssa-dse.c (live_bytes_read): New function.
10980 (dse_classify_store): Ignore reads of dead bytes.
10982 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
10983 typos and whitespace errors.
10984 * config/i386/predicates.md (address_no_seg_operand): Likewise.
10985 * config/s390/s390.c (s390_emit_prologue): Likewise.
10987 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10990 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
10991 with a symbol for LRA.
10993 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10995 PR rtl-optimization/82338
10996 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
10998 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
11000 * genmodes.c (calc_wider_mode): Suppress qsort macro.
11001 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
11002 (qsort_chk): Declare.
11003 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
11004 (qsort_chk): New function.
11006 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11008 PR tree-optimization/82337
11009 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
11010 phi definition if the PHI result appears in an abnormal PHI.
11011 (find_basis_for_base_expr): Don't record a basis if the LHS of the
11012 basis appears in an abnormal PHI.
11014 2017-09-29 Richard Biener <rguenther@suse.de>
11016 * graphite-isl-ast-to-gimple.c
11017 (translate_isl_ast_to_gimple::set_codegen_error): New function.
11018 (binary_op_to_tree): Use it.
11019 (get_rename_from_scev): Likewise.
11020 (copy_loop_phi_nodes): Likewise.
11021 (copy_bb_and_scalar_dependences): Likewise.
11022 (translate_pending_phi_nodes): Likewise.
11024 2017-09-29 Jakub Jelinek <jakub@redhat.com>
11027 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
11028 for movabsq $(i32 << shift), r64.
11030 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
11032 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
11033 index when encoding %esp as %rsp to avoid 0x67 prefix.
11035 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
11037 * config/i386/i386.md (*movsf_internal, *movdf_internal):
11038 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
11040 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
11042 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
11043 Extensions with more than 16 double VFP registers.
11044 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
11045 to_clear_mask and all code related to it. Replace the remaining
11046 entry by a sbitmap and adapt code accordingly.
11048 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
11050 * brig-builtins.def: Change pure attributes to const.
11052 2017-09-28 Joseph Myers <joseph@codesourcery.com>
11054 * config.gcc (default_gnu_indirect_function): Default to yes for
11055 sparc*-*-linux* with glibc.
11057 2017-09-28 Joseph Myers <joseph@codesourcery.com>
11059 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
11060 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
11061 when creating .init_array and .fini_array sections with priority
11064 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
11067 * config/aarch64/aarch64.c
11068 (aarch64_builtin_support_vector_misalignment): Always return false
11069 when misalignment is unknown.
11071 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
11073 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
11074 this function to return false if the definition used by the swap
11075 instruction is artificial, or if the memory address from which the
11076 constant value is loaded is not represented by a base address held
11077 in a register or if the base address register is a frame or stack
11078 pointer. Additionally, return false if the base address of the
11079 loaded constant is a SYMBOL_REF but is not considered to be a
11081 (replace_swapped_load_constant): New function.
11082 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
11083 loaded constant vector with a load of a swapped constant vector.
11085 2017-09-27 Carl Love <cel@us.ibm.com>
11087 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
11088 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
11089 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
11092 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
11094 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
11095 first, always call autopref_rank_data otherwise.
11097 2017-09-27 Richard Biener <rguenther@suse.de>
11099 * graphite-scop-detection.c (find_scop_parameters): Move
11100 loop bound handling ...
11101 (gather_bbs::before_dom_children): ... here, avoiding the need
11102 to build scop_info->loop_nest.
11103 (record_loop_in_sese): Remove.
11104 * sese.h (sese_info_t::loop_nest): Remove.
11105 * sese.c (new_sese_info): Do not allocate loop_nest.
11106 (free_sese_info): Do not free loop_nest.
11108 2017-09-27 Jakub Jelinek <jakub@redhat.com>
11111 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
11112 lhs from calls if the lhs has addressable type.
11114 2017-09-27 Richard Biener <rguenther@suse.de>
11116 * graphite.h (scop::max_alias_set): New member.
11117 * graphite-scop-detection.c: Remove references to non-existing
11118 --param in comments.
11119 (build_alias_sets): Record the maximum alias set used for drs.
11120 (build_scops): Support zero as unlimited for
11121 --param graphite-max-arrays-per-scop.
11122 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
11123 and inline into ...
11124 (build_poly_sr_1): ... here. Compute alias set based on the
11125 maximum alias set used for drs rather than
11126 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
11128 2017-09-27 Richard Biener <rguenther@suse.de>
11130 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
11131 --param loop-block-tile-size=0 to disable tiling.
11133 2017-09-27 Richard Biener <rguenther@suse.de>
11135 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
11136 (graphite-max-nb-scop-params): Document special value zero.
11137 * domwalk.h (dom_walker::STOP): New symbolical constant.
11138 (dom_walker::dom_walker): Add optional parameter for bb to
11140 (dom_walker::~dom_walker): Declare.
11141 (dom_walker::before_dom_children): Document STOP return value.
11142 (dom_walker::m_user_bb_to_rpo): New member.
11143 (dom_walker::m_bb_to_rpo): Likewise.
11144 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
11145 mapping here if not provided by the user.
11146 (dom_walker::~dom_walker): Free bb to RPO mapping if not
11147 provided by the user.
11148 (dom_walker::STOP): Define.
11149 (dom_walker::walk): Do not compute bb to RPO mapping here.
11150 Support STOP return value from before_dom_children to stop
11152 * graphite-optimize-isl.c (optimize_isl): If the schedule
11153 is the same still generate code if -fgraphite-identity
11154 or -floop-parallelize-all are given.
11155 * graphite-scop-detection.c: Include cfganal.h.
11156 (gather_bbs::gather_bbs): Get and pass through bb to RPO
11158 (gather_bbs::before_dom_children): Return STOP for BBs
11160 (build_scops): Compute bb to RPO mapping and pass it to
11161 the domwalk. Treat --param graphite-max-nb-scop-params=0
11162 as not limiting the number of params.
11163 * graphite.c (graphite_initialize): Remove limit on the
11164 number of basic-blocks in a function.
11165 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
11166 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
11167 default value of 10.
11169 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11171 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
11172 Adjust code to eliminate needing to do the shift right 32-bits
11173 operation after XSCVDPSPN.
11175 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11177 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
11178 ((X / Y) != 0 -> X >= Y): Likewise.
11180 2017-09-26 Carl Love <cel@us.ibm.com>
11182 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
11183 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
11184 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
11185 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
11186 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
11187 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
11188 definitions and overloading.
11189 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
11190 statement for P9V_BUILTIN_XST_LEN_R.
11191 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
11192 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
11193 define_expand and define_insn for the instructions and builtins.
11194 * doc/extend.texi: Update the built-in documentation file for the new
11195 built-in functions.
11196 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
11197 define_insn for the instructions
11199 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
11202 * gcc/config/netbsd-protos.h: New file.
11203 * gcc/config/netbsd.c: New file.
11204 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
11205 * gcc/config/t-netbsd: New file.
11206 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
11207 (tmake_file) Add t-netbsd.
11208 (extra_objs) Add netbsd.o.
11210 2017-09-26 Janus Weil <janus@gcc.gnu.org>
11214 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
11216 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11218 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
11219 sign extension from a vector register to a GPR by doing a 32-bit
11220 direct move and then an EXTSW.
11221 (extendsi<mode>2 splitter): Likewise.
11222 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
11223 right or vector extract after doing XSCVDPSPN. Use
11224 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
11226 (movdi_from_sf_zero_ext): Likewise.
11227 (reload_gpr_from_vsxsf): Likewise.
11228 (p8_mfvsrd_4_disf): Delete, no longer used.
11229 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
11230 then needing to move the SFmode to a GPR to use the XSCVDPSP
11231 instruction instead of FRSP and XSCVDPSPN.
11232 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
11233 it is adjacent to the other XSCVSPDP insns.
11234 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
11235 SFmode to be in traditional Altivec registers.
11236 (vsx_xscvdpspn): Eliminate useless alternative constraint.
11237 (vsx_xscvspdpn): Likewise.
11238 (vsx_xscvspdpn_scalar): Likewise.
11240 2017-09-26 Martin Jambor <mjambor@suse.cz>
11242 * tree-sra.c (compare_access_positions): Put integral types first,
11243 stabilize sorting of integral types, remove conditions putting
11244 non-full-precision integers last.
11245 (sort_and_splice_var_accesses): Disable scalarization if a
11246 non-integert would be represented by a non-full-precision integer.
11248 2017-09-26 Joseph Myers <joseph@codesourcery.com>
11250 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
11251 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
11252 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
11253 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
11254 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
11255 conditionals inside the function instead of around it. Call
11256 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
11257 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
11259 2017-09-26 Richard Biener <rguenther@suse.de>
11261 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
11263 (scop_detection::build_scop_breadth): ... this. Removed.
11264 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
11265 (scop_detection::harmful_stmt_in_bb): Likewise.
11266 (scop_detection::graphite_can_represent_stmt): Likewise.
11267 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
11268 (scop_detection::can_represent_loop): Remove recursion, fold in ...
11269 (scop_detection::can_represent_loop_1): ... this. Removed.
11270 (scop_detection::harmful_loop_in_region): Simplify after inlining
11271 the above and remove more quadraticness.
11272 (build_scops): Adjust.
11273 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
11276 2017-09-26 Jakub Jelinek <jakub@redhat.com>
11279 * config/i386/i386.c (ix86_print_operand_address_as): Only test
11280 REGNO (base) == SP_REG if base is a REG.
11282 PR middle-end/35691
11283 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
11284 if it is different SSA_NAME.
11285 (optimize_range_tests_cmp_bitwise): New function.
11286 (optimize_range_tests): Call it.
11288 2017-09-26 Richard Biener <rguenther@suse.de>
11290 PR tree-optimization/82321
11291 * graphite.c (canonicalize_loop_closed_ssa): Properly check
11292 for the def being inside the loop.
11294 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11296 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
11298 * config/s390/s390-builtins.def: Fix constraint on op4.
11300 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11302 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
11303 independent expanders.
11304 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
11305 ("vec_ordered", "vec_unordered"): New expanders.
11307 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11309 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
11312 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11314 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
11315 vec_unpacks_lo_v16qi.
11316 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
11318 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11320 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
11321 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
11322 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
11324 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11326 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
11328 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
11330 ("*vec_slb<mode>"): New insn pattern.
11331 ("vec_shr_<mode>"): New expander.
11332 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
11333 and force the shift count operand to V16QImode.
11334 ("vec_srb<mode>"): Set shift count mode to V16QI.
11336 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11338 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
11339 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
11340 ("vec_widen_smult_hi_<mode>"): New expander definitions.
11342 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
11345 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
11347 2017-09-26 Richard Biener <rguenther@suse.de>
11349 PR tree-optimization/82320
11350 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
11353 2017-09-25 Jeff Law <law@redhat.com>
11355 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
11356 prototype for new argument.
11357 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
11358 mostly extracted from rs6000_emit_allocate_stack.
11359 (rs6000_emit_probe_stack_range_stack_clash): New function.
11360 (rs6000_emit_allocate_stack): Call
11361 rs6000_emit_probe_stack_range_stack_clash as needed.
11362 (rs6000_emit_probe_stack_range): Add additional argument
11363 to call to gen_probe_stack_range{si,di}.
11364 (output_probe_stack_range): New.
11365 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
11366 (output_probe_stack_range_stack_clash): New.
11367 (rs6000_emit_prologue): Emit notes into dump file as requested.
11368 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
11369 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
11370 Add additional operand and pass it to output_probe_stack_range.
11372 2017-09-25 Bin Cheng <bin.cheng@arm.com>
11374 PR tree-optimization/82163
11375 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
11376 (checking_verify_loop_closed_ssa): New parameter.
11377 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
11378 (check_loop_closed_ssa_stmt): Delete.
11379 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
11380 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
11381 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
11384 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
11386 * brig-builtins.def: Treat HSAIL barrier builtins as
11387 setjmp/longjump style functions.
11389 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
11391 * target.def (constant_alignment): New hook.
11392 * defaults.h (CONSTANT_ALIGNMENT): Delete.
11393 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
11394 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
11395 * doc/tm.texi: Regenerate.
11396 * targhooks.h (default_constant_alignment): Declare.
11397 (constant_alignment_word_strings): Likewise.
11398 * targhooks.c (default_constant_alignment): New function.
11399 (constant_alignment_word_strings): Likewise.
11400 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
11401 instead of CONSTANT_ALIGNMENT.
11402 * varasm.c (align_variable, get_variable_align, build_constant_desc)
11403 (force_const_mem): Likewise.
11404 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
11405 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
11406 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
11407 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11408 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
11410 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
11411 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11412 constant_alignment_word_strings.
11413 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
11414 (CONSTANT_ALIGNMENT): Likewise.
11415 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11416 (arm_constant_alignment): New function.
11417 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
11418 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11419 constant_alignment_word_strings.
11420 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
11421 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11422 constant_alignment_word_strings.
11423 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
11424 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11425 (cris_constant_alignment): New function.
11426 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
11427 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11428 (epiphany_constant_alignment): New function.
11429 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
11430 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11431 constant_alignment_word_strings.
11432 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
11433 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11434 constant_alignment_word_strings.
11435 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
11436 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11437 constant_alignment_word_strings.
11438 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
11439 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
11440 * config/i386/i386.c (ix86_constant_alignment): Make static.
11441 Use the same interface as the target hook.
11442 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11443 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
11444 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11445 constant_alignment_word_strings.
11446 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
11447 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
11448 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11449 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
11450 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11451 constant_alignment_word_strings.
11452 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
11453 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11454 constant_alignment_word_strings.
11455 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
11456 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11457 constant_alignment_word_strings.
11458 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
11459 * config/microblaze/microblaze.c (microblaze_constant_alignment):
11461 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11462 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
11463 * config/mips/mips.c (mips_constant_alignment): New function.
11464 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11465 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
11466 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
11467 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11468 (mmix_constant_alignment): Make static. Use the same interface
11469 as the target hook.
11470 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
11471 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11472 constant_alignment_word_strings.
11473 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
11474 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11475 constant_alignment_word_strings.
11476 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
11477 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11478 constant_alignment_word_strings.
11479 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
11480 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11481 (rs6000_constant_alignment): New function.
11482 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
11483 * config/riscv/riscv.c (riscv_constant_alignment): New function.
11484 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11485 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
11486 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11487 (rs6000_constant_alignment): New function.
11488 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
11489 * config/s390/s390.c (s390_constant_alignment): New function.
11490 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11491 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
11492 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11493 constant_alignment_word_strings.
11494 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
11495 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11496 (sparc_constant_alignment): New function.
11497 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
11498 * config/spu/spu.c (spu_constant_alignment): New function.
11499 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11500 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
11501 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11502 constant_alignment_word_strings.
11503 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
11504 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11505 constant_alignment_word_strings.
11506 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
11507 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11508 constant_alignment_word_strings.
11509 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
11510 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11511 (visium_constant_alignment): New function.
11512 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
11513 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11514 (xtensa_constant_alignment): New function.
11515 * system.h (CONSTANT_ALIGNMENT): Poison.
11517 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
11519 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11520 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
11521 (rs6000_builtin_valid_without_lhs): New helper function.
11522 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11523 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
11525 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
11527 * target.h (vec_perm_indices): Use unsigned short rather than
11529 (auto_vec_perm_indices): Likewise.
11530 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
11531 Use unsigned int rather than unsigned char.
11532 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
11534 2017-09-25 Richard Biener <rguenther@suse.de>
11536 * cfgloop.h (sort_sibling_loops): Declare.
11537 * cfgloop.c (sort_sibling_loops_cmp): New helper.
11538 (sort_sibling_loops): New function sorting the sibling loop list
11540 * graphite.c (graphite_transform_loops): Sort sibling loops.
11542 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
11544 * target.def (vec_perm_const_ok): Change sel parameter to
11546 * optabs-query.c (can_vec_perm_p): Update accordingly.
11547 * doc/tm.texi: Regenerate.
11548 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
11549 auto_vec_perm_indices and remove separate nelt field.
11550 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
11551 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
11552 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
11553 (aarch64_expand_vec_perm_const): Update accordingly.
11554 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
11555 to vec_perm_indices.
11556 * config/arm/arm.c (expand_vec_perm_d): Change perm to
11557 auto_vec_perm_indices and remove separate nelt field.
11558 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
11559 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
11560 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
11562 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
11563 to vec_perm_indices.
11564 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
11565 sel to vec_perm_indices.
11566 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
11567 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
11568 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
11570 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
11573 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
11576 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
11577 on the FUNCTION_DECL function context if it has a DIE that is a
11580 2017-09-25 Richard Biener <rguenther@suse.de>
11582 PR tree-optimization/82285
11583 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
11586 2017-09-25 Tom de Vries <tom@codesourcery.com>
11590 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
11593 2017-09-25 Richard Biener <rguenther@suse.de>
11595 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
11596 ISL errors other than isl_error_quota happen. Dump if the
11597 schedule is the same.
11598 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
11599 errors instead of aborting inside ISL.
11601 2017-09-25 Iain Sandoe <iain@codesourcery.com>
11604 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
11605 of libgcc_eh for m64.
11606 * config/i386/darwin64.h: Likewise.
11608 2017-09-25 Richard Biener <rguenther@suse.de>
11610 PR middle-end/82144
11611 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
11612 attribute for incomplete types nor twice for complete ones.
11614 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
11617 * config/i386/i386.c (ix86_print_operand_address_as): Encode
11618 %esp as %rsp to avoid 0x67 prefix if there is no index or base
11621 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
11624 * config/i386/i386.opt (mprefer-avx256): Use
11625 ix86_target_flags variable.
11626 * config/i386/i386.c (ix86_target_string): Move
11627 -mprefer-avx256 to flag2_opts.
11629 2017-09-22 Jakub Jelinek <jakub@redhat.com>
11631 PR middle-end/35691
11632 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
11633 and x != -1 | y != -1 into (x & y) != -1.
11635 2017-09-22 Steve Ellcey <sellcey@cavium.com>
11637 * config.gcc: Add new case statement to set
11638 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
11639 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
11640 s390x-*-linux* case statements. Added aarch64 to the list of
11641 supported architectures.
11643 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11645 PR tree-optimization/82289
11646 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
11647 STMT_VINFO_RELEVANT_P.
11649 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11650 Alan Hayward <alan.hayward@arm.com>
11651 David Sherwood <david.sherwood@arm.com>
11653 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
11654 for VR_RANGE only; don't allow VR_ANTI_RANGE.
11655 (extract_range_from_binary_expr_1): Don't call
11656 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
11658 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11659 Alan Hayward <alan.hayward@arm.com>
11660 David Sherwood <david.sherwood@arm.com>
11662 * target.def (preferred_vector_alignment): New hook.
11663 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
11665 * doc/tm.texi: Regenerate.
11666 * targhooks.h (default_preferred_vector_alignment): Declare.
11667 * targhooks.c (default_preferred_vector_alignment): New function.
11668 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
11670 (DR_TARGET_ALIGNMENT): New macro.
11671 (aligned_access_p): Update commentary.
11672 (vect_known_alignment_in_bytes): New function.
11673 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
11675 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
11676 Calculate the misalignment based on the target alignment rather than
11678 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
11679 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
11680 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
11681 the target alignment, rather than masking the element misalignment
11682 with the number of elements in a vector. Also use the target
11683 alignment when calculating the maximum number of peels.
11684 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
11685 instead of TYPE_ALIGN_UNIT.
11686 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
11687 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
11688 (vect_create_addr_base_for_vector_ref): Update call accordingly.
11689 (vect_create_data_ref_ptr): Likewise.
11690 (vect_setup_realignment): Realign by ANDing with
11691 -DR_TARGET_MISALIGNMENT.
11692 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
11693 the number of peels based on DR_TARGET_ALIGNMENT.
11694 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
11695 with the guaranteed alignment boundary when deciding whether
11697 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
11698 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
11699 (ensure_base_align): Remove stmt_info parameter. Get the
11700 target base alignment from DR_TARGET_ALIGNMENT.
11701 (vectorizable_store): Update call accordingly. Interpret
11702 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
11704 (vectorizable_load): Likewise.
11706 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11707 Alan Hayward <alan.hayward@arm.com>
11708 David Sherwood <david.sherwood@arm.com>
11710 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
11711 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
11712 (vect_enhance_data_refs_alignment): Likewise.
11714 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11716 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
11717 error. Only quit immediately if parsing is complete.
11718 (BEGIN): Initialize fatal_err and parse_done.
11719 (begin fpu, end fpu): Check number of arguments.
11720 (begin arch, end arch): Likewise.
11721 (begin cpu, end cpu): Likewise.
11722 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
11723 (optalias): Likewise.
11725 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11727 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
11728 * config/arm/arm-isa.h: Delete. Move definitions to ...
11729 * arm-cpus.in: ... here. Use new feature and fgroup values.
11730 * config/arm/arm.c (arm_option_override): Use lower case for feature
11732 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
11733 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
11734 * config/arm/parsecpu.awk (END): Add new command 'isa'.
11736 (print_isa_bits_for): New function.
11737 (gen_isa): New function.
11738 (gen_comm_data): Use print_isa_bits_for.
11739 (define feature): New keyword.
11740 (define fgroup): New keyword.
11741 * config/arm/t-arm (TM_H): Remove.
11742 (GTM_H): Add arm-isa.h.
11743 (arm-isa.h): Add rule to generate file.
11744 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
11745 case for feature bit names.
11747 2017-09-22 Richard Biener <rguenther@suse.de>
11749 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
11751 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
11752 print of no dependency loops ...
11753 * graphite.c (graphite_transform_loops): ... here.
11754 (canonicalize_loop_closed_ssa_form): Work from inner to outer
11756 (same_close_phi_node, remove_duplicate_close_phi,
11757 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
11758 (canonicalize_loop_closed_ssa): ... here and simplify.
11759 * graphite-optimize-isl.c: Include tree-vectorizer.h.
11760 (optimize_isl): Use dump_printf_loc to tell when we stopped
11761 optimizing because of an ISL timeout.
11763 2017-09-22 Richard Biener <rguenther@suse.de>
11765 PR tree-optimization/82291
11766 * tree-if-conv.c (predicate_mem_writes): Make sure to
11767 remove writes in blocks predicated with false.
11769 2017-09-22 Richard Biener <rguenther@suse.de>
11771 * sese.c: Include cfganal.h.
11772 (if_region_set_false_region): Remove.
11773 (create_if_region_on_edge): Likewise.
11774 (move_sese_in_condition): Re-implement without destroying
11777 2017-09-22 Richard Biener <rguenther@suse.de>
11779 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
11780 Verify both BBs contain loop PHI nodes before dispatching to
11781 copy_loop_phi_args.
11782 (graphite_regenerate_ast_isl): Do not recompute dominators,
11783 do not verify three times. Restructure for clarity.
11784 * graphite-scop-detection.c (same_close_phi_node,
11785 remove_duplicate_close_phi, make_close_phi_nodes_unique,
11786 defined_in_loop_p, canonicalize_loop_closed_ssa,
11787 canonicalize_loop_closed_ssa_form): Simplify, remove excess
11788 checking and SSA rewrite, move to ...
11789 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
11790 (graphite_initialize): Do not pass in ctx, do not reset the
11791 SCEV cache, compute only dominators.
11792 (graphite_transform_loops): Allocate ISL ctx after
11793 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
11794 Maintain post-dominators only around build_scops.
11795 * sese.c (if_region_set_false_region): Make static. Free
11796 and recompute dominators.
11797 (move_sese_in_condition): Assert we don't get called with
11798 post-dominators computed.
11799 * sese.h (if_region_set_false_region): Remove.
11801 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
11803 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
11804 mode attribute for TARGET_AVX512VL.
11806 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
11808 * config/i386/i386.opt (mprefer-avx256): New option.
11809 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
11811 (ix86_preferred_simd_mode): Return 256-bit AVX modes
11812 for TARGET_PREFER_AVX256.
11813 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
11815 2017-09-21 Jeff Law <law@redhat.com>
11817 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
11818 Fix dump output if the only stack space is for pushed registers.
11820 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11822 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
11825 2017-09-21 Martin Sebor <msebor@redhat.com>
11828 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
11829 code (in C++) or code that triggers warnings.
11831 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
11833 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
11835 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
11837 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
11838 * sched-rgn.c: Ditto.
11839 * sel-sched-ir.c: Ditto.
11841 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
11843 * toplev.h (set_random_seed): Adjust return type.
11844 * toplev.c (init_local_tick): Move eager initialization of random_seed
11845 to get_random_seed. Adjust comment.
11846 (init_random_seed): Inline to get_random_seed, delete.
11847 (get_random_seed): Initialize random_seed lazily.
11848 (set_random_seed): Do not return previous value.
11849 (print_switch_value): Do not call get_random_seed.
11851 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
11853 * cgraph.c (delete_function_version): New, broken out from...
11854 (cgraph_node::delete_function_version): ...here. Rename to
11855 cgraph_node::delete_function_version_by_decl. Update all uses.
11856 (cgraph_node::remove): Call delete_function_version.
11858 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11861 * tree-inline.c (expand_call_inline): Emit clobber stmts for
11862 VAR_DECLs to which addressable non-volatile parameters are mapped
11863 and for id->retvar after the return value assignment. Clear
11864 id->retval and id->retbnd after inlining.
11866 2017-09-21 Richard Biener <rguenther@suse.de>
11868 PR tree-optimization/82276
11869 PR tree-optimization/82244
11870 * tree-vrp.c (build_assert_expr_for): Set
11871 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
11873 (remove_range_assertions): Revert earlier change.
11875 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
11878 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
11880 2017-09-21 Richard Biener <rguenther@suse.de>
11882 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11883 Restore valid IL after code generation errors.
11884 * graphite.c (graphite_transform_loops): Diagnose code
11885 generation issues as MSG_MISSED_OPTIMIZATION and continue
11886 with processing SCOPs.
11888 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11889 Alan Hayward <alan.hayward@arm.com>
11890 David Sherwood <david.sherwood@arm.com>
11892 * calls.c (compute_argument_addresses): Use simplify_gen_binary
11893 rather than choosing between plus_constant and gen_rtx_<CODE>.
11894 * expr.c (emit_push_insn): Likewise.
11895 (expand_expr_real_2): Likewise.
11897 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11898 Alan Hayward <alan.hayward@arm.com>
11899 David Sherwood <david.sherwood@arm.com>
11901 * loop-unroll.c (split_iv): Call copy_rtx on the step.
11903 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11904 Alan Hayward <alan.hayward@arm.com>
11905 David Sherwood <david.sherwood@arm.com>
11907 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
11908 calling tree_to_uhwi.
11910 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11911 Alan Hayward <alan.hayward@arm.com>
11912 David Sherwood <david.sherwood@arm.com>
11914 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
11915 INTEGER_CST rather than a negative test for ADDR_EXPR.
11917 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11918 Alan Hayward <alan.hayward@arm.com>
11919 David Sherwood <david.sherwood@arm.com>
11921 * tree-vrp.c (extract_range_from_binary_expr_1): Check
11922 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
11924 2017-09-21 Richard Biener <rguenther@suse.de>
11926 PR tree-optimization/71351
11927 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
11928 graphite_create_new_loop_guard): Remove, fold remaining parts
11930 (translate_isl_ast_node_for): ... here and simplify.
11932 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11935 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
11936 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
11937 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
11938 alternative always use QI mode, for -Os imov (=R,R) alternative
11939 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
11942 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11943 Jeff Law <law@redhat.com>
11945 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
11946 (allocate_stack_space): New function, partially extracted from
11947 s390_emit_prologue.
11948 (s390_emit_prologue): Track offset to most recent stack probe.
11949 Code to allocate space moved into allocate_stack_space.
11950 Dump actions when no stack is allocated.
11951 (s390_prologue_plus_offset): New function.
11952 (s390_emit_stack_probe): Likewise.
11954 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
11956 * common.opt (Wa, Wl, Wp, g, gz=): Add
11958 (gno-column-info): Remove.
11959 (gcolumn-info): Drop RejectNegative.
11960 (gno-): New prefix.
11961 (gno-record-gcc-switches): Remove.
11962 (grecord-gcc-switches): Drop RejectNegative.
11963 (gno-split-dwarf): Remove.
11964 (gsplit-dwarf): Drop RejectNegative.
11965 (gno-strict-dwarf): Remove.
11966 (gstrict-dwarf): Drop RejectNegative.
11967 * config/darwin.opt (gfull, gused): Add RejectNegative.
11968 * dwarf2out.c (gen_producer_string): Drop
11969 gno-record-gcc-switches handler.
11970 * optc-gen.awk: Add g to prefixes with negative forms.
11971 * opts-common.c (remapping_prefix_p): New.
11972 (find_opt): Check it.
11973 (generate_canonical_option): Test g prefix.
11974 (option_map): Add -gno- mapping.
11975 (add_misspelling_candidates): Check remapping_prefix_p.
11977 2017-09-20 Jeff Law <law@redhat.com>
11979 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
11980 thinko in stack clash protection support.
11982 * explow.c (compute_stack_clash_protection_loop_data): Use
11983 CONST_INT_P instead of explicit test. Verify object is a
11984 CONST_INT_P before looking at INTVAL.
11985 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
11986 instead of explicit test.
11988 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
11991 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
11992 address instead of to r1 and r11.
11994 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
11996 * config.gcc: Support "knm".
11997 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
11998 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
12000 * config/i386/i386.c (m_KNM): Define.
12001 (processor_target_table): Add "knm".
12003 (ix86_option_override_internal): Add "knm".
12004 (ix86_issue_rate): Add PROCESSOR_KNM.
12005 (ix86_adjust_cost): Ditto.
12006 (ia32_multipass_dfa_lookahead): Ditto.
12007 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
12008 (fold_builtin_cpu): Add M_INTEL_KNM.
12009 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
12010 (processor_type): Add PROCESSOR_KNM.
12011 * config/i386/x86-tune.def: Add m_KNM.
12012 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
12014 2017-09-20 Richard Biener <rguenther@suse.de>
12016 PR tree-optimization/80213
12017 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
12018 are allowed in empty BBs as well.
12019 (canonicalize_loop_closed_ssa): Also look for other complex
12021 (scop_detection::get_sese): Include the loop-closed PHI block
12023 (scop_detection::merge_sese): Remove code adding extra blocks.
12024 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
12025 (build_scops): Assert the final returned scop is invalid.
12027 2017-09-20 Richard Biener <rguenther@suse.de>
12029 PR tree-optimization/82264
12030 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
12032 (vn_phi_lookup): Likewise.
12033 (vn_phi_insert): Likewise.
12035 2017-09-20 Jakub Jelinek <jakub@redhat.com>
12037 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
12038 that fits into uhwi or shwi, add DW_AT_const_value regardless
12039 of early_dwarf without going through RTL, using add_AT_unsigned
12042 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
12043 (DEBUG_LTO_ABBREV_SECTION): Likewise.
12044 (DEBUG_LTO_MACINFO_SECTION): Likewise.
12045 (DEBUG_MACRO_SECTION): Likewise.
12046 (DEBUG_LTO_MACRO_SECTION): Likewise.
12047 (DEBUG_STR_DWO_SECTION): Likewise.
12048 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
12049 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
12050 (DEBUG_LTO_DWO_LINE_SECTION): Define.
12051 (DEBUG_LTO_LINE_STR_SECTION): Define.
12052 (init_sections_and_labels): Initialize debug_line_str_section
12053 variable. Initialize debug_loc_section for -gdwarf-5 to
12054 DEBUG_LOCLISTS_SECTION. Formatting fixes.
12056 2017-09-20 Richard Biener <rguenther@suse.de>
12058 * graphite-sese-to-poly.c (extract_affine): Properly handle
12059 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
12061 2017-09-20 Richard Biener <rguenther@suse.de>
12063 PR tree-optimization/81373
12064 * graphite-scop-detection.c (build_cross_bb_scalars_def):
12065 Force SESE live-out defs to be handled even if they are
12068 2017-09-19 Jeff Law <law@redhat.com>
12070 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
12071 nothing for stack adjustments with REG_STACK_CHECK.
12072 * sched-deps.c (parse_add_or_inc): Reject insns with
12073 REG_STACK_CHECK from dependency breaking.
12074 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
12075 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
12076 * reg-notes.def (STACK_CHECK): New note.
12078 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
12079 (ix86_expand_prologue): Dump stack clash info as needed.
12080 Call ix86_adjust_stack_and_probe_stack_clash as needed.
12082 * function.c (dump_stack_clash_frame_info): New function.
12083 * function.h (dump_stack_clash_frame_info): Prototype.
12084 (enum stack_clash_probes): New enum.
12086 * config/alpha/alpha.c (alpha_expand_prologue): Also check
12087 flag_stack_clash_protection.
12088 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
12089 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
12090 (arm_frame_pointer_required): Likewise.
12091 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
12092 (ia64_expand_prologue): Likewise.
12093 * config/mips/mips.c (mips_expand_prologue): Likewise.
12094 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
12095 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
12096 (sparc_flat_expand_prologue): Likewise.
12097 * config/spu/spu.c (spu_expand_prologue): Likewise.
12099 * explow.c: Include "params.h".
12100 (anti_adjust_stack_and_probe_stack_clash): New function.
12101 (get_stack_check_protect): Likewise.
12102 (compute_stack_clash_protection_loop_data): Likewise.
12103 (emit_stack_clash_protection_loop_start): Likewise.
12104 (emit_stack_clash_protection_loop_end): Likewise.
12105 (allocate_dynamic_stack_space): Use get_stack_check_protect.
12106 Use anti_adjust_stack_and_probe_stack_clash.
12107 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
12108 (emit_stack_clash_protection_loop_start): Likewise.
12109 (emit_stack_clash_protection_loop_end): Likewise.
12110 * rtl.h (get_stack_check_protect): Prototype.
12111 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
12112 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
12113 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
12115 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
12117 * doc/tm.texi: Rebuilt.
12118 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
12119 get_stack_check_protect.
12120 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
12121 * config/arm/arm.c (arm_expand_prologue): Likewise.
12122 (arm_frame_pointer_required): Likewise.
12123 * config/i386/i386.c (ix86_expand_prologue): Likewise.
12124 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
12125 * config/mips/mips.c (mips_expand_prologue): Likewise.
12126 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
12127 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
12128 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
12129 (sparc_flat_expand_prologue): Likewise.
12131 * common.opt (-fstack-clash-protection): New option.
12132 * flag-types.h (enum stack_check_type): Note difference between
12133 -fstack-check= and -fstack-clash-protection.
12134 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
12135 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
12136 * toplev.c (process_options): Issue warnings/errors for cases
12137 not handled with -fstack-clash-protection.
12138 * doc/invoke.texi (-fstack-clash-protection): Document new option.
12139 (-fstack-check): Note additional problem with -fstack-check=generic.
12140 Note that -fstack-check is primarily for Ada and refer users
12141 to -fstack-clash-protection for stack-clash-protection.
12142 Document new params for stack clash protection.
12144 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
12146 * config/i386/i386.c (ix86_split_long_move): Do not handle
12147 address used for LEA in a special way.
12149 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
12151 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
12153 2017-09-19 Martin Sebor <msebor@redhat.com>
12156 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
12157 of incompatible types.
12159 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
12161 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12162 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
12163 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12164 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
12166 2017-09-19 Richard Biener <rguenther@suse.de>
12168 PR tree-optimization/82244
12169 * tree-vrp.c (remove_range_assertions): Do not propagate
12170 a constant to abnormals but replace the assert with a copy.
12172 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
12174 PR rtl-optimization/57878
12175 PR rtl-optimization/68988
12176 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
12177 avoidance test involving non_reload_pseudos. Move frequency test
12178 below the general fragmentation avoidance test.
12180 2017-09-19 Richard Biener <rguenther@suse.de>
12182 PR tree-optimization/69728
12183 * graphite-sese-to-poly.c (schedule_error): New global.
12184 (add_loop_schedule): Handle empty domain by failing the
12186 (build_original_schedule): Handle schedule_error.
12188 2017-09-19 Richard Biener <rguenther@suse.de>
12190 * graphite-scop-detection.c (scop_detection::can_represent_loop):
12191 Do not iterate to sibling loops but only to siblings of inner
12194 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
12197 * config/m68k/m68k.md (moveq feeding equality comparison): Check
12198 that the registers are different.
12200 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
12202 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
12203 to processor_model and "amdfam17h" to arch_names_table.
12204 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
12206 2017-09-18 Jakub Jelinek <jakub@redhat.com>
12209 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
12211 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12212 Alan Hayward <alan.hayward@arm.com>
12213 David Sherwood <david.sherwood@arm.com>
12215 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
12217 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
12219 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
12220 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
12221 vect_schedule_slp_instance.
12222 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
12223 Update call to vect_slp_analyze_node_operations. Simplify return
12225 (vect_slp_analyze_bb_1): Update call accordingly.
12226 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
12227 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
12228 (vect_schedule_slp): Update call accordingly.
12230 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12231 Alan Hayward <alan.hayward@arm.com>
12232 David Sherwood <david.sherwood@arm.com>
12234 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
12235 with types that aren't in fact scalar.
12237 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12239 * tree-vect-slp.c (vect_record_max_nunits): New function,
12241 (vect_build_slp_tree_1): ...here.
12242 (vect_build_slp_tree_2): Call it for phis too.
12244 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12246 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
12247 to vect_get_vec_def_for_operand when getting the mask operand.
12249 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12250 Alan Hayward <alan.hayward@arm.com>
12251 David Sherwood <david.sherwood@arm.com>
12253 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
12256 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12257 Alan Hayward <alan.hayward@arm.com>
12258 David Sherwood <david.sherwood@arm.com>
12260 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
12261 calculation for vector booleans.
12263 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12264 Alan Hayward <alan.hayward@arm.com>
12265 David Sherwood <david.sherwood@arm.com>
12267 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
12269 (vect_transform_stmt): ...here.
12270 (vect_analyze_stmt): Use it instead of calling
12271 vectorizable_live_operation directly.
12273 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
12275 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
12276 non-SIMT targets in acc vector loops.
12278 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
12280 * configure.ac: Add arc and check if assembler supports gdwarf2.
12281 * configure: Regenerate.
12283 2017-09-18 Richard Biener <rguenther@suse.de>
12285 PR tree-optimization/82220
12286 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
12287 epilogue niters from the min_profitable_iters compute.
12289 2017-09-18 Jakub Jelinek <jakub@redhat.com>
12292 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
12293 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
12294 (ix86_init_pic_reg): Revert 2017-09-01 changes.
12296 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
12299 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
12300 switching to a new text section.
12302 2017-09-18 Richard Biener <rguenther@suse.de>
12304 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
12306 (build_alias_set): Reject aliases with no access function.
12308 2017-09-18 Richard Biener <rguenther@suse.de>
12310 PR tree-optimization/79622
12311 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
12313 (build_cross_bb_scalars_use): Likewise.
12315 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
12317 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
12319 2017-09-18 Alan Modra <amodra@gmail.com>
12322 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
12323 stack_pointer_rtx for count 0. Update comments. Break up
12324 large rtl expression.
12326 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
12328 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
12329 Increase to 20 bytes.
12330 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
12331 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
12332 or avx version of the stub.
12334 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
12337 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
12338 compute the minimum stack alignment. Also update preferred stack
12339 boundary for leaf functions.
12341 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
12343 PR tree-optimization/82228
12344 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
12347 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
12349 * common/config/nds32/nds32-common.c
12350 (nds32_option_optimization_table): Refine formatting.
12351 (nds32_option_optimization_table): Use -fsched-pressure and
12352 -fomit-frame-pointer for specific optimization level.
12354 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
12356 * config/nds32/nds32.c: Refine formatting and comments.
12357 * config/nds32/nds32.h: Likewise.
12358 * config/nds32/nds32.md: Likewise.
12359 * config/nds32/nds32-cost.c: Likewise.
12360 * config/nds32/nds32-isr.c: Likewise.
12361 * config/nds32/nds32-md-auxiliary.c: Likewise.
12362 * config/nds32/nds32-multiple.md: Likewise.
12363 * config/nds32/nds32-predicates.c: Likewise.
12365 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
12366 Jakub Jelinek <jakub@redhat.com>
12368 Add support for -std=c++2a.
12369 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
12371 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
12373 2017-09-15 Steve Ellcey <sellcey@cavium.com>
12376 * doc/extend.texi (Common Function Attributes): Add
12377 references to ARM, AArch64, and S/390 specific attributes.
12378 (Function Specific Option Pragmas): Add AArch64 and S/390
12379 to list of back ends that support the target pragma.
12381 2017-09-15 Nathan Sidwell <nathan@acm.org>
12383 * doc/standards.texi: Fix C++17 description. Update URLs for
12386 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
12388 * common.opt (Wcast-align=strict): New warning option.
12389 * doc/invoke.texi: Document -Wcast-align=strict.
12391 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
12393 * cgraph.h (cgraph_thunk_info): Add comments.
12394 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
12395 assert for VIRTUAL_* arguments stricter.
12397 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
12399 PR tree-optimization/71026
12400 * match.pd: Move RDIV patterns from fold-const.c
12401 * fold-const.c (distribute_real_division): Removed.
12402 (fold_binary_loc): Remove calls to distribute_real_divison.
12404 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12406 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
12407 c++1z and gnu++1z as deprecated. Change other references to
12408 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
12409 Change -Wc++1z-compat to -Wc++17-compat.
12410 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
12411 * dwarf2out.c (highest_c_language): Handle C++17.
12412 (gen_compile_unit_die): Likewise.
12414 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12416 PR rtl-optimization/82192
12417 * combine.c (make_extraction): Don't look through non-paradoxical
12418 SUBREGs or TRUNCATE if pos + len is or might be bigger than
12421 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
12422 Alan Hayward <alan.hayward@arm.com>
12423 David Sherwood <david.sherwood@arm.com>
12425 * target.def (function_arg_offset): New hook.
12426 * targhooks.h (default_function_arg_offset): Declare.
12427 * targhooks.c (default_function_arg_offset): New function.
12428 * function.c (locate_and_pad_parm): Use
12429 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
12430 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
12431 (TARGET_FUNCTION_ARG_OFFSET): ...this.
12432 * doc/tm.texi: Regenerate.
12433 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
12434 * config/spu/spu.c (spu_function_arg_offset): New function.
12435 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
12436 * system.h (FUNCTION_ARG_OFFSET): Poison.
12438 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
12439 Alan Hayard <alan.hayward@arm.com>
12440 David Sherwood <david.sherwood@arm.com>
12442 * target.def (truly_noop_truncation): New hook.
12443 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
12444 than TRULY_NOOP_TRUNCATION.
12445 * hooks.h (hook_bool_uint_uint_true): Declare.
12446 * hooks.c (hook_bool_uint_uint_true): New function.
12447 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
12448 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
12449 * doc/tm.texi: Regenerate.
12450 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
12451 rather than TRULY_NOOP_TRUNCATION in comments.
12452 (simplify_comparison): Likewise.
12453 (record_truncated_value): Likewise.
12454 * expmed.c (extract_bit_field_1): Likewise.
12455 (extract_split_bit_field): Likewise.
12456 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
12457 instead of TRULY_NOOP_TRUNCATION.
12458 * function.c (assign_parm_setup_block): Likewise.
12459 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
12460 * rtlhooks.c: Include target.h.
12461 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
12462 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
12463 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
12464 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
12465 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
12466 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
12467 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
12468 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
12469 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
12470 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
12471 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
12472 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
12473 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
12474 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
12475 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
12476 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
12477 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
12478 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
12479 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
12480 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
12481 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
12482 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
12483 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
12484 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
12485 * config/mips/mips.c (mips_truly_noop_truncation): New function.
12486 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
12487 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
12488 rather than TRULY_NOOP_TRUNCATION in comments.
12489 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
12490 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
12491 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
12492 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
12493 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
12494 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
12495 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
12496 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
12497 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
12498 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
12499 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
12500 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
12501 rather than TRULY_NOOP_TRUNCATION in comments.
12502 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
12503 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
12504 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
12505 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
12506 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
12507 TRULY_NOOP_TRUNCATION condition.
12508 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
12509 (TRULY_NOOP_TRUNCATION): Delete.
12510 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
12511 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
12512 * config/spu/spu.c (spu_truly_noop_truncation): New function.
12513 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
12514 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
12515 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
12516 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
12517 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
12518 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
12519 rather than TRULY_NOOP_TRUNCATION in comments.
12520 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
12521 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
12522 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
12523 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
12524 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
12525 * system.h (TRULY_NOOP_TRUNCATION): Poison.
12527 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
12530 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
12531 (*cmp_ior): Likewise.
12532 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
12533 (*ior_scc_scc_cmp): Likewise.
12534 (*and_scc_scc): Likewise.
12535 (*and_scc_scc_cmp): Likewise.
12537 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
12538 Alan Hayard <alan.hayward@arm.com>
12539 David Sherwood <david.sherwood@arm.com>
12541 * target.def (can_change_mode_class): New hook.
12542 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
12543 (hard_regno_nregs): Likewise.
12544 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
12545 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
12546 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
12547 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
12548 (LOAD_EXTEND_OP): Update accordingly.
12549 * doc/tm.texi: Regenerate.
12550 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
12551 CANNOT_CHANGE_MODE_CLASS.
12552 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
12553 (REG_CAN_CHANGE_MODE_P): ...this new macro.
12554 * combine.c (simplify_set): Update accordingly.
12555 * emit-rtl.c (validate_subreg): Likewise.
12556 * recog.c (general_operand): Likewise.
12557 * regcprop.c (mode_change_ok): Likewise.
12558 * reload1.c (choose_reload_regs): Likewise.
12559 (inherit_piecemeal_p): Likewise.
12560 * rtlanal.c (simplify_subreg_regno): Likewise.
12561 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
12562 instead of CANNOT_CHANGE_MODE_CLASS.
12563 (reload_cse_simplify_operands): Likewise.
12564 * reload.c (push_reload): Use targetm.can_change_mode_class
12565 instead of CANNOT_CHANGE_MODE_CLASS.
12566 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
12567 REG_CANNOT_CHANGE_MODE_P.
12568 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12569 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
12570 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12571 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12572 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12573 (arm_can_change_mode_class): New function.
12574 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
12575 than CANNOT_CHANGE_MODE_CLASS in comments.
12576 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12577 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
12578 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
12579 (ix86_can_change_mode_class): ...this new function, inverting the
12580 sense of the return value.
12581 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12582 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12583 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12584 (ia64_can_change_mode_class): New function.
12585 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12586 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
12587 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
12588 (m32c_can_change_mode_class): ...this new function, inverting the
12589 sense of the return value.
12590 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12591 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12592 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
12593 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
12594 (mips_can_change_mode_class): ...this new function, inverting the
12595 sense of the return value.
12596 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12597 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12598 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12599 (msp430_can_change_mode_class): New function.
12600 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12601 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
12602 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12603 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12604 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12605 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
12606 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12607 (pa_cannot_change_mode_class): Replace with...
12608 (pa_can_change_mode_class): ...this new function, inverting the
12609 sense of the return value.
12610 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
12611 than CANNOT_CHANGE_MODE_CLASS in comments.
12612 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12613 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
12614 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12615 (pdp11_cannot_change_mode_class): Replace with...
12616 (pdp11_can_change_mode_class): ...this new function, inverting the
12617 sense of the return value.
12618 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12619 * config/powerpcspe/powerpcspe-protos.h
12620 (rs6000_cannot_change_mode_class_ptr): Delete.
12621 * config/powerpcspe/powerpcspe.c
12622 (rs6000_cannot_change_mode_class_ptr): Delete.
12623 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12624 (rs6000_option_override_internal): Assign to
12625 targetm.can_change_mode_class instead of
12626 rs6000_cannot_change_mode_class_ptr.
12627 (rs6000_cannot_change_mode_class): Replace with...
12628 (rs6000_can_change_mode_class): ...this new function, inverting the
12629 sense of the return value.
12630 (rs6000_debug_cannot_change_mode_class): Replace with...
12631 (rs6000_debug_can_change_mode_class): ...this new function.
12632 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12633 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
12634 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12635 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12636 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
12638 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
12639 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12640 (rs6000_option_override_internal): Assign to
12641 targetm.can_change_mode_class instead of
12642 rs6000_cannot_change_mode_class_ptr.
12643 (rs6000_cannot_change_mode_class): Replace with...
12644 (rs6000_can_change_mode_class): ...this new function, inverting the
12645 sense of the return value.
12646 (rs6000_debug_cannot_change_mode_class): Replace with...
12647 (rs6000_debug_can_change_mode_class): ...this new function.
12648 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12649 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
12650 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
12651 (s390_can_change_mode_class): ...this new function, inverting the
12652 sense of the return value.
12653 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12654 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12655 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
12656 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12657 (sh_cannot_change_mode_class): Replace with...
12658 (sh_can_change_mode_class): ...this new function, inverting the
12659 sense of the return value.
12660 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12661 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12662 (sparc_can_change_mode_class): New function.
12663 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12664 * config/spu/spu.c (spu_can_change_mode_class): New function.
12665 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12666 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12667 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12668 (visium_can_change_mode_class): New function.
12669 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
12671 2017-09-15 Richard Biener <rguenther@suse.de>
12673 PR tree-optimization/82217
12674 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
12675 but not undefined case.
12677 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12680 * postreload.c (reload_cse_simplify_operands): Skip
12681 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
12683 2017-09-15 Richard Biener <rguenther@suse.de>
12685 PR tree-optimization/68823
12686 * graphite-scop-detection.c (build_alias_set): If we have a
12687 possible dependence check whether we can handle them by just
12688 looking at the DRs DR_ACCESS_FNs.
12689 (build_scops): If build_alias_set fails, fail the SCOP.
12691 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
12693 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
12694 to support float128 built-in functions that require the ISA 3.0
12696 (BU_FLOAT128_3_HW): Likewise.
12697 (SQRTF128): Add support for the IEEE 128-bit square root and fma
12698 built-in functions.
12699 (FMAF128): Likewise.
12701 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
12702 support for built-in functions that need the ISA 3.0 IEEE 128-bit
12703 floating point instructions.
12704 (rs6000_invalid_builtin): Likewise.
12705 (rs6000_builtin_mask_names): Likewise.
12706 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
12707 (RS6000_BTM_FLOAT128_HW): Likewise.
12708 (RS6000_BTM_COMMON): Likewise.
12709 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
12711 * doc/extend.texi (RS/6000 built-in functions): Document the
12712 IEEE 128-bit floating point square root and fused multiply-add
12713 built-in functions.
12715 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
12717 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
12718 reg (r2) isn't in the set of registers defined in the prologue.
12720 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12721 Alan Hayward <alan.hayward@arm.com>
12722 David Sherwood <david.sherwood@arm.com>
12724 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
12725 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
12726 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
12727 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
12728 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
12730 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
12731 max_vectorization_factor.
12732 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
12734 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12735 Alan Hayward <alan.hayward@arm.com>
12736 David Sherwood <david.sherwood@arm.com>
12738 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
12739 (vect_worthwhile_without_simd_p): Declare.
12740 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
12741 (vectorizable_reduction): Use it.
12742 * tree-vect-stmts.c (vectorizable_shift): Likewise.
12743 (vectorizable_operation): Likewise.
12745 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12746 Alan Hayward <alan.hayward@arm.com>
12747 David Sherwood <david.sherwood@arm.com>
12749 * tree-vectorizer.h (vect_get_num_copies): New function.
12750 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
12751 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12752 (vectorizable_induction): Likewise.
12753 (vectorizable_live_operation): Likewise.
12754 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12755 (vectorizable_bswap): Likewise.
12756 (vectorizable_call): Likewise.
12757 (vectorizable_conversion): Likewise.
12758 (vectorizable_assignment): Likewise.
12759 (vectorizable_shift): Likewise.
12760 (vectorizable_operation): Likewise.
12761 (vectorizable_store): Likewise.
12762 (vectorizable_load): Likewise.
12763 (vectorizable_condition): Likewise.
12764 (vectorizable_comparison): Likewise.
12765 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
12767 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12768 Alan Hayward <alan.hayward@arm.com>
12769 David Sherwood <david.sherwood@arm.com>
12771 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
12772 of vect_init_vector.
12774 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12775 Alan Hayward <alan.hayward@arm.com>
12776 David Sherwood <david.sherwood@arm.com>
12778 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
12779 an inline wrapper that provides a location.
12780 (gimple_build_vector): Likewise.
12781 * gimple-fold.c (gimple_build_vector_from_val): New function.
12782 (gimple_build_vector): Likewise.
12783 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
12784 functions to build the initial value. Always return a gimple value.
12785 (get_initial_defs_for_reduction): Likewise. Only compute
12787 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
12788 vect_init_vector on the results from get_initial_def(s)_for_reduction.
12789 (vectorizable_induction): Use gimple_build_vector rather than
12792 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12793 Alan Hayward <alan.hayward@arm.com>
12794 David Sherwood <david.sherwood@arm.com>
12796 * target.h (vec_perm_indices): New typedef.
12797 (auto_vec_perm_indices): Likewise.
12798 * optabs-query.h: Include target.h
12799 (can_vec_perm_p): Take a vec_perm_indices *.
12800 * optabs-query.c (can_vec_perm_p): Likewise.
12801 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
12802 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12803 * tree-vect-generic.c (lower_vec_perm): Likewise.
12804 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
12805 (vect_grouped_load_supported): Likewise.
12806 (vect_shift_permute_load_chain): Likewise.
12807 (vect_permute_store_chain): Use auto_vec_perm_indices.
12808 (vect_permute_load_chain): Likewise.
12809 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
12810 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
12811 Update uses of can_vec_perm_p.
12812 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
12813 mode with a number of elements. Take a vec_perm_indices *.
12814 (vect_create_epilog_for_reduction): Update accordingly.
12815 Use auto_vec_perm_indices.
12816 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
12817 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
12818 (vect_transform_slp_perm_load): Likewise.
12819 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
12820 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
12821 (vect_gen_perm_mask_checked): Likewise.
12822 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
12823 (vect_gen_perm_mask_checked): Likewise.
12824 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
12825 (vectorizable_store): Likewise.
12826 (vectorizable_load): Likewise.
12827 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
12828 (vectorizable_bswap): Likewise.
12830 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12831 Alan Hayward <alan.hayward@arm.com>
12832 David Sherwood <david.sherwood@arm.com>
12834 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
12835 * tree.c (build_vector): Likewise.
12836 (build_vector_from_ctor): Update accordingly.
12837 (build_vector_from_val): Likewise.
12838 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
12839 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12840 * tree-vect-generic.c (add_rshift): Likewise.
12841 (expand_vector_divmod): Likewise.
12842 (optimize_vector_constructor): Likewise.
12843 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
12844 (vect_transform_slp_perm_load): Likewise.
12845 (vect_schedule_slp_instance): Likewise.
12846 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
12847 (vectorizable_call): Likewise.
12848 (vect_gen_perm_mask_any): Likewise. Add elements in order.
12849 * expmed.c (make_tree): Likewise.
12850 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
12851 a vector passed to build_vector.
12852 (fold_convert_const): Likewise.
12853 (exact_inverse): Likewise.
12854 (fold_ternary_loc): Likewise.
12855 (fold_relational_const): Likewise.
12856 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
12857 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
12858 (const_unop): Likewise. Store the reduction accumulator in a
12859 variable rather than an array.
12860 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
12861 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
12862 the new vector, rather than constructing it after the input arrays.
12863 (native_interpret_vector): Use auto_vec<tree> when building
12864 a vector passed to build_vector. Add elements in order.
12865 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
12866 auto_vec<tree> when building a vector passed to build_vector.
12867 (vect_create_epilog_for_reduction): Likewise.
12868 (vectorizable_induction): Likewise.
12869 (get_initial_def_for_reduction): Likewise. Fix indentation of
12871 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
12873 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
12874 passed to build_vector.
12876 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12877 Alan Hayward <alan.hayward@arm.com>
12878 David Sherwood <david.sherwood@arm.com>
12880 * tree-core.h (tree_base::u): Add an "nelts" field.
12881 (tree_vector): Use VECTOR_CST_NELTS as the length.
12882 * tree.c (tree_size): Likewise.
12883 (make_vector): Initialize VECTOR_CST_NELTS.
12884 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
12885 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
12886 TYPE_VECTOR_SUBPARTS.
12887 * expr.c (const_vector_mask_from_tree): Consistently use "units"
12888 as the number of units, setting it from VECTOR_CST_NELTS.
12889 (const_vector_from_tree): Likewise.
12890 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
12891 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
12892 (fold_negate_expr_1): Likewise.
12893 (fold_convert_const): Likewise.
12894 (const_binop): Likewise. Differentiate the number of output and
12896 (const_unop): Likewise.
12897 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
12898 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
12899 in cases that did the opposite.
12901 2017-09-14 Richard Biener <rguenther@suse.de>
12903 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
12906 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
12908 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
12910 2017-09-14 Jakub Jelinek <jakub@redhat.com>
12913 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
12914 if and where to split a bb, except for splitting before debug insn
12915 sequences followed by non-label real insn. Delete debug insns
12916 in between basic blocks.
12918 * combine.c (make_compound_operation_int): Formatting fixes.
12920 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
12921 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12922 * config/netbsd.h (LINK_EH_SPEC): Likewise.
12923 * config/sol2.h (LINK_EH_SPEC): Likewise.
12924 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12925 * config/s390/linux.h (LINK_SPEC): Likewise.
12926 * config/freebsd.h (LINK_EH_SPEC): Likewise.
12927 * config/openbsd.h (LINK_EH_SPEC): Likewise.
12928 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12929 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
12930 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
12931 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12932 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
12933 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
12935 2017-09-13 Jakub Jelinek <jakub@redhat.com>
12937 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
12939 (ENDFILE_LINUX_SPEC): Likewise.
12940 (LINK_EH_SPEC): Likewise.
12941 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
12942 (LINK_OS_LINUX_SPEC32): Likewise.
12943 (LINK_OS_LINUX_SPEC64): Likewise.
12944 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
12945 (LINK_OS_LINUX_SPEC): Likewise.
12947 2017-09-13 Martin Liska <mliska@suse.cz>
12949 PR middle-end/82154
12950 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
12951 CASE_HIGH is NULL_TREE.
12953 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12954 Alan Hayward <alan.hayward@arm.com>
12955 David Sherwood <david.sherwood@arm.com>
12957 * target.def (secondary_memory_needed): New hook.
12958 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
12959 instead of SECONDARY_MEMORY_NEEDED.
12960 (secondary_memory_needed_mode): Likewise.
12961 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
12962 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
12963 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
12964 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
12965 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
12966 * doc/tm.texi: Regenerate.
12967 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
12968 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
12969 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12970 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
12971 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
12972 * config/i386/i386.c (inline_secondary_memory_needed): Put the
12973 mode argument first and change the reg_class arguments to reg_class_t.
12974 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
12975 Make static. Update the call to inline_secondary_memory_needed.
12976 (ix86_register_move_cost): Update the call to
12977 inline_secondary_memory_needed.
12978 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12979 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
12981 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
12982 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12984 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
12985 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
12986 * config/mips/mips.c (mips_secondary_memory_needed): Make static
12987 and match hook interface. Add comment from mips.h.
12988 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12989 * config/mmix/mmix.md (truncdfsf2): Refer to
12990 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12992 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
12993 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
12994 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12995 (pa_secondary_memory_needed): New function.
12996 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
12997 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
12998 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12999 (pdp11_secondary_memory_needed): Make static and match hook interface.
13000 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
13001 * config/powerpcspe/powerpcspe-protos.h
13002 (rs6000_secondary_memory_needed_ptr): Delete.
13003 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
13005 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13006 (rs6000_option_override_internal): Assign to
13007 targetm.secondary_memory_needed rather than
13008 rs6000_secondary_memory_needed_ptr.
13009 (rs6000_secondary_memory_needed): Match hook interface.
13010 (rs6000_debug_secondary_memory_needed): Likewise.
13011 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
13012 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
13013 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
13014 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13015 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
13016 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
13018 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
13019 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13020 (rs6000_option_override_internal): Assign to
13021 targetm.secondary_memory_needed rather than
13022 rs6000_secondary_memory_needed_ptr.
13023 (rs6000_secondary_memory_needed): Match hook interface.
13024 (rs6000_debug_secondary_memory_needed): Likewise.
13025 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
13026 * config/s390/s390.c (s390_secondary_memory_needed): New function.
13027 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13028 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
13029 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13030 (sparc_secondary_memory_needed): New function.
13031 * lra-constraints.c (check_and_process_move): Refer to
13032 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
13034 (curr_insn_transform): Likewise.
13035 (process_alt_operands): Use targetm.secondary_memory_needed
13036 instead of TARGET_SECONDARY_MEMORY_NEEDED.
13037 (check_secondary_memory_needed_p): Likewise.
13038 (choose_split_class): Likewise.
13039 * reload.c: Unconditionally include code that was previously
13040 conditional on SECONDARY_MEMORY_NEEDED.
13041 (push_secondary_reload): Use targetm.secondary_memory_needed
13042 instead of TARGET_SECONDARY_MEMORY_NEEDED.
13043 (push_reload): Likewise.
13044 * reload1.c: Unconditionally include code that was previously
13045 conditional on SECONDARY_MEMORY_NEEDED.
13046 (choose_reload_regs): Use targetm.secondary_memory_needed
13047 instead of TARGET_SECONDARY_MEMORY_NEEDED.
13048 (gen_reload): Likewise.
13049 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
13051 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
13052 Alan Hayward <alan.hayward@arm.com>
13053 David Sherwood <david.sherwood@arm.com>
13055 * target.def (secondary_memory_needed_mode): New hook:
13056 * targhooks.c (default_secondary_memory_needed_mode): Declare.
13057 * targhooks.h (default_secondary_memory_needed_mode): New function.
13058 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
13059 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
13060 * doc/tm.texi: Regenerate.
13061 * lra-constraints.c (check_and_process_move): Use
13062 targetm.secondary_memory_needed_mode instead of
13063 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
13064 (curr_insn_transform): Likewise.
13065 * reload.c (get_secondary_mem): Likewise.
13066 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13067 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
13069 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13070 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13071 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
13072 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13073 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
13075 * config/powerpcspe/powerpcspe-protos.h
13076 (rs6000_secondary_memory_needed_mode): Delete.
13077 * config/powerpcspe/powerpcspe.c
13078 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13079 (rs6000_secondary_memory_needed_mode): Make static.
13080 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13081 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
13083 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
13085 (rs6000_secondary_memory_needed_mode): Make static.
13086 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13087 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
13088 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13089 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13090 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
13092 (sparc_secondary_memory_needed_mode): New function.
13093 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
13095 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
13097 * config/aarch64/constraints.md (Umq): New constraint.
13098 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
13100 (mov<mode>): Update condition.
13102 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13104 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
13105 when bitposition is the same.
13107 2017-09-13 Richard Biener <rguenther@suse.de>
13109 * dwarf2out.c (output_die_symbol): Remove.
13110 (output_die): Do not output a DIEs symbol.
13112 2017-09-13 Richard Biener <rguenther@suse.de>
13114 PR middle-end/82128
13115 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
13116 default-def to avoid breaking iterator update with the weird
13117 interaction with cgraph_update_edges_for_call_stmt_node.
13119 2017-09-13 Richard Biener <rguenther@suse.de>
13121 * tree-cfg.c (verify_gimple_assign_binary): Add verification
13122 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
13123 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
13124 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
13126 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
13128 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
13129 Disable pc relative literal load irrespective of
13130 TARGET_FIX_ERR_A53_84341 for default.
13132 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
13134 * config/sparc/sparc.c (output_return): Output the source location of
13135 the insn in the delay slot, if any.
13136 (output_sibcall): Likewise.
13138 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
13141 * common.opt (-static-pie): New alias.
13142 (shared): Negate static-pie.
13143 (-no-pie): Update help text.
13145 (static-pie): New option.
13146 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
13147 -static-pie support.
13148 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
13149 (LINK_EH_SPEC): Likewise.
13150 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13151 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
13152 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
13153 * gcc.c (LINK_COMMAND_SPEC): Likewise.
13154 (init_gcc_specs): Likewise.
13155 (init_spec): Likewise.
13156 (display_help): Update help message for -pie.
13157 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
13160 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
13162 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
13163 (movdi_aarch64): Likewise.
13164 (movti_aarch64): Likewise.
13166 2017-09-12 Simon Wright <simon@pushface.org>
13169 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
13170 calculation of the minor version, always output as 0.
13172 2017-09-12 Jakub Jelinek <jakub@redhat.com>
13175 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
13176 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
13177 on it early, rather than manual conversion late. For
13178 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
13179 instead of performing manual conversion.
13181 2017-09-12 Carl Love <cel@us.ibm.com>
13183 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
13184 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
13186 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13187 VMULOSW): Add definitions.
13188 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13189 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13190 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13191 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13192 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13194 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
13196 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
13198 (movti_aarch64): Likewise.
13199 (movdf_aarch64): Likewise.
13200 (movtf_aarch64): Likewise.
13201 (load_pairdi): Likewise.
13202 (store_pairdi): Likewise.
13203 (load_pairdf): Likewise.
13204 (store_pairdf): Likewise.
13205 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
13206 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
13207 (ldr_got_small_<mode>): Likewise.
13208 (ldr_got_small_28k_<mode>): Likewise.
13209 (ldr_got_tiny): Likewise.
13210 * config/aarch64/iterators.md (ldst_sz): New.
13211 (ldpstp_sz): Likewise.
13212 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
13214 (thunderx_load): Split load_8 to load_16.
13215 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
13217 (thunderx2t99_storepair_basic): Split store_8 to store_16.
13218 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
13219 (xgene1_store_pair): Split store_8 to store_16.
13220 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
13221 (falkor_st_0_st_sd): Split store_8 to store_16.
13223 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
13225 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
13226 and store1/2/3/4 to store_4/8/12/16.
13227 * config/aarch64/aarch64.md: Update for rename.
13228 * config/arm/arm.md: Likewise.: Likewise.
13229 * config/arm/arm.c: Likewise.
13230 * config/arm/thumb1.md: Likewise.
13231 * config/arm/thumb2.md: Likewise.
13232 * config/arm/vfp.md: Likewise.
13233 * config/arm/arm-generic.md: Likewise.
13234 * config/arm/arm1020e.md: Likewise.
13235 * config/arm/arm1026ejs.md: Likewise.
13236 * config/arm/arm1136jfs.md: Likewise.
13237 * config/arm/arm926ejs.md: Likewise.
13238 * config/arm/cortex-a15.md: Likewise.
13239 * config/arm/cortex-a17.md: Likewise.
13240 * config/arm/cortex-a5.md: Likewise.
13241 * config/arm/cortex-a53.md: Likewise.
13242 * config/arm/cortex-a57.md: Likewise.
13243 * config/arm/cortex-a7.md: Likewise.
13244 * config/arm/cortex-a8.md: Likewise.
13245 * config/arm/cortex-a9.md: Likewise.
13246 * config/arm/cortex-m4.md: Likewise.
13247 * config/arm/cortex-m7.md: Likewise.
13248 * config/arm/cortex-r4.md: Likewise.
13249 * config/arm/exynos-m1.md: Likewise.
13250 * config/arm/fa526.md: Likewise.
13251 * config/arm/fa606te.md: Likewise.
13252 * config/arm/fa626te.md: Likewise.
13253 * config/arm/fa726te.md: Likewise.
13254 * config/arm/fmp626.md: Likewise.
13255 * config/arm/iwmmxt.md: Likewise.
13256 * config/arm/ldmstm.md: Likewise.
13257 * config/arm/marvell-pj4.md: Likewise.
13258 * config/arm/xgene1.md: Likewise.
13259 * config/aarch64/thunderx.md: Likewise.
13260 * config/aarch64/thunderx2t99.md: Likewise.
13261 * config/aarch64/falkor.md: Likewise.
13263 2017-09-12 Martin Liska <mliska@suse.cz>
13265 * attribs.c (private_lookup_attribute): New function.
13266 * attribs.h (private_lookup_attribute): Declared here.
13267 (lookup_attribute): Called from this place.
13269 2017-09-12 Richard Biener <rguenther@suse.de>
13271 PR tree-optimization/82157
13272 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
13273 stmts with side-effects.
13275 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13276 Alan Hayward <alan.hayward@arm.com>
13277 David Sherwood <david.sherwood@arm.com>
13279 * target.def (hard_regno_nregs): New hook.
13280 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
13281 * targhooks.h (default_hard_regno_nregs): Declare.
13282 * targhooks.c (default_hard_regno_nregs): New function.
13283 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
13284 (TARGET_HARD_REGNO_NREGS): ...this hook.
13285 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
13286 (CLASS_MAX_NREGS): Likewise.
13287 * doc/tm.texi: Regenerate.
13288 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
13289 instead of HARD_REGNO_NREGS.
13290 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
13291 HARD_REGNO_NREGS in the comment.
13292 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
13293 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
13294 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
13295 Return an unsigned int.
13296 (TARGET_HARD_REGNO_NREGS): Redefine.
13297 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
13298 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
13299 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
13300 (arc_hard_regno_nregs): New function.
13301 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
13302 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
13303 (arm_hard_regno_nregs): New function.
13304 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
13305 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
13306 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
13307 (TARGET_HARD_REGNO_NREGS): Redefine.
13308 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
13309 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
13310 (HARD_REGNO_NREGS): Delete.
13311 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
13312 (cr16_hard_regno_nregs): New function.
13313 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
13314 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
13315 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
13316 (cris_hard_regno_nregs): New function.
13317 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
13318 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
13319 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
13320 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
13321 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
13322 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
13323 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
13324 (frv_hard_regno_nregs): Make static. Take and return an
13326 (frv_class_max_nregs): Remove outdated copy of documentation.
13327 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
13328 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
13329 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
13330 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
13331 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
13332 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
13333 (TARGET_HARD_REGNO_NREGS): Redefine.
13334 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
13335 (CLASS_MAX_NREGS): Update comment.
13336 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
13337 (ia64_hard_regno_nregs): New function.
13338 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
13339 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
13340 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
13341 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
13342 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
13344 (m32c_hard_regno_nregs): Likewise. Make static.
13345 (TARGET_HARD_REGNO_NREGS): Redefine.
13346 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
13347 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
13348 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
13349 (m68k_hard_regno_nregs): New function.
13350 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
13351 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
13352 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
13353 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
13354 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
13355 Take and return an unsigned int.
13356 (TARGET_HARD_REGNO_NREGS): Redefine.
13357 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
13358 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
13359 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
13360 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
13361 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
13362 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
13363 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
13364 (msp430_hard_regno_nregs): Make static. Take and return an
13366 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
13367 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
13368 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
13369 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
13370 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
13371 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
13372 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
13373 (TARGET_HARD_REGNO_NREGS): Redefine.
13374 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
13375 (PA_HARD_REGNO_NREGS): ...this.
13376 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
13377 (PA_HARD_REGNO_NREGS): ...this.
13378 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
13379 (pa_hard_regno_nregs): New function.
13380 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
13381 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
13382 (pdp11_hard_regno_nregs): New function.
13383 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
13384 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
13385 (rs6000_hard_regno_nregs_hook): New function.
13386 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
13387 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
13388 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
13389 Take and return an unsigned int. Move earlier in file.
13390 (TARGET_HARD_REGNO_NREGS): Redefine.
13391 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
13392 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
13393 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
13394 (rl78_hard_regno_nregs): Make static. Take and return an
13396 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
13397 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
13398 (rs6000_hard_regno_nregs_hook): New function.
13399 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
13400 * config/rx/rx.c (rx_hard_regno_nregs): New function.
13401 (TARGET_HARD_REGNO_NREGS): Redefine.
13402 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
13403 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
13404 instead of HARD_REGNO_NREGS.
13405 (s390_hard_regno_nregs): New function.
13406 (s390_hard_regno_mode_ok): Add comment from s390.h.
13407 (TARGET_HARD_REGNO_NREGS): Redefine.
13408 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
13409 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
13410 (sh_hard_regno_nregs): New function.
13411 (sh_pass_in_reg_p): Use it.
13412 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
13413 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
13414 (sparc_hard_regno_nregs): New function.
13415 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
13416 * config/spu/spu.c (spu_hard_regno_nregs): New function.
13417 (spu_function_arg_advance): Use it, supplying a valid register number.
13418 (TARGET_HARD_REGNO_NREGS): Redefine.
13419 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
13420 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
13421 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
13422 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
13423 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
13424 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
13425 (CLASS_MAX_NREGS): Remove copy of old documentation.
13426 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
13427 (visium_hard_regno_nregs): New function.
13428 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
13429 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
13430 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
13431 xtensa_hard_regno_nregs): New function.
13432 * system.h (HARD_REGNO_NREGS): Poison.
13434 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13436 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
13437 hard_regno_nregs instead of HARD_REGNO_NREGS.
13438 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
13439 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
13440 (c6x_expand_epilogue): Likewise.
13441 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
13442 (frv_read_iacc_argument): Likewise.
13443 * config/sh/sh.c: Include regs.h.
13444 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
13445 (regs_used): Likewise.
13446 (output_stack_adjust): Likewise.
13447 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
13448 * expmed.c: Include regs.h.
13449 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
13450 * ree.c: Include regs.h.
13451 (combine_reaching_defs): Use hard_regno_nregs instead of
13453 (add_removable_extension): Likewise.
13455 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13457 * regs.h (hard_regno_nregs): Turn into a function.
13458 (end_hard_regno): Update accordingly.
13459 * caller-save.c (setup_save_areas): Likewise.
13460 (save_call_clobbered_regs): Likewise.
13461 (replace_reg_with_saved_mem): Likewise.
13462 (insert_restore): Likewise.
13463 (insert_save): Likewise.
13464 * combine.c (can_change_dest_mode): Likewise.
13465 (move_deaths): Likewise.
13466 (distribute_notes): Likewise.
13467 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
13468 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
13469 (rs6000_split_multireg_move): Likewise.
13470 (rs6000_register_move_cost): Likewise.
13471 (rs6000_memory_move_cost): Likewise.
13472 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
13473 (rs6000_split_multireg_move): Likewise.
13474 (rs6000_register_move_cost): Likewise.
13475 (rs6000_memory_move_cost): Likewise.
13476 * cselib.c (cselib_reset_table): Likewise.
13477 (cselib_lookup_1): Likewise.
13478 * emit-rtl.c (set_mode_and_regno): Likewise.
13479 * function.c (aggregate_value_p): Likewise.
13480 * ira-color.c (setup_profitable_hard_regs): Likewise.
13481 (check_hard_reg_p): Likewise.
13482 (calculate_saved_nregs): Likewise.
13483 (assign_hard_reg): Likewise.
13484 (improve_allocation): Likewise.
13485 (calculate_spill_cost): Likewise.
13486 * ira-emit.c (modify_move_list): Likewise.
13487 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
13488 (ira_hard_reg_in_set_p): Likewise.
13489 * ira.c (setup_reg_mode_hard_regset): Likewise.
13490 (clarify_prohibited_class_mode_regs): Likewise.
13491 (check_allocation): Likewise.
13492 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13493 (lra_setup_reg_renumber): Likewise.
13494 (setup_try_hard_regno_pseudos): Likewise.
13495 (spill_for): Likewise.
13496 (assign_hard_regno): Likewise.
13497 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
13498 * lra-constraints.c (in_class_p): Likewise.
13499 (lra_constraint_offset): Likewise.
13500 (simplify_operand_subreg): Likewise.
13501 (lra_constraints): Likewise.
13502 (split_reg): Likewise.
13503 (split_if_necessary): Likewise.
13504 (invariant_p): Likewise.
13505 (inherit_in_ebb): Likewise.
13506 * lra-lives.c (process_bb_lives): Likewise.
13507 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
13508 (get_hard_regs): Likewise.
13509 (do_remat): Likewise.
13510 * lra-spills.c (assign_spill_hard_regs): Likewise.
13511 * mode-switching.c (create_pre_exit): Likewise.
13512 * postreload.c (reload_combine_recognize_pattern): Likewise.
13513 * recog.c (peep2_find_free_register): Likewise.
13514 * regcprop.c (kill_value_regno): Likewise.
13515 (set_value_regno): Likewise.
13516 (copy_value): Likewise.
13517 (maybe_mode_change): Likewise.
13518 (find_oldest_value_reg): Likewise.
13519 (copyprop_hardreg_forward_1): Likewise.
13520 * regrename.c (check_new_reg_p): Likewise.
13521 (regrename_do_replace): Likewise.
13522 * reload.c (push_reload): Likewise.
13523 (combine_reloads): Likewise.
13524 (find_dummy_reload): Likewise.
13525 (operands_match_p): Likewise.
13526 (find_reloads): Likewise.
13527 (find_equiv_reg): Likewise.
13528 (reload_adjust_reg_for_mode): Likewise.
13529 * reload1.c (count_pseudo): Likewise.
13530 (count_spilled_pseudo): Likewise.
13531 (find_reg): Likewise.
13532 (clear_reload_reg_in_use): Likewise.
13533 (free_for_value_p): Likewise.
13534 (allocate_reload_reg): Likewise.
13535 (choose_reload_regs): Likewise.
13536 (reload_adjust_reg_for_temp): Likewise.
13537 (emit_reload_insns): Likewise.
13538 (delete_output_reload): Likewise.
13539 * rtlanal.c (subreg_get_info): Likewise.
13540 * sched-deps.c (sched_analyze_reg): Likewise.
13541 * sel-sched.c (init_regs_for_mode): Likewise.
13542 (mark_unavailable_hard_regs): Likewise.
13543 (choose_best_reg_1): Likewise.
13544 (verify_target_availability): Likewise.
13545 * valtrack.c (dead_debug_insert_temp): Likewise.
13546 * var-tracking.c (track_loc_p): Likewise.
13547 (emit_note_insn_var_location): Likewise.
13548 * varasm.c (make_decl_rtl): Likewise.
13549 * reginfo.c (choose_hard_reg_mode): Likewise.
13550 (init_reg_modes_target): Refer directly to
13551 this_target_regs->x_hard_regno_nregs.
13553 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13555 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
13556 instead of hard_regno_nregs.
13558 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13560 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
13561 end_hard_regno instead of hard_regno_nregs.
13562 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
13563 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
13564 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
13565 * ira-color.c (improve_allocation): Likewise.
13566 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13567 * lra-lives.c (mark_regno_live): Likewise.
13568 (mark_regno_dead): Likewise.
13569 * lra-remat.c (operand_to_remat): Likewise.
13570 * lra.c (collect_non_operand_hard_regs): Likewise.
13571 * postreload.c (reload_combine_note_store): Likewise.
13572 (move2add_valid_value_p): Likewise.
13573 * reload.c (regno_clobbered_p): Likewise.
13575 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13577 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
13579 * config/v850/v850.c (v850_reorg): Likewise.
13580 * reload.c (refers_to_regno_for_reload_p): Likewise.
13581 (find_equiv_reg): Likewise.
13582 * reload1.c (reload_reg_reaches_end_p): Likewise.
13584 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13586 * caller-save.c (add_used_regs): Use REG_NREGS instead of
13588 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
13589 * config/arm/arm.c (output_move_neon): Likewise.
13590 (arm_attr_length_move_neon): Likewise.
13591 (neon_split_vcombine): Likewise.
13592 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
13593 (c6x_mark_reg_written): Likewise.
13594 (c6x_dwarf_register_span): Likewise.
13595 * config/i386/i386.c (ix86_save_reg): Likewise.
13596 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
13597 (rws_access_reg): Likewise.
13598 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
13599 * mode-switching.c (create_pre_exit): Likewise.
13600 * ree.c (combine_reaching_defs): Likewise.
13601 (add_removable_extension): Likewise.
13602 * regcprop.c (find_oldest_value_reg): Likewise.
13603 (copyprop_hardreg_forward_1): Likewise.
13604 * reload.c (reload_inner_reg_of_subreg): Likewise.
13605 (push_reload): Likewise.
13606 (combine_reloads): Likewise.
13607 (find_dummy_reload): Likewise.
13608 (reload_adjust_reg_for_mode): Likewise.
13609 * reload1.c (find_reload_regs): Likewise.
13610 (forget_old_reloads_1): Likewise.
13611 (reload_reg_free_for_value_p): Likewise.
13612 (reload_adjust_reg_for_temp): Likewise.
13613 (emit_reload_insns): Likewise.
13614 (delete_output_reload): Likewise.
13615 * sel-sched.c (choose_best_reg_1): Likewise.
13616 (choose_best_pseudo_reg): Likewise.
13618 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13619 Alan Hayward <alan.hayward@arm.com>
13620 David Sherwood <david.sherwood@arm.com>
13622 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
13623 * target.def (slow_unaligned_access): New hook.
13624 * targhooks.h (default_slow_unaligned_access): Declare.
13625 * targhooks.c (default_slow_unaligned_access): New function.
13626 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
13627 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
13628 * doc/tm.texi: Regenerate.
13629 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
13630 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
13631 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
13633 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
13634 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
13636 (rs6000_slow_unaligned_access): New function.
13637 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13638 (expand_block_compare): Likewise.
13639 (expand_strn_compare): Likewise.
13640 (rs6000_rtx_costs): Likewise.
13641 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
13642 (riscv_slow_unaligned_access): Likewise.
13643 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
13644 (riscv_slow_unaligned_access_p): ...this and make static.
13645 (riscv_option_override): Update accordingly.
13646 (riscv_slow_unaligned_access): New function.
13647 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13648 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
13649 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13650 (rs6000_slow_unaligned_access): New function.
13651 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13652 (rs6000_rtx_costs): Likewise.
13653 * config/rs6000/rs6000-string.c (expand_block_compare)
13654 (expand_strn_compare): Use targetm.slow_unaligned_access instead
13655 of SLOW_UNALIGNED_ACCESS.
13656 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
13657 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
13658 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
13659 of SLOW_UNALIGNED_ACCESS.
13660 * expmed.c (simple_mem_bitfield_p): Likewise.
13661 * expr.c (alignment_for_piecewise_move): Likewise.
13662 (emit_group_load_1): Likewise.
13663 (emit_group_store): Likewise.
13664 (copy_blkmode_from_reg): Likewise.
13665 (emit_push_insn): Likewise.
13666 (expand_assignment): Likewise.
13667 (store_field): Likewise.
13668 (expand_expr_real_1): Likewise.
13669 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
13670 * lra-constraints.c (simplify_operand_subreg): Likewise.
13671 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13672 * gimple-ssa-store-merging.c: Likewise in block comment at start
13674 * tree-ssa-strlen.c: Include target.h.
13675 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
13676 of SLOW_UNALIGNED_ACCESS.
13677 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
13679 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13681 PR rtl-optimization/82185
13682 * expmed.c (emit_store_flag_int): Only test tem if it has been
13685 2017-09-12 Richard Biener <rguenther@suse.de>
13687 PR middle-end/82149
13688 * match.pd ((FTYPE) N CMP CST): Fix typo.
13690 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
13692 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
13694 (mips_near_type_p): Add 'short_call' attribute as a synonym
13696 * doc/extend.texi (short_call): Document new function attribute.
13698 2017-09-12 Jakub Jelinek <jakub@redhat.com>
13701 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
13702 assertion check that in the condition.
13703 (get_atomic_generic_size): Likewise. Before testing if parameter
13704 has pointer type, if it has array type, call for C++
13705 default_conversion to perform array-to-pointer conversion.
13707 2017-09-12 Richard Biener <rguenther@suse.de>
13709 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
13710 for operations we cannot scalarize.
13712 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
13714 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
13715 vectors heap vectors. Clean up comments.
13716 Make visited_bbs a reference.
13717 (profitable_jump_thread_path): Make GC
13718 vectors heap vectors. Clean up comments.
13720 (convert_and_register_jump_thread_path): Make GC vectors heap
13722 (check_subpath_and_update_thread_path): Same. Clean up comments.
13723 Make visited_bbs a reference.
13724 (handle_phi): Abstract common code to to
13725 register_jump_thread_path_if_profitable.
13726 Rename VAR_BB to DEF_BB.
13728 Make GC vectors heap vectors.
13729 Make visited_bbs a reference.
13730 (handle_assignment): Same.
13731 (register_jump_thread_path_if_profitable): New.
13732 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
13734 Make GC vectors heap vectors. Clean up comments.
13735 Make visited_bbs a reference.
13736 (find_jump_threads_backwards): Make visited_bbs live in the stack.
13737 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
13740 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
13743 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
13744 words of E_DImode object are reachable by xtensa_uimm8x4 access.
13746 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
13748 Revert r251800 and r251799.
13750 2017-09-11 Martin Jambor <mjambor@suse.cz>
13753 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
13754 arguments in advance.
13755 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
13756 use it to find predecessor edges.
13757 (naive_outof_ssa): Collect vector of predecessors.
13759 2017-09-08 Jason Merrill <jason@redhat.com>
13761 PR c++/70029 - ICE with ref-qualifier and -flto
13762 * langhooks.h (struct lang_hooks_for_types): Add
13763 copy_lang_qualifiers.
13764 * attribs.c (build_type_attribute_qual_variant): Use it.
13765 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
13767 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
13768 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
13770 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
13773 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
13774 (*mulsi3_sp64): New instruction.
13775 (mulsi3): New expander.
13777 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
13779 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
13781 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13783 * sancov.c: Include memmodel.h.
13785 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
13788 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
13791 2017-09-07 Carl Love <cel@us.ibm.com>
13793 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
13794 the sldi instruction.
13796 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
13798 * sancov.c: Include tm_p.h.
13800 2017-09-07 Jakub Jelinek <jakub@redhat.com>
13803 * output.h (switch_to_other_text_partition): New declaration.
13804 * varasm.c (switch_to_other_text_partition): New function.
13805 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
13806 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
13807 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
13808 to the other text partition before emitting LCL label and switch back
13809 after emitting the word after it.
13811 2017-09-07 Richard Biener <rguenther@suse.de>
13813 * passes.def (pass_split_crit_edges): Remove instance before PRE.
13814 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
13815 critical edges here, after loop init.
13816 (pass_data_pre): Remove PROP_no_crit_edges flags.
13817 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
13818 for valueization of call args to avoid leaking VN_TOP.
13819 (visit_use): Assert we do not visit default defs.
13820 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
13821 Use error_mark_node to more easily detect leaking VN_TOP.
13822 All default-defs are varying, not VN_TOP. Mark them visited.
13823 (run_scc_vn): Make code match comment.
13825 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13827 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
13828 OPTION_MASK_FLOAT128_KEYWORD.
13829 (POWERPC_MASKS): Likewise.
13830 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
13831 support for the -mfloat128-type option, and make -mfloat128
13832 default on PowerPC Linux systems. Define or undefine
13833 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
13834 Define __float128 to be __ieee128 if IEEE 128-bit support is
13835 enabled, or undefine it.
13836 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
13837 Delete defining __FLOAT128_TYPE__.
13838 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
13839 -mfloat128-type option and make -mfloat128 default on PowerPC
13841 (TARGET_FLOAT128_TYPE): Likewise.
13842 (-mfloat128-type): Likewise.
13843 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13844 Delete the -mfloat128-type option and make -mfloat128 default on
13845 PowerPC Linux systems. Always use __ieee128 to be the keyword for
13846 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
13847 128-bit floating point is enabled. Change tests from using
13848 -mfloat128-type to -mfloat128.
13849 (rs6000_mangle_type): Use the correct mangling for the __float128
13850 type even if normal long double is restricted to 64-bits.
13851 (floatn_mode): Enable the _Float128 type by default on VSX Linux
13853 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
13854 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
13855 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
13857 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
13858 documentation for -mfloat128.
13860 2017-09-06 Olivier Hainque <hainque@adacore.com>
13862 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
13864 2017-09-06 Wish Wu <wishwu007@gmail.com>
13865 Jakub Jelinek <jakub@redhat.com>
13867 * asan.c (initialize_sanitizer_builtins): Add
13868 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
13869 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
13870 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
13871 BT_FN_VOID_UINT64_PTR variables.
13872 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
13873 (BT_FN_VOID_UINT16_UINT16): Likewise.
13874 (BT_FN_VOID_UINT32_UINT32): Likewise.
13875 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
13876 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
13877 (BT_FN_VOID_UINT64_PTR): Likewise.
13878 * common.opt (flag_sanitize_coverage): New variable.
13879 (fsanitize-coverage=trace-pc): Remove.
13880 (fsanitize-coverage=): Add.
13881 * flag-types.h (enum sanitize_coverage_code): New enum.
13882 * fold-const.c (fold_range_test): Disable non-short-circuit
13883 optimization if flag_sanitize_coverage.
13884 (fold_truth_andor): Likewise.
13885 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
13886 * opts.c (COVERAGE_SANITIZER_OPT): Define.
13887 (coverage_sanitizer_opts): New array.
13888 (get_closest_sanitizer_option): Add OPTS argument, handle also
13889 OPT_fsanitize_coverage_.
13890 (parse_sanitizer_options): Adjusted to also handle
13891 OPT_fsanitize_coverage_.
13892 (common_handle_option): Add OPT_fsanitize_coverage_.
13893 * sancov.c (instrument_comparison, instrument_switch): New function.
13894 (sancov_pass): Add trace-cmp support.
13895 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
13896 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
13897 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
13898 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
13899 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
13900 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
13901 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
13902 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
13903 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
13904 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
13906 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13908 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13909 error. Only quit immediately if parsing is complete.
13910 (BEGIN): Initialize fatal_err and parse_done.
13911 (begin fpu, end fpu): Check number of arguments.
13912 (begin arch, end arch): Likewise.
13913 (begin cpu, end cpu): Likewise.
13914 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13915 (optalias): Likewise.
13917 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13919 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13920 * config/arm/arm-isa.h: Delete. Move definitions to ...
13921 * arm-cpus.in: ... here. Use new feature and fgroup values.
13922 * config/arm/arm.c (arm_option_override): Use lower case for feature
13924 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13925 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13926 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13928 (print_isa_bits_for): New function.
13929 (gen_isa): New function.
13930 (gen_comm_data): Use print_isa_bits_for.
13931 (define feature): New keyword.
13932 (define fgroup): New keyword.
13933 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
13934 (arm-isa.h): Add rule to generate file.
13935 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13936 case for feature bit names.
13938 2017-09-06 Richard Biener <rguenther@suse.de>
13940 * tree-ssa-pre.c (NECESSARY): Remove.
13941 (create_expression_by_pieces): Do not touch pass-local flags.
13942 (insert_into_preds_of_block): Likewise.
13943 (do_pre_regular_insertion): Likewise.
13944 (eliminate_insert): Likewise.
13945 (eliminate_dom_walker::before_dom_children): Likewise.
13946 (fini_eliminate): Do not look at inserted_exprs.
13947 (mark_operand_necessary): Remove.
13948 (remove_dead_inserted_code): Replace with simple work-list
13949 algorithm based on inserted_exprs and SSA uses.
13950 (pass_pre::execute): Re-order fini_eliminate and
13951 remove_dead_inserted_code.
13953 2017-09-06 Olivier Hainque <hainque@adacore.com>
13955 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
13956 for VxWorks 7. Adjust surrounding comments.
13958 2017-09-06 Richard Biener <rguenther@suse.de>
13960 * gimple-ssa-strength-reduction.c
13961 (find_candidates_dom_walker::before_dom_children): Also allow
13964 2017-09-06 Richard Biener <rguenther@suse.de>
13966 PR tree-optimization/82108
13967 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
13968 for gap in the non-permutation SLP case.
13970 2017-09-06 Martin Jambor <mjambor@suse.cz>
13972 PR tree-optimization/82078
13973 * tree-sra.c (sort_and_splice_var_accesses): Move call to
13974 add_access_to_work_queue...
13975 (build_accesses_from_assign): ...here.
13976 (propagate_all_subaccesses): Make sure racc is the group
13977 representative, if there is one.
13979 2017-09-06 Jakub Jelinek <jakub@redhat.com>
13981 PR middle-end/82095
13982 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
13985 2017-09-06 Richard Biener <rguenther@suse.de>
13987 * gimple-ssa-strength-reduction.c
13988 (find_candidates_dom_walker::before_doom_children): Use a
13989 type and not a mode check.
13991 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
13994 * config/arm/predicates.md (arm_general_adddi_operand): Create new
13996 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
13998 2017-09-05 Jeff Law <law@redhat.com>
14000 PR tree-optimization/64910
14001 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
14002 cases where we have 3 or more operands.
14004 2017-09-05 Jakub Jelinek <jakub@redhat.com>
14006 PR middle-end/81768
14007 * omp-low.c (lower_omp_for): Recompute tree invariant if
14008 gimple_omp_for_initial/final is ADDR_EXPR.
14010 PR middle-end/81768
14011 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
14012 into gimple val before gimplification fo the COND_EXPR.
14014 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
14016 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
14017 REGION_COPY argument.
14018 (thread_through_all_blocks): Remove unused argument to
14019 duplicate_thread_path.
14021 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14022 Alan Hayward <alan.hayward@arm.com>
14023 David Sherwood <david.sherwood@arm.com>
14025 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
14026 Take a scalar_mode rather than a machine_mode.
14027 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14028 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
14029 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14030 (aarch64_gen_adjusted_ldpstp): Likewise.
14031 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
14033 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14034 Alan Hayward <alan.hayward@arm.com>
14035 David Sherwood <david.sherwood@arm.com>
14037 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
14038 Take a scalar_int_mode instead of a machine_mode.
14039 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
14040 (aarch64_output_scalar_simd_mov_immediate): Likewise.
14041 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
14042 (aarch64_simd_attr_length_rglist): Delete.
14043 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
14044 a scalar_int_mode instead of a machine_mode.
14045 (aarch64_add_offset): Likewise.
14046 (aarch64_internal_mov_immediate): Likewise
14047 (aarch64_add_constant_internal): Likewise.
14048 (aarch64_add_constant): Likewise.
14049 (aarch64_movw_imm): Likewise.
14050 (aarch64_rtx_arith_op_extract_p): Likewise.
14051 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
14052 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
14053 Remove assert that the mode isn't a vector.
14054 (aarch64_output_scalar_simd_mov_immediate): Likewise.
14055 (aarch64_expand_mov_immediate): Update calls after above changes.
14056 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
14057 (aarch64_and_bitmask_imm): Check for scalar integer modes.
14058 (aarch64_move_imm): Likewise.
14059 (aarch64_can_const_movi_rtx_p): Likewise.
14060 (aarch64_strip_extend): Likewise.
14061 (aarch64_extr_rtx_p): Likewise.
14062 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
14063 a CONST_INT when the mode parameter is VOIDmode.
14064 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
14066 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14068 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
14069 * stor-layout.c (bitwise_mode_for_mode): Likewise.
14070 (bitwise_type_for_mode): Update accordingly.
14072 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14074 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
14075 * stor-layout.c (mode_for_size_tree): Likewise.
14076 (mode_for_array): Update accordingly.
14077 (layout_decl): Likewise.
14078 (compute_record_mode): Likewise. Only set the mode once.
14080 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14082 * target.def (get_mask_mode): Change return type to opt_mode.
14084 * doc/tm.texi: Regenerate.
14085 * targhooks.h (default_get_mask_mode): Return an opt_mode.
14086 * targhooks.c (default_get_mask_mode): Likewise.
14087 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
14088 * optabs-query.c (can_vec_mask_load_store_p): Update use of
14089 targetm.get_mask_mode.
14090 * tree.c (build_truth_vector_type): Likewise.
14092 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14094 * machmode.h (mode_for_vector): Return an opt_mode.
14095 * stor-layout.c (mode_for_vector): Likewise.
14096 (mode_for_int_vector): Update accordingly.
14097 (layout_type): Likewise.
14098 * config/i386/i386.c (emit_memmov): Likewise.
14099 (ix86_expand_set_or_movmem): Likewise.
14100 (ix86_expand_vector_init): Likewise.
14101 (ix86_get_mask_mode): Likewise.
14102 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
14104 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
14105 * expmed.c (extract_bit_field_1): Likewise.
14106 * expr.c (expand_expr_real_2): Likewise.
14107 * optabs-query.c (can_vec_perm_p): Likewise.
14108 (can_vec_mask_load_store_p): Likewise.
14109 * optabs.c (expand_vec_perm): Likewise.
14110 * targhooks.c (default_get_mask_mode): Likewise.
14111 * tree-vect-stmts.c (vectorizable_store): Likewise.
14112 (vectorizable_load): Likewise.
14113 (get_vectype_for_scalar_type_and_size): Likewise.
14115 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14117 * machmode.h (mode_for_int_vector): New function.
14118 * stor-layout.c (mode_for_int_vector): Likewise.
14119 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
14120 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
14121 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
14122 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
14123 (s390_expand_vcond): Likewise.
14125 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14127 * machmode.h (opt_machine_mode): New type.
14128 (opt_mode<T>): Allow construction from anything that can be
14130 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
14131 (mode_for_size): Return an opt_machine_mode.
14132 * stor-layout.c (mode_for_size): Likewise.
14133 (mode_for_size_tree): Update call accordingly.
14134 (bitwise_mode_for_mode): Likewise.
14135 (make_fract_type): Likewise.
14136 (make_accum_type): Likewise.
14137 * caller-save.c (replace_reg_with_saved_mem): Update call
14139 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14140 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14141 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14142 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14143 * expmed.c (extract_bit_field_1): Likewise.
14144 * reload.c (get_secondary_mem): Likewise.
14145 * varasm.c (assemble_integer): Likewise.
14146 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
14149 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14151 * machmode.h (decimal_float_mode_for_size): New function.
14152 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
14153 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
14154 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
14155 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
14156 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
14157 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
14159 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14161 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
14162 (get_builtin_sync_mode): Likewise.
14163 (expand_ifn_atomic_compare_exchange): Likewise.
14164 (expand_builtin_atomic_clear): Likewise.
14165 (expand_builtin_atomic_test_and_set): Likewise.
14166 (fold_builtin_atomic_always_lock_free): Likewise.
14167 * calls.c (compute_argument_addresses): Likewise.
14168 (emit_library_call_value_1): Likewise.
14169 (store_one_arg): Likewise.
14170 * combine.c (combine_instructions): Likewise.
14171 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
14172 * config/arm/arm.c (arm_function_value): Likewise.
14173 (aapcs_allocate_return_reg): Likewise.
14174 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
14175 * config/i386/i386.c (construct_container): Likewise.
14176 (ix86_gimplify_va_arg): Likewise.
14177 (ix86_expand_sse_cmp): Likewise.
14178 (emit_memmov): Likewise.
14179 (emit_memset): Likewise.
14180 (expand_small_movmem_or_setmem): Likewise.
14181 (ix86_expand_pextr): Likewise.
14182 (ix86_expand_pinsr): Likewise.
14183 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
14184 * config/microblaze/microblaze.c (microblaze_block_move_straight):
14186 * config/mips/mips.c (mips_function_value_1) Likewise.
14187 (mips_block_move_straight): Likewise.
14188 (mips_expand_ins_as_unaligned_store): Likewise.
14189 * config/powerpcspe/powerpcspe.c
14190 (rs6000_darwin64_record_arg_advance_flush): Likewise.
14191 (rs6000_darwin64_record_arg_flush): Likewise.
14192 * config/rs6000/rs6000.c
14193 (rs6000_darwin64_record_arg_advance_flush): Likewise.
14194 (rs6000_darwin64_record_arg_flush): Likewise.
14195 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
14196 (sparc_function_value_1): Likewise.
14197 * config/spu/spu.c (adjust_operand): Likewise.
14198 (spu_emit_branch_or_set): Likewise.
14199 (arith_immediate_p): Likewise.
14200 * emit-rtl.c (gen_lowpart_common): Likewise.
14201 * expr.c (expand_expr_real_1): Likewise.
14202 * function.c (assign_parm_setup_block): Likewise.
14203 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
14204 * reload1.c (alter_reg): Likewise.
14205 * stor-layout.c (mode_for_vector): Likewise.
14206 (layout_type): Likewise.
14208 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14210 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
14211 (spu_convert_move): Likewise.
14212 * lower-subreg.c (resolve_simple_move): Likewise.
14214 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14217 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
14218 define_insn to a define_expand.
14219 (altivec_vsum2sws_direct): New define_insn.
14220 (altivec_vsumsws): Convert from a define_insn to a define_expand.
14222 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
14224 * config/arm/arm.c (arm_option_params_internal): Improve setting of
14227 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
14232 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
14233 realign stack if stack alignment needed is less than incoming
14236 2017-09-05 Marek Polacek <polacek@redhat.com>
14239 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
14242 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
14244 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
14246 2017-09-05 Richard Biener <rguenther@suse.de>
14248 PR tree-optimization/82084
14249 * fold-const.c (can_native_encode_string_p): Handle wide characters.
14251 2017-09-05 Richard Biener <rguenther@suse.de>
14253 PR tree-optimization/82102
14254 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
14256 2017-09-05 Martin Liska <mliska@suse.cz>
14258 PR tree-optimization/82032
14259 * tree-cfg.c (generate_range_test): New function.
14260 * tree-cfg.h (generate_range_test): Declared here.
14261 * tree-cfgcleanup.c (convert_single_case_switch): New function.
14262 (cleanup_control_expr_graph): Use it.
14263 * tree-switch-conversion.c (try_switch_expansion): Remove
14265 (emit_case_nodes): Use generate_range_test.
14267 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
14270 * config/i386/i386.md (*<btsc><mode>_mask): Add
14271 TARGET_USE_BT to insn constraint.
14272 (*btr<mode>_mask): Ditto.
14274 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
14276 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
14277 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
14279 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
14282 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
14283 TARGET_NEON and TARGET_IWMMXT.
14284 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
14285 TARGET_NEON and TARGET_IWMMXT.
14286 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
14288 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
14290 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
14291 (ix86_rewrite_tls_address): Ditto.
14292 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
14293 (ix86_rewrite_tls_address_1): Ditto.
14294 (ix86_rewrite_tls_address): Ditto.
14295 * config/i386/predicates.md (tls_address_pattern): New predicate.
14296 * config/i386/i386.md (TLS address splitter): New splitter.
14298 2017-09-04 Richard Biener <rguenther@suse.de>
14300 PR tree-optimization/82084
14301 * fold-const.h (can_native_encode_string_p): Declare.
14302 * fold-const.c (can_native_encode_string_p): Factor out from ...
14303 (native_encode_string): ... here.
14304 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
14305 vectorizing stores from constants we later cannot handle.
14307 2017-09-04 Marek Polacek <polacek@redhat.com>
14310 * doc/invoke.texi: Update -Wtautological-compare documentation.
14312 2017-09-04 Jeff Law <law@redhat.com>
14314 PR tree-optimization/64910
14315 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
14316 swap the first and last operand if the last is a constant.
14318 2017-09-04 Marek Polacek <polacek@redhat.com>
14321 * convert.c (do_narrow): When sanitizing signed integer overflows,
14322 bail out for signed types.
14323 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
14325 2017-09-04 Richard Biener <rguenther@suse.de>
14327 PR tree-optimization/82060
14328 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
14329 Move devirtualization after stmt folding and before EH/AB/noreturn
14330 cleanup to get the stmt refs canonicalized. Use a bool instead
14331 of gimple_modified_p since that doesn't work for NOPs. Schedule
14332 NOPs generated by folding for removal.
14334 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14335 Alan Hayward <alan.hayward@arm.com>
14336 David Sherwood <david.sherwood@arm.com>
14338 * coretypes.h (pad_direction): New enum.
14339 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
14340 (FUNCTION_ARG_PADDING): Likewise.
14341 * target.def (function_arg_padding): New hook.
14342 * targhooks.h (default_function_arg_padding): Declare.
14343 * targhooks.c (default_function_arg_padding): New function.
14344 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
14345 (TARGET_FUNCTION_ARG_PADDING): ...this.
14346 * doc/tm.texi: Regenerate.
14347 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
14348 instead of direction.
14349 (compute_argument_addresses): Likewise.
14350 (load_register_parameters): Likewise.
14351 (emit_library_call_value_1): Likewise.
14352 (store_one_arg): Use targetm.calls.function_arg_padding instead
14353 of FUNCTION_ARG_PADDING.
14354 (must_pass_in_stack_var_size_or_pad): Likewise.
14355 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
14356 (emit_group_store): Likewise.
14357 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
14358 instead of FUNCTION_ARG_PADDING.
14359 (emit_push_insn): Likewise, and propagate enum change throughout
14361 * function.h (direction): Delete.
14362 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
14364 * function.c (assign_parm_find_stack_rtl): Likewise.
14365 (assign_parm_setup_block_p): Likewise.
14366 (assign_parm_setup_block): Likewise.
14367 (gimplify_parameters): Likewise.
14368 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
14369 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
14371 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
14372 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
14373 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
14374 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
14375 (aarch64_function_arg_padding): ...this new function.
14376 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
14377 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14378 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
14379 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
14380 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
14381 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14382 (arm_pad_arg_upward): Replace with...
14383 (arm_function_arg_padding): ...this new function.
14384 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
14386 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
14387 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
14388 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14389 (ia64_hpux_function_arg_padding): Replace with...
14390 (ia64_function_arg_padding): ...this new function. Use pad_direction
14391 instead of direction. Check for TARGET_HPUX.
14392 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
14393 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14394 (iq2000_function_arg_padding): New function.
14395 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
14396 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
14397 (mips_function_arg_padding): ...this new function.
14398 (mips_pad_reg_upward): Update accordingly.
14399 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14400 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
14401 targetm.calls.function_arg_padding.
14402 (FUNCTION_ARG_PADDING): Delete.
14403 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
14404 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
14405 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
14406 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
14407 (nios2_block_reg_padding): Return pad_direction instead of direction.
14408 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
14409 instead of direction.
14410 (nios2_function_arg_padding): Likewise. Make static.
14411 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14412 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
14413 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
14414 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
14415 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14416 (pa_function_arg_padding): Make static. Return pad_direction instead
14418 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
14419 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
14420 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
14421 instead of direction. Use targetm.calls.function_arg_padding.
14422 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
14423 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
14424 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
14425 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
14426 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
14428 (function_arg_padding): Rename to...
14429 (rs6000_function_arg_padding): ...this. Make static. Return
14430 pad_direction instead of direction.
14431 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
14432 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
14433 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
14434 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
14435 instead of direction. Use targetm.calls.function_arg_padding.
14436 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
14437 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
14438 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
14439 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
14440 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14441 (function_arg_padding): Rename to...
14442 (rs6000_function_arg_padding): ...this. Make static. Return
14443 pad_direction instead of direction.
14444 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
14445 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
14446 * config/s390/s390.c (s390_function_arg_padding): New function.
14447 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14448 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
14449 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
14450 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14451 (function_arg_padding): Rename to...
14452 (sparc_function_arg_padding): ...this. Make static. Return
14453 pad_direction instead of direction.
14454 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
14455 * config/spu/spu.c (spu_function_arg_padding): New function.
14456 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14457 * system.h (FUNCTION_ARG_PADDING): Poison.
14459 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14460 Alan Hayward <alan.hayward@arm.com>
14461 David Sherwood <david.sherwood@arm.com>
14463 * target.def (modes_tieable_p): New hook.
14464 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
14465 (TARGET_MODES_TIEABLE_P): ...this.
14466 * doc/tm.texi.in: Regenerate.
14467 * hooks.h (hook_bool_mode_mode_true): Declare.
14468 * hooks.c (hook_bool_mode_mode_true): New function.
14469 * combine.c (subst): Use targetm.modes_tieable_p instead of
14471 * dse.c (find_shift_sequence): Likewise.
14472 * expmed.c (extract_low_bits): Likewise.
14473 * lower-subreg.c: Include target.h.
14474 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
14476 * rtlanal.c (rtx_cost): Likewise.
14477 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
14478 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
14479 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
14480 (TARGET_MODES_TIEABLE_P): Redefine.
14481 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
14482 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
14483 (TARGET_MODES_TIEABLE_P): Redefine.
14484 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
14485 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
14486 (arc_modes_tieable_p): New function.
14487 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
14488 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
14489 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
14490 (arm_modes_tieable_p): Make static.
14491 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
14492 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
14493 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
14494 (TARGET_MODES_TIEABLE_P): Redefine.
14495 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
14496 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
14497 (TARGET_MODES_TIEABLE_P): Redefine.
14498 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
14499 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
14500 (cr16_modes_tieable_p): New function.
14501 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
14502 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
14503 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
14504 (TRULY_NOOP_TRUNCATION): Update comment.
14505 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
14506 (TRULY_NOOP_TRUNCATION): Update comment.
14507 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
14508 (frv_modes_tieable_p): New function.
14509 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
14510 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
14511 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
14512 (TARGET_MODES_TIEABLE_P): Redefine.
14513 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
14514 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
14515 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
14516 (TARGET_MODES_TIEABLE_P): Redefine.
14517 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
14518 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
14519 (ia64_modes_tieable_p): New function.
14520 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
14521 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
14522 (iq2000_modes_tieable_p): New function.
14523 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
14524 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
14525 (lm32_modes_tieable_p): New function.
14526 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
14527 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
14528 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
14529 (TARGET_MODES_TIEABLE_P): Redefine.
14530 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
14531 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
14532 (m32r_modes_tieable_p): New function.
14533 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
14534 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
14535 (m68k_modes_tieable_p): New function.
14536 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
14537 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
14538 (mcore_modes_tieable_p): New function.
14539 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
14540 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
14542 (TARGET_MODES_TIEABLE_P): Redefine.
14543 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
14544 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
14545 * config/mips/mips.c (mips_modes_tieable_p): Make static.
14546 (TARGET_MODES_TIEABLE_P): Redefine.
14547 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
14548 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
14549 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
14550 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
14551 (mn10300_modes_tieable_p): ...this and make static.
14552 (TARGET_MODES_TIEABLE_P): Redefine.
14553 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
14554 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
14555 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
14556 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
14557 (msp430_modes_tieable_p): Make static.
14558 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
14559 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
14560 (TARGET_MODES_TIEABLE_P): Redefine.
14561 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
14562 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
14563 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
14564 (TARGET_MODES_TIEABLE_P): Redefine.
14565 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
14566 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
14567 * config/pa/pa.c (pa_modes_tieable_p): Make static.
14568 (TARGET_MODES_TIEABLE_P): Redefine.
14569 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
14570 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
14571 (pdp11_modes_tieable_p): New function.
14572 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
14573 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
14574 (rs6000_modes_tieable_p): New function.
14575 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
14576 * config/powerpcspe/powerpcspe.md: Update comment.
14577 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
14578 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
14579 (TARGET_MODES_TIEABLE_P): Redefine.
14580 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
14581 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
14582 (rl78_modes_tieable_p): New function.
14583 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
14584 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
14585 (rs6000_modes_tieable_p): New function.
14586 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
14587 * config/rs6000/rs6000.md: Update comment.
14588 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
14589 * config/rx/rx.c (rx_modes_tieable_p): New function.
14590 (TARGET_MODES_TIEABLE_P): Redefine.
14591 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
14592 * config/s390/s390.c (s390_modes_tieable_p): New function.
14593 (TARGET_MODES_TIEABLE_P): Redefine.
14594 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
14595 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
14596 (sh_modes_tieable_p): New function.
14597 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
14598 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
14599 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
14600 (sparc_modes_tieable_p): Make static.
14601 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
14602 * config/spu/spu.c (spu_modes_tieable_p): New function.
14603 (TARGET_MODES_TIEABLE_P): Redefine.
14604 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
14605 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
14606 (TARGET_MODES_TIEABLE_P): Redefine.
14607 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
14608 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
14609 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
14610 * config/v850/v850.c (v850_modes_tieable_p): New function.
14611 (TARGET_MODES_TIEABLE_P): Redefine.
14612 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
14613 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
14614 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
14615 (visium_modes_tieable_p): New function.
14616 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
14617 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
14618 (xtensa_modes_tieable_p): New function.
14619 * system.h (MODES_TIEABLE_P): Poison.
14621 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14622 Alan Hayward <alan.hayward@arm.com>
14623 David Sherwood <david.sherwood@arm.com>
14625 * target.def (hard_regno_mode_ok): New hook.
14626 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
14627 (TARGET_HARD_REGNO_MODE_OK): ...this.
14628 * doc/tm.texi.in: Regenerate.
14629 * hooks.h (hook_bool_uint_mode_true): Declare.
14630 * hooks.c (hook_bool_uint_mode_true): New function.
14631 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
14632 HARD_REGNO_MODE_OK.
14633 * genpreds.c (write_insn_preds_c): Add an include of target.h.
14634 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
14635 instead of HARD_REGNO_MODE_OK.
14636 * caller-save.c: Include target.h.
14637 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
14638 HARD_REGNO_MODE_OK.
14639 * combine.c (can_combine_p): Likewise.
14640 (combinable_i3pat): Likewise.
14641 (can_change_dest_mode): Likewise.
14642 * expr.c (init_expr_target): Likewise.
14643 (convert_move): Likewise.
14644 (convert_modes): Likewise.
14645 * ira.c (setup_prohibited_class_mode_regs): Likewise.
14646 (setup_prohibited_mode_move_regs): Likewise.
14647 * ira.h (target_ira): Likewise.
14648 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14649 * lra-constraints.c (process_alt_operands): Likewise.
14650 (split_reg): Likewise.
14651 * recog.c (peep2_find_free_register): Likewise.
14652 * ree.c (combine_reaching_defs): Likewise.
14653 * regcprop.c (maybe_mode_change): Likewise.
14654 * reginfo.c (init_reg_sets_1): Likewise.
14655 (choose_hard_reg_mode): Likewise.
14656 (simplifiable_subregs): Likewise.
14657 * regrename.c (check_new_reg_p): Likewise.
14658 * reload.c (find_valid_class): Likewise.
14659 (find_valid_class_1): Likewise.
14660 (reload_inner_reg_of_subreg): Likewise.
14661 (push_reload): Likewise.
14662 (combine_reloads): Likewise.
14663 (find_dummy_reload): Likewise.
14664 (find_reloads): Likewise.
14665 * reload1.c (find_reg): Likewise.
14666 (set_reload_reg): Likewise.
14667 (allocate_reload_reg): Likewise.
14668 (choose_reload_regs): Likewise.
14669 (reload_adjust_reg_for_temp): Likewise.
14670 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
14671 (simplify_subreg_regno): Likewise.
14672 * sel-sched.c (init_regs_for_mode): Likewise.
14673 * varasm.c (make_decl_rtl): Likewise.
14674 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
14675 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
14676 HARD_REGNO_MODE_OK.
14677 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
14678 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
14679 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14680 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
14681 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
14682 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14683 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
14684 (arc_mode_class): Delete.
14685 (HARD_REGNO_MODE_OK): Delete.
14686 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14687 (arc_hard_regno_mode_ok): Rename old array to...
14688 (arc_hard_regno_mode_ok_modes): ...this.
14689 (arc_conditional_register_usage): Update accordingly.
14690 (arc_mode_class): Make static.
14691 (arc_hard_regno_mode_ok): New function.
14692 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
14693 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
14694 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14695 (arm_hard_regno_mode_ok): Make static.
14696 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
14697 HARD_REGNO_MODE_OK.
14698 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
14699 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
14700 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
14702 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14703 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
14704 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
14705 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
14706 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
14707 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14708 * config/bfin/predicates.md (valid_reg_operand): Use
14709 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
14710 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
14711 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
14712 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14713 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
14714 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
14715 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14716 (cr16_hard_regno_mode_ok): Make static and return a bool.
14717 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
14718 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14719 (cris_hard_regno_mode_ok): New function.
14720 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
14721 (epiphany_mode_class): Delete.
14722 (HARD_REGNO_MODE_OK): Delete.
14723 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
14724 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14725 (hard_regno_mode_ok): Rename to...
14726 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
14727 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
14728 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
14729 HARD_REGNO_MODE_OK.
14730 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
14731 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
14732 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14733 (frv_hard_regno_mode_ok): Make static and return a bool.
14734 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
14735 HARD_REGNO_MODE_OK.
14736 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
14737 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
14738 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
14739 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
14741 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14742 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
14743 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
14744 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
14746 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14747 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
14748 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14749 (ia64_hard_regno_mode_ok): New function.
14750 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
14751 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14752 (iq2000_hard_regno_mode_ok): New function.
14753 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
14754 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14755 (lm32_hard_regno_mode_ok): New function.
14756 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
14757 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
14758 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
14759 instead of HARD_REGNO_MODE_OK.
14760 (m32c_hard_regno_ok): Rename to...
14761 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
14762 (m32c_cannot_change_mode_class): Update accordingly.
14763 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14764 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
14765 (m32r_mode_class): Delete.
14766 (HARD_REGNO_MODE_OK): Delete.
14767 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14768 (m32r_hard_regno_mode_ok): Rename to...
14769 (m32r_hard_regno_modes): ...this.
14770 (m32r_mode_class): Make static.
14771 (m32r_hard_regno_mode_ok): New function.
14772 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
14773 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
14774 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14775 (m68k_hard_regno_mode_ok): Make static.
14776 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
14777 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14778 (mcore_hard_regno_mode_ok): New function.
14779 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
14780 (HARD_REGNO_MODE_OK): Delete.
14781 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
14783 (microblaze_hard_regno_mode_ok_p): ...this and make static.
14784 (microblaze_hard_regno_mode_ok): New function.
14785 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14786 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
14787 (mips_hard_regno_mode_ok): Delete.
14788 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
14789 (mips_hard_regno_mode_ok_p): ...this and make static.
14790 (mips_hard_regno_mode_ok_p): Rename to...
14791 (mips_hard_regno_mode_ok_uncached): ...this.
14792 (mips_hard_regno_mode_ok): New function.
14793 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
14794 of HARD_REGNO_MODE_OK.
14795 (mips_option_override): Update after above name changes.
14796 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14797 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
14798 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
14799 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
14800 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
14801 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14802 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
14803 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
14804 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
14805 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14806 (msp430_hard_regno_mode_ok): Make static and return a bool.
14807 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
14808 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
14809 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
14811 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14812 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
14813 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
14814 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
14815 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
14816 (PA_HARD_REGNO_MODE_OK): ...this
14817 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
14818 (PA_HARD_REGNO_MODE_OK): ...this.
14819 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14820 (pa_hard_regno_mode_ok): New function.
14821 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
14822 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14823 (pdp11_hard_regno_mode_ok): New function.
14824 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
14825 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
14827 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
14829 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14830 (rs6000_hard_regno_mode_ok): Rename to...
14831 (rs6000_hard_regno_mode_ok_uncached): ...this.
14832 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14833 (rs6000_hard_regno_mode_ok): New function.
14834 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
14835 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
14836 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
14837 (riscv_hard_regno_mode_ok): ...this and make static.
14838 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14839 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
14840 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
14841 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14842 (rl78_hard_regno_mode_ok): Make static and return bool.
14843 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
14844 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
14846 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
14847 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14848 (rs6000_hard_regno_mode_ok): Rename to...
14849 (rs6000_hard_regno_mode_ok_uncached): ...this.
14850 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14851 (rs6000_hard_regno_mode_ok): New function.
14852 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
14853 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
14854 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14855 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
14856 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
14857 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
14858 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14859 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
14860 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
14861 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14862 (sh_hard_regno_mode_ok): Make static.
14863 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
14864 instead of HARD_REGNO_MODE_OK.
14865 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
14866 (sparc_mode_class): Delete.
14867 (HARD_REGNO_MODE_OK): Delete.
14868 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14869 (hard_regno_mode_classes): Make static.
14870 (sparc_mode_class): Likewise.
14871 (sparc_hard_regno_mode_ok): New function.
14872 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
14873 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
14874 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
14876 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14877 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
14878 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
14879 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
14880 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
14881 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14882 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
14883 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
14884 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14885 (visium_hard_regno_mode_ok): New function.
14886 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
14887 instead of HARD_REGNO_MODE_OK.
14888 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
14889 (HARD_REGNO_MODE_OK): Delete.
14890 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
14891 (xtensa_hard_regno_mode_ok_p): ...this and make static.
14892 (xtensa_option_override): Update accordingly.
14893 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14894 (xtensa_hard_regno_mode_ok): New function.
14895 * system.h (HARD_REGNO_MODE_OK): Poison.
14897 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14898 Alan Hayward <alan.hayward@arm.com>
14899 David Sherwood <david.sherwood@arm.com>
14901 * target.def (hard_regno_call_part_clobbered): New hook.
14902 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
14903 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
14904 * doc/tm.texi: Regenerate.
14905 * hooks.h (hook_bool_uint_mode_false): Declare.
14906 * hooks.c (hook_bool_uint_mode_false): New function.
14907 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14908 * cselib.c (cselib_process_insn): Use
14909 targetm.hard_regno_call_part_clobbered instead of
14910 HARD_REGNO_CALL_PART_CLOBBERED.
14911 * ira-conflicts.c (ira_build_conflicts): Likewise.
14912 * ira-costs.c (ira_tune_allocno_costs): Likewise.
14913 * lra-constraints.c (need_for_call_save_p): Likewise.
14914 * lra-lives.c: Include target.h.
14915 (check_pseudos_live_through_calls): Use
14916 targetm.hard_regno_call_part_clobbered instead of
14917 HARD_REGNO_CALL_PART_CLOBBERED.
14918 * regcprop.c: Include target.h.
14919 (copyprop_hardreg_forward_1): Use
14920 targetm.hard_regno_call_part_clobbered instead of
14921 HARD_REGNO_CALL_PART_CLOBBERED.
14922 * reginfo.c (choose_hard_reg_mode): Likewise.
14923 * regrename.c (check_new_reg_p): Likewise.
14924 * reload.c (find_equiv_reg): Likewise.
14925 * reload1.c (emit_reload_insns): Likewise.
14926 * sched-deps.c (deps_analyze_insn): Likewise.
14927 * sel-sched.c (init_regs_for_mode): Likewise.
14928 (mark_unavailable_hard_regs): Likewise.
14929 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
14930 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14931 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
14933 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14934 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14935 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
14937 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
14939 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14940 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14941 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
14943 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14944 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14945 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
14947 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14948 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
14950 * config/powerpcspe/powerpcspe.c
14951 (rs6000_hard_regno_call_part_clobbered): New function.
14952 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14953 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14954 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
14956 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14957 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14958 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
14960 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14961 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14962 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
14964 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14965 Alan Hayward <alan.hayward@arm.com>
14966 David Sherwood <david.sherwood@arm.com>
14968 * rtl.h (subreg_memory_offset): Declare.
14969 * emit-rtl.c (subreg_memory_offset): New function.
14970 * expmed.c (store_bit_field_1): Use it.
14971 * expr.c (undefined_operand_subword_p): Likewise.
14972 * simplify-rtx.c (simplify_subreg): Likewise.
14974 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
14976 PR rtl-optimization/57448
14979 * optabs.c (expand_atomic_load): Place compiler memory barriers if
14980 using atomic_load pattern.
14981 (expand_atomic_store): Likewise.
14983 2017-09-04 Jakub Jelinek <jakub@redhat.com>
14986 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
14987 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
14988 handling. Use replace_call_with_value with NULL instead of
14989 gsi_replace, unlink_stmt_vdef and release_defs.
14991 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
14994 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
14996 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14999 * rtl.h (emit_library_call_value_1): Declare.
15000 (emit_library_call): Replace declaration with a series of overloads.
15001 Remove the parameter count argument.
15002 (emit_library_call_value): Likewise.
15003 * calls.c (emit_library_call_value_1): Make global. Replace varargs
15004 with an "rtx_mode_t *".
15005 (emit_library_call_value): Delete.
15006 (emit_library_call): Likewise.
15007 * asan.c (asan_emit_stack_protection): Update calls accordingly.
15008 (asan_emit_allocas_unpoison): Likewise.
15009 * builtins.c (expand_builtin_powi): Likewise.
15010 (expand_asan_emit_allocas_unpoison): Likewise.
15011 * cfgexpand.c (expand_main_function): Likewise.
15012 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
15013 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
15014 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
15015 * config/arm/arm.c (arm_trampoline_init): Likewise.
15016 (arm_call_tls_get_addr): Likewise.
15017 (arm_expand_divmod_libfunc): Likewise.
15018 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
15019 (smulsi3_highpart): Likewise.
15020 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
15021 (c6x_expand_compare): Likewise.
15022 (c6x_expand_movmem): Likewise.
15023 * config/frv/frv.c (frv_trampoline_init): Likewise.
15024 * config/i386/i386.c (ix86_trampoline_init): Likewise.
15025 (ix86_expand_divmod_libfunc): Likewise.
15026 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
15027 (ia64_expand_compare): Likewise.
15028 (ia64_profile_hook): Likewise.
15029 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
15030 (nonlocal_goto): Likewise.
15031 (restore_stack_nonlocal): Likewise.
15032 * config/m32r/m32r.c (block_move_call): Likewise.
15033 (m32r_trampoline_init): Likewise.
15034 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
15035 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
15036 (m68k_call_m68k_read_tp): Likewise.
15037 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
15038 (microblaze_expand_divide): Likewise.
15039 * config/mips/mips.h (mips_args): Likewise.
15040 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
15041 (MIPS_ICACHE_SYNC): Likewise.
15042 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
15043 (nios2_trampoline_init): Likewise.
15044 * config/pa/pa.c (hppa_tls_call): Likewise.
15045 (pa_trampoline_init): Likewise.
15046 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
15047 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
15048 (expand_strn_compare): Likewise.
15049 (rs6000_generate_compare): Likewise.
15050 (rs6000_expand_float128_convert): Likewise.
15051 (output_profile_hook): Likewise.
15052 (rs6000_trampoline_init): Likewise.
15053 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
15054 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
15055 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
15056 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
15057 (rs6000_generate_compare): Likewise.
15058 (rs6000_expand_float128_convert): Likewise.
15059 (output_profile_hook): Likewise.
15060 (rs6000_trampoline_init): Likewise.
15061 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
15062 * config/sh/sh.c (sh_trampoline_init): Likewise.
15063 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
15064 (sparc_emit_float_lib_cmp): Likewise.
15065 (sparc32_initialize_trampoline): Likewise.
15066 (sparc64_initialize_trampoline): Likewise.
15067 (sparc_profile_hook): Likewise.
15068 * config/spu/spu.c (ea_load_store): Likewise.
15069 * config/spu/spu.md (floatunssidf2): Likewise.
15070 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
15071 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
15072 * config/visium/visium.c (expand_block_move_4): Likewise.
15073 (expand_block_move_2): Likewise.
15074 (expand_block_move_1): Likewise.
15075 (expand_block_set_4): Likewise.
15076 (expand_block_set_2): Likewise.
15077 (expand_block_set_1): Likewise.
15078 (visium_trampoline_init): Likewise.
15079 (visium_profile_hook): Likewise.
15080 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
15081 (xtensa_setup_frame_addresses): Likewise.
15082 (xtensa_trampoline_init): Likewise.
15083 * except.c (sjlj_emit_function_enter): Likewise.
15084 (sjlj_emit_function_exit): Likewise.
15085 * explow.c (allocate_dynamic_stack_space): Likewise.
15086 (probe_stack_range): Likewise.
15087 * expr.c (convert_mode_scalar): Likewise.
15088 * optabs.c (expand_binop): Likewise.
15089 (expand_twoval_binop_libfunc): Likewise.
15090 (expand_unop): Likewise.
15091 (prepare_cmp_insn): Likewise.
15092 (prepare_float_lib_cmp): Likewise.
15093 (expand_float): Likewise.
15094 (expand_fix): Likewise.
15095 (expand_fixed_convert): Likewise.
15096 (maybe_emit_sync_lock_test_and_set): Likewise.
15097 (expand_atomic_compare_and_swap): Likewise.
15098 (expand_mem_thread_fence): Likewise.
15099 (expand_atomic_fetch_op): Likewise.
15101 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
15103 * doc/generic.texi (OpenACC): Adjust URL.
15104 * doc/invoke.texi (C Dialect Options): Ditto.
15106 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
15108 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
15109 predicate for operand 1. Add (m,<S>) constraint.
15110 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
15111 Prevent memory operand 1 with register operand 2.
15113 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
15115 PR rtl-optimization/82024
15116 * combine.c (try_combine): If the combination result is a PARALLEL,
15117 and we only need to retain the SET in there that would be placed
15118 at I2, check that we can place that at I3 instead, before doing so.
15120 2017-09-01 Jakub Jelinek <jakub@redhat.com>
15123 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
15125 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
15126 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
15129 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
15130 Jeff Law <law@redhat.com>
15132 * varasm.c (bss_initializer_p): Do not put constants into .bss
15133 (categorize_decl_for_section): Handle bss_initializer_p returning
15134 false when DECL_INITIAL is NULL.
15136 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15139 * config/s390/s390.c (s390_can_inline_p): New function.
15141 2017-09-01 Jeff Law <law@redhat.com>
15143 PR tree-optimization/82052
15144 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
15145 Always initialize the returned slot after a hash table miss
15146 when INSERT is true.
15148 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
15150 * config/s390/s390.md (mem_signal_fence): Remove.
15151 * doc/md.texi (mem_signal_fence): Remove.
15152 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
15154 * target-insns.def (mem_signal_fence): Remove.
15156 2017-09-01 Jakub Jelinek <jakub@redhat.com>
15159 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
15162 * asan.c (create_odr_indicator): Strip name encoding from assembler
15163 name before appending it after __odr_asan_.
15165 2017-09-01 Martin Liska <mliska@suse.cz>
15167 PR tree-optimization/82059
15168 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
15169 frequency only when an edge is redirected.
15171 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15173 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
15174 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
15175 (arc_conditional_register_usage): Remove ARC600 lp_count
15177 (arc_file_start): Emit Tag_ARC_CPU_variation.
15178 (arc_can_use_doloop_p): New conditions to use ZOLs.
15179 (hwloop_fail): New function.
15180 (hwloop_optimize): Likewise.
15181 (hwloop_pattern_reg): Likewise.
15182 (arc_doloop_hooks): New struct, to be used with reorg_loops.
15183 (arc_reorg_loops): New function, calls reorg_loops.
15184 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
15185 (arc600_corereg_hazard): Remove ZOL checking, case handled by
15187 (arc_loop_hazard): Remove function, functionality moved into
15189 (arc_hazard): Remove arc_loop_hazard call.
15190 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
15191 into hwloop_optimize.
15192 (arc_label_align): Remove ZOL handling.
15193 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
15194 * config/arc/arc.md (doloop_begin): Remove pattern.
15195 (doloop_begin_i): Likewise.
15196 (doloop_end_i): Likewise.
15197 (doloop_fallback): Likewise.
15198 (doloop_fallback_m): Likewise.
15199 (doloop_end): Reimplement expand.
15200 (arc_lp): New pattern for LP instruction.
15201 (loop_end): New pattern.
15202 (loop_fail): Likewise.
15203 (decrement_and_branch_until_zero): Likewise.
15204 * config/arc/arc.opt (mlpc-width): New option.
15205 * doc/invoke.texi (mlpc-width): Document option.
15207 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15209 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
15210 (arc_ccfsm_advance): Fix checking for delay slots.
15211 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
15213 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15215 * config/arc/arc.md (movqi_insn): Add stores to save constant long
15217 (movhi_insn): Update store instruction constraint which are saving
15218 6-bit short immediates.
15219 (movsi_insn): Consider also short scaled load operations.
15220 (zero_extendhisi2_i): Use Usd constraint instead of T.
15221 (extendhisi2_i): Add q constraint.
15222 (arc_clzsi2): Add type and length attributes.
15223 (arc_ctzsi2): Likewise.
15224 * config/arc/constraints.md (Usc): Update constraint, the
15225 assembler can parse two relocations for a single instruction.
15227 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15229 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
15230 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
15232 2017-08-31 Olivier Hainque <hainque@adacore.com>
15234 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
15235 match as powerpc-wrs-vxworks*.
15237 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
15239 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
15240 register constraint for by-element operand.
15241 (aarch64_mls_elt_merge<mode>): Likewise.
15243 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
15245 * config/arc/arc.c (arc_can_follow_jump): Check for short
15248 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
15250 * config.gcc: Use g.opt for arc.
15251 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
15252 functionality moved to ...
15253 (legitimate_scaled_address_p): New function, ...here.
15254 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
15255 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
15256 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
15258 (arc_override_options): Handle G option.
15259 (arc_output_pic_addr_const): Correct function definition.
15260 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
15261 (arc_decl_anon_ns_mem_p): Delete.
15262 (arc_in_small_data_p): Overhaul this function to take into
15263 consideration the value given via G option.
15264 (arc_rewrite_small_data_1): Renamed and corrected old
15265 arc_rewrite_small_data function.
15266 (arc_rewrite_small_data): New function.
15267 (small_data_pattern): Don't use pic_offset_table_rtx.
15268 * config/arc/arc.h (CC1_SPEC): Recognize G option.
15269 * config/arc/simdext.md (movmisalignv2hi): Use
15270 prepare_move_operands function.
15272 (movmisalign*): Likewise.
15273 * doc/invoke.texi (ARC options): Document -G option.
15275 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
15277 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
15279 * config/arc/arc.c (arc_print_operand): Output scalled address for
15280 sdata whenever is possible.
15281 (arc_in_small_data_p): Allow sdata for 64bit datum when double
15282 load/stores are available.
15283 (compact_sda_memory_operand): Check for the alignment required by
15284 code density instructions.
15285 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
15287 * config/arc/constraints.md (Usd): Update constraint.
15288 (Us0): New constraint.
15289 (Usc): Update constraint.
15291 2017-08-31 Richard Biener <rguenther@suse.de>
15293 PR middle-end/82054
15294 * dwarf2out.c (dwarf2out_early_global_decl): Process each
15295 function only once.
15297 2017-08-31 Tamar Christina <tamar.christina@arm.com>
15299 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
15300 Resize type_signature.
15302 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
15303 Alan Hayward <alan.hayward@arm.com>
15304 David Sherwood <david.sherwood@arm.com>
15306 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
15307 subregs whose inner modes can be stored in GPRs.
15308 (aarch64_classify_index): Likewise.
15310 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
15311 Alan Hayward <alan.hayward@arm.com>
15312 David Sherwood <david.sherwood@arm.com>
15314 * config/aarch64/iterators.md (V_cmp_result): Rename to...
15315 (V_INT_EQUIV): ...this.
15316 (v_cmp_result): Rename to...
15317 (v_int_equiv): ...this.
15318 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
15319 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
15320 (copysign<mode>3): Likewise.
15321 (aarch64_simd_bsl<mode>_internal): Likewise.
15322 (aarch64_simd_bsl<mode>): Likewise.
15323 (vec_cmp<mode><mode>): Likewise.
15324 (vcond<mode><mode>): Likewise.
15325 (vcond<v_cmp_mixed><mode>): Likewise.
15326 (vcondu<mode><v_cmp_mixed>): Likewise.
15327 (aarch64_cm<optab><mode>): Likewise.
15328 (aarch64_cmtst<mode>): Likewise.
15329 (aarch64_fac<optab><mode>): Likewise.
15330 (vec_perm_const<mode>): Likewise.
15331 (vcond_mask_<mode><v_cmp_result>): Rename to...
15332 (vcond_mask_<mode><v_int_equiv>): ...this.
15333 (vec_cmp<mode><v_cmp_result>): Rename to...
15334 (vec_cmp<mode><v_int_equiv>): ...this.
15336 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
15337 Alan Hayward <alan.hayward@arm.com>
15338 David Sherwood <david.sherwood@arm.com>
15340 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
15342 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
15343 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
15344 (UNSPEC_LD4_DREG): New unspecs.
15345 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
15346 (aarch64_ld2<mode>_dreg_be): Replace with...
15347 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
15349 (aarch64_ld3<mode>_dreg_le)
15350 (aarch64_ld3<mode>_dreg_be): Replace with...
15351 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
15353 (aarch64_ld4<mode>_dreg_le)
15354 (aarch64_ld4<mode>_dreg_be): Replace with...
15355 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
15358 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15360 PR tree-optimization/81987
15361 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
15362 insert an initializer in a location not dominated by the stride
15365 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
15367 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
15368 on the entire header of the finally block in the fallthru case.
15370 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
15372 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
15374 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
15376 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
15377 rs6000_emit_move_from_cr and call renamed function.
15378 (rs6000_emit_prologue): Call renamed functions.
15379 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
15380 movesi_from_cr, remove volatile CRs.
15382 2017-08-30 Jon Beniston <jon@beniston.com>
15383 Richard Biener <rguenther@suse.de>
15385 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
15386 of VECTOR_MODE_P check.
15387 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
15388 element vector types.
15390 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15392 * df.h (df_read_modify_subreg_p): Remove in favor of...
15393 * rtl.h (read_modify_subreg_p): ...this new function. Take a
15394 const_rtx instead of an rtx.
15395 * cprop.c (local_cprop_find_used_regs): Update accordingly.
15396 * df-problems.c (df_word_lr_mark_ref): Likewise.
15397 * ira-lives.c (mark_pseudo_reg_live): Likewise.
15398 (mark_pseudo_reg_dead): Likewise.
15399 (mark_ref_dead): Likewise.
15400 * reginfo.c (init_subregs_of_mode): Likewise.
15401 * sched-deps.c (sched_analyze_1): Likewise.
15402 * df-scan.c (df_def_record_1): Likewise.
15403 (df_uses_record): Likewise.
15404 (df_read_modify_subreg_p): Remove in favor of...
15405 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
15406 const_rtx instead of an rtx.
15408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15409 Alan Hayward <alan.hayward@arm.com>
15410 David Sherwood <david.sherwood@arm.com>
15412 * rtl.h (partial_subreg_p): New function.
15413 * caller-save.c (save_call_clobbered_regs): Use it.
15414 * calls.c (expand_call): Likewise.
15415 * combine.c (combinable_i3pat): Likewise.
15416 (simplify_set): Likewise.
15417 (make_extraction): Likewise.
15418 (make_compound_operation_int): Likewise.
15419 (gen_lowpart_or_truncate): Likewise.
15420 (force_to_mode): Likewise.
15421 (make_field_assignment): Likewise.
15422 (reg_truncated_to_mode): Likewise.
15423 (record_truncated_value): Likewise.
15424 (move_deaths): Likewise.
15425 * cse.c (record_jump_cond): Likewise.
15426 (cse_insn): Likewise.
15427 * cselib.c (cselib_lookup_1): Likewise.
15428 * expmed.c (extract_bit_field_using_extv): Likewise.
15429 * function.c (assign_parm_setup_reg): Likewise.
15430 * ifcvt.c (noce_convert_multiple_sets): Likewise.
15431 * ira-build.c (create_insn_allocnos): Likewise.
15432 * lra-coalesce.c (merge_pseudos): Likewise.
15433 * lra-constraints.c (match_reload): Likewise.
15434 (simplify_operand_subreg): Likewise.
15435 (curr_insn_transform): Likewise.
15436 * lra-lives.c (process_bb_lives): Likewise.
15437 * lra.c (new_insn_reg): Likewise.
15438 (lra_substitute_pseudo): Likewise.
15439 * regcprop.c (mode_change_ok): Likewise.
15440 (maybe_mode_change): Likewise.
15441 (copyprop_hardreg_forward_1): Likewise.
15442 * reload.c (push_reload): Likewise.
15443 (find_reloads): Likewise.
15444 (find_reloads_subreg_address): Likewise.
15445 * reload1.c (alter_reg): Likewise.
15446 (eliminate_regs_1): Likewise.
15447 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15449 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
15451 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
15452 back to if statements, including unpack.
15454 2017-08-30 Martin Liska <mliska@suse.cz>
15456 PR inline-asm/82001
15457 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
15459 (func_checker::compare_asm_inputs_outputs): ... this function.
15460 (func_checker::compare_gimple_asm): Use the function to compare
15461 also ASM constrains.
15462 * ipa-icf-gimple.h: Rename the function.
15464 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15465 Alan Hayward <alan.hayward@arm.com>
15466 David Sherwood <david.sherwood@arm.com>
15468 * coretypes.h (complex_mode): New type.
15469 * gdbhooks.py (build_pretty_printer): Handle it.
15470 * machmode.h (complex_mode): New class.
15471 (complex_mode::includes_p): New function.
15472 (is_complex_int_mode): Likewise.
15473 (is_complex_float_mode): Likewise.
15474 * genmodes.c (get_mode_class): Handle complex mode classes.
15475 * function.c (expand_function_end): Use is_complex_int_mode.
15477 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15478 Alan Hayward <alan.hayward@arm.com>
15479 David Sherwood <david.sherwood@arm.com>
15481 * coretypes.h (scalar_mode_pod): New typedef.
15482 * gdbhooks.py (build_pretty_printer): Handle it.
15483 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
15484 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
15485 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
15486 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
15487 as_a <scalar_mode>.
15489 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15490 Alan Hayward <alan.hayward@arm.com>
15491 David Sherwood <david.sherwood@arm.com>
15493 * machmode.h (mode_for_vector): Take a scalar_mode instead
15495 * stor-layout.c (mode_for_vector): Likewise.
15496 * explow.c (promote_mode): Use as_a <scalar_mode>.
15497 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
15499 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15500 Alan Hayward <alan.hayward@arm.com>
15501 David Sherwood <david.sherwood@arm.com>
15503 * target.def (preferred_simd_mode): Take a scalar_mode
15504 instead of a machine_mode.
15505 * targhooks.h (default_preferred_simd_mode): Likewise.
15506 * targhooks.c (default_preferred_simd_mode): Likewise.
15507 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
15508 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
15509 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
15510 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
15511 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
15512 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
15513 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
15514 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
15516 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
15517 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
15518 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
15519 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
15520 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
15521 * doc/tm.texi: Regenerate.
15522 * optabs-query.c (can_vec_mask_load_store_p): Return false for
15525 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15526 Alan Hayward <alan.hayward@arm.com>
15527 David Sherwood <david.sherwood@arm.com>
15529 * target.def (scalar_mode_supported_p): Take a scalar_mode
15530 instead of a machine_mode.
15531 * targhooks.h (default_scalar_mode_supported_p): Likewise.
15532 * targhooks.c (default_scalar_mode_supported_p): Likewise.
15533 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
15534 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
15535 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
15536 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
15537 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
15538 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
15539 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
15540 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
15541 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
15542 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
15543 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
15544 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
15546 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
15547 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
15548 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
15549 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
15550 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
15552 * doc/tm.texi: Regenerate.
15554 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15555 Alan Hayward <alan.hayward@arm.com>
15556 David Sherwood <david.sherwood@arm.com>
15558 * coretypes.h (opt_scalar_mode): New typedef.
15559 * gdbhooks.py (build_pretty_printers): Handle it.
15560 * machmode.h (mode_iterator::get_2xwider): Add overload for
15562 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
15564 * expr.c (convert_mode_scalar): Likewise.
15565 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15566 * optabs.c (expand_float): Likewise.
15567 (expand_fix): Likewise.
15568 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15570 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15571 Alan Hayward <alan.hayward@arm.com>
15572 David Sherwood <david.sherwood@arm.com>
15574 * optabs.c (expand_float): Explicitly check for scalars before
15575 using a branching expansion.
15576 (expand_fix): Likewise.
15578 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15579 Alan Hayward <alan.hayward@arm.com>
15580 David Sherwood <david.sherwood@arm.com>
15582 * expr.c (convert_mode): Split scalar handling out into...
15583 (convert_mode_scalar): ...this new function. Treat the modes
15586 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15587 Alan Hayward <alan.hayward@arm.com>
15588 David Sherwood <david.sherwood@arm.com>
15590 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
15592 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
15594 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15595 Alan Hayward <alan.hayward@arm.com>
15596 David Sherwood <david.sherwood@arm.com>
15598 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
15599 rather than a machine_mode.
15600 (fixed_from_string): Likewise.
15601 (fixed_convert): Likewise.
15602 (fixed_convert_from_int): Likewise.
15603 (fixed_convert_from_real): Likewise.
15604 (real_convert_from_fixed): Likewise.
15605 * fixed-value.c (fixed_from_double_int): Likewise.
15606 (fixed_from_string): Likewise.
15607 (fixed_convert): Likewise.
15608 (fixed_convert_from_int): Likewise.
15609 (fixed_convert_from_real): Likewise.
15610 (real_convert_from_fixed): Likewise.
15611 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
15613 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15614 Alan Hayward <alan.hayward@arm.com>
15615 David Sherwood <david.sherwood@arm.com>
15617 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
15618 of separate mode class checks. Do not allow vector modes here.
15619 (immed_wide_int_const): Use as_a <scalar_mode>.
15620 * explow.c (trunc_int_for_mode): Likewise.
15621 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
15622 (wi::shwi): Likewise.
15623 (wi::min_value): Likewise.
15624 (wi::max_value): Likewise.
15625 * dwarf2out.c (loc_descriptor): Likewise.
15626 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
15627 for CONST_WIDE_INT.
15629 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15630 Alan Hayward <alan.hayward@arm.com>
15631 David Sherwood <david.sherwood@arm.com>
15633 * tree.h (SCALAR_TYPE_MODE): New macro.
15634 * expr.c (expand_expr_addr_expr_1): Use it.
15635 (expand_expr_real_2): Likewise.
15636 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
15637 (fold_convert_const_fixed_from_int): Likewise.
15638 (fold_convert_const_fixed_from_real): Likewise.
15639 (native_encode_fixed): Likewise
15640 (native_encode_complex): Likewise
15641 (native_encode_vector): Likewise.
15642 (native_interpret_fixed): Likewise.
15643 (native_interpret_real): Likewise.
15644 (native_interpret_complex): Likewise.
15645 (native_interpret_vector): Likewise.
15646 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
15647 (simd_clone_adjust_argument_types): Likewise.
15648 (simd_clone_init_simd_arrays): Likewise.
15649 (simd_clone_adjust): Likewise.
15650 * stor-layout.c (layout_type): Likewise.
15651 * tree.c (build_minus_one_cst): Likewise.
15652 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15653 * tree-inline.c (estimate_move_cost): Likewise.
15654 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
15655 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
15656 (vectorizable_reduction): Likewise.
15657 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
15658 (vect_recog_mixed_size_cond_pattern): Likewise.
15659 (check_bool_pattern): Likewise.
15660 (adjust_bool_pattern): Likewise.
15661 (search_type_for_mask_1): Likewise.
15662 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
15663 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15664 (vectorizable_load): Likewise.
15665 (vectorizable_store): Likewise.
15666 * ubsan.c (ubsan_encode_value): Likewise.
15667 * varasm.c (output_constant): Likewise.
15669 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15670 Alan Hayward <alan.hayward@arm.com>
15671 David Sherwood <david.sherwood@arm.com>
15673 * coretypes.h (scalar_mode): New class.
15674 * machmode.h (scalar_mode): Likewise.
15675 (scalar_mode::includes_p): New function.
15676 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
15677 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
15678 * genmodes.c (get_mode_class): Handle remaining scalar modes.
15679 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
15680 * expmed.c (store_bit_field_1): Likewise.
15681 (extract_bit_field_1): Likewise.
15682 * expr.c (write_complex_part): Likewise.
15683 (read_complex_part): Likewise.
15684 (emit_move_complex_push): Likewise.
15685 (expand_expr_real_2): Likewise.
15686 * function.c (assign_parm_setup_reg): Likewise.
15687 (assign_parms_unsplit_complex): Likewise.
15688 * optabs.c (expand_binop): Likewise.
15689 * rtlanal.c (subreg_get_info): Likewise.
15690 * simplify-rtx.c (simplify_immed_subreg): Likewise.
15691 * varasm.c (output_constant_pool_2): Likewise.
15693 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15694 Alan Hayward <alan.hayward@arm.com>
15695 David Sherwood <david.sherwood@arm.com>
15697 * expmed.c (extract_high_half): Use scalar_int_mode and remove
15699 (expmed_mult_highpart_optab): Likewise.
15700 (expmed_mult_highpart): Likewise.
15702 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15703 Alan Hayward <alan.hayward@arm.com>
15704 David Sherwood <david.sherwood@arm.com>
15706 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
15707 instead of a machine_mode.
15708 (builtin_memset_read_str): Likewise.
15709 * builtins.c (c_readstr): Likewise.
15710 (builtin_memcpy_read_str): Likewise.
15711 (builtin_strncpy_read_str): Likewise.
15712 (builtin_memset_read_str): Likewise.
15713 (builtin_memset_gen_str): Likewise.
15714 (expand_builtin_signbit): Use scalar_int_mode for local variables.
15715 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
15716 instead of a machine_mode.
15717 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
15719 (make_extraction): Likewise.
15720 (try_widen_shift_mode): Take and return scalar_int_modes instead
15722 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
15723 a scalar_int_mode instead of a machine_mode.
15724 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
15725 (avr_addr_space_pointer_mode): Likewise.
15726 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
15727 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
15728 (msp430_unwind_word_mode): Likewise.
15729 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
15730 (spu_addr_space_pointer_mode): Likewise.
15731 (spu_addr_space_address_mode): Likewise.
15732 (spu_libgcc_cmp_return_mode): Likewise.
15733 (spu_libgcc_shift_count_mode): Likewise.
15734 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
15735 (rl78_addr_space_pointer_mode): Likewise.
15736 (fl78_unwind_word_mode): Likewise.
15737 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
15739 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
15740 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
15741 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
15742 (mips_valid_pointer_mode): Likewise.
15743 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
15744 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
15745 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
15747 (ft32_addr_space_address_mode): Likewise.
15748 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
15749 scalar_int_mode instead of a machine_mode.
15750 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
15752 (m32c_addr_space_address_mode): Likewise.
15753 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
15754 (rs6000_eh_return_filter_mode): Likewise.
15755 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
15756 (rs6000_eh_return_filter_mode): Likewise.
15757 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
15758 (s390_libgcc_shift_count_mode): Likewise.
15759 (s390_unwind_word_mode): Likewise.
15760 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
15762 * target.def (mode_rep_extended): Likewise.
15763 (valid_pointer_mode): Likewise.
15764 (addr_space.valid_pointer_mode): Likewise.
15765 (eh_return_filter_mode): Return a scalar_int_mode rather than
15767 (libgcc_cmp_return_mode): Likewise.
15768 (libgcc_shift_count_mode): Likewise.
15769 (unwind_word_mode): Likewise.
15770 (addr_space.pointer_mode): Likewise.
15771 (addr_space.address_mode): Likewise.
15772 * doc/tm.texi: Regenerate.
15773 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
15775 (do_jump): Use scalar_int_mode for local variables.
15776 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
15777 rather than a machine_mode.
15778 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
15779 (scompare_loc_descriptor_wide): Likewise.
15780 (scompare_loc_descriptor_narrow): Likewise.
15781 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
15783 * except.c (sjlj_emit_dispatch_table): Likewise.
15784 (expand_builtin_eh_copy_values): Likewise.
15785 * explow.c (convert_memory_address_addr_space_1): Likewise.
15786 Take a scalar_int_mode rather than a machine_mode.
15787 (convert_memory_address_addr_space): Take a scalar_int_mode rather
15788 than a machine_mode.
15789 (memory_address_addr_space): Use scalar_int_mode for local variables.
15790 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
15791 rather than a machine_mode.
15792 * expmed.c (mask_rtx): Likewise.
15793 (init_expmed_one_conv): Likewise.
15794 (expand_mult_highpart_adjust): Likewise.
15795 (extract_high_half): Likewise.
15796 (expmed_mult_highpart_optab): Likewise.
15797 (expmed_mult_highpart): Likewise.
15798 (expand_smod_pow2): Likewise.
15799 (expand_sdiv_pow2): Likewise.
15800 (emit_store_flag_int): Likewise.
15801 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
15803 (extract_low_bits): Likewise.
15804 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
15806 * expr.c (pieces_addr::adjust): Likewise.
15807 (can_store_by_pieces): Likewise.
15808 (store_by_pieces): Likewise.
15809 (clear_by_pieces_1): Likewise.
15810 (expand_expr_addr_expr_1): Likewise.
15811 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
15812 (expand_expr_real_1): Likewise.
15813 (try_casesi): Likewise.
15814 * final.c (shorten_branches): Likewise.
15815 * fold-const.c (fold_convert_const_int_from_fixed): Change the
15816 type of "mode" to machine_mode.
15817 * internal-fn.c (expand_arith_overflow_result_store): Take a
15818 scalar_int_mode rather than a machine_mode.
15819 (expand_mul_overflow): Use scalar_int_mode for local variables.
15820 * loop-doloop.c (doloop_modify): Likewise.
15821 (doloop_optimize): Likewise.
15822 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
15823 than a machine_mode.
15824 (expand_doubleword_shift_condmove): Likewise.
15825 (expand_doubleword_shift): Likewise.
15826 (expand_doubleword_clz): Likewise.
15827 (expand_doubleword_popcount): Likewise.
15828 (expand_doubleword_parity): Likewise.
15829 (expand_absneg_bit): Use scalar_int_mode for local variables.
15830 (prepare_float_lib_cmp): Likewise.
15831 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
15832 rather than a machine_mode.
15833 (convert_memory_address_addr_space): Likewise.
15834 (get_mode_bounds): Likewise.
15835 (get_address_mode): Return a scalar_int_mode rather than a
15837 * rtlanal.c (get_address_mode): Likewise.
15838 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
15839 than a machine_mode.
15840 * targhooks.c (default_mode_rep_extended): Likewise.
15841 (default_valid_pointer_mode): Likewise.
15842 (default_addr_space_valid_pointer_mode): Likewise.
15843 (default_eh_return_filter_mode): Return a scalar_int_mode rather
15844 than a machine_mode.
15845 (default_libgcc_cmp_return_mode): Likewise.
15846 (default_libgcc_shift_count_mode): Likewise.
15847 (default_unwind_word_mode): Likewise.
15848 (default_addr_space_pointer_mode): Likewise.
15849 (default_addr_space_address_mode): Likewise.
15850 * targhooks.h (default_eh_return_filter_mode): Likewise.
15851 (default_libgcc_cmp_return_mode): Likewise.
15852 (default_libgcc_shift_count_mode): Likewise.
15853 (default_unwind_word_mode): Likewise.
15854 (default_addr_space_pointer_mode): Likewise.
15855 (default_addr_space_address_mode): Likewise.
15856 (default_mode_rep_extended): Take a scalar_int_mode rather than
15858 (default_valid_pointer_mode): Likewise.
15859 (default_addr_space_valid_pointer_mode): Likewise.
15860 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
15862 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
15863 rather than a machine_mode.
15864 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
15865 for local variables.
15866 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15867 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
15868 than a machine_mode.
15870 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15871 Alan Hayward <alan.hayward@arm.com>
15872 David Sherwood <david.sherwood@arm.com>
15874 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
15875 the mode argument to scalar_int_mode.
15876 (do_jump_by_parts_zero_rtx): Likewise.
15877 (do_jump_by_parts_equality_rtx): Likewise.
15878 (do_jump_by_parts_greater): Take a mode argument.
15879 (do_jump_by_parts_equality): Likewise.
15880 (do_jump_1): Update calls accordingly.
15882 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15883 Alan Hayward <alan.hayward@arm.com>
15884 David Sherwood <david.sherwood@arm.com>
15886 * is-a.h (safe_dyn_cast): New function.
15887 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
15888 (jump_table_for_label): Likewise.
15889 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
15890 instead of an rtx_insn *.
15891 (shorten_branches): Use dyn_cast instead of LABEL_P and
15892 JUMP_TABLE_DATA_P. Use jump_table_for_label and
15893 rtx_jump_table_data::get_data_mode.
15894 (final_scan_insn): Likewise.
15896 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15897 Alan Hayward <alan.hayward@arm.com>
15898 David Sherwood <david.sherwood@arm.com>
15900 * combine.c (try_combine): Use is_a <scalar_int_mode> when
15901 trying to combine a full-register integer set with a subreg
15904 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15905 Alan Hayward <alan.hayward@arm.com>
15906 David Sherwood <david.sherwood@arm.com>
15908 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
15909 that is always either address_mode or pointer_mode.
15911 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15912 Alan Hayward <alan.hayward@arm.com>
15913 David Sherwood <david.sherwood@arm.com>
15915 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
15916 when the two are known to be equal.
15918 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15919 Alan Hayward <alan.hayward@arm.com>
15920 David Sherwood <david.sherwood@arm.com>
15922 * simplify-rtx.c (simplify_const_unary_operation): Use
15923 is_a <scalar_int_mode> instead of checking for a nonzero
15924 precision. Forcibly convert op_mode to a scalar_int_mode
15925 in that case. More clearly differentiate the operand and
15926 result modes and use the former when deciding what the value
15927 of a count-bits operation should be. Use is_int_mode instead
15928 of checking for a MODE_INT. Remove redundant check for whether
15929 this mode has a zero precision.
15931 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15932 Alan Hayward <alan.hayward@arm.com>
15933 David Sherwood <david.sherwood@arm.com>
15935 * optabs.c (widen_leading): Change the type of the mode argument
15936 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
15937 (widen_bswap): Likewise.
15938 (expand_parity): Likewise.
15939 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
15940 (expand_ffs): Likewise.
15941 (epand_unop): Check for scalar integer modes before calling the
15944 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15945 Alan Hayward <alan.hayward@arm.com>
15946 David Sherwood <david.sherwood@arm.com>
15948 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
15950 (expand_expr_real_1): Update call accordingly.
15952 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15953 Alan Hayward <alan.hayward@arm.com>
15954 David Sherwood <david.sherwood@arm.com>
15956 * expmed.c (store_bit_field_using_insv): Add op0_mode and
15957 value_mode arguments. Use scalar_int_mode internally.
15958 (store_bit_field_1): Rename the new integer mode from imode
15959 to op0_mode and use it instead of GET_MODE (op0). Update calls
15960 to store_split_bit_field, store_bit_field_using_insv and
15961 store_fixed_bit_field.
15962 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
15963 Use scalar_int_mode internally. Use a bit count rather than a mode
15964 when calculating the largest bit size for get_best_mode.
15965 Update calls to store_split_bit_field and store_fixed_bit_field_1.
15966 (store_fixed_bit_field_1): Add mode and value_mode arguments.
15967 Remove assertion that OP0 has a scalar integer mode.
15968 (store_split_bit_field): Add op0_mode and value_mode arguments.
15969 Update calls to extract_fixed_bit_field.
15970 (extract_bit_field_using_extv): Add an op0_mode argument.
15971 Use scalar_int_mode internally.
15972 (extract_bit_field_1): Rename the new integer mode from imode to
15973 op0_mode and use it instead of GET_MODE (op0). Update calls to
15974 extract_split_bit_field, extract_bit_field_using_extv and
15975 extract_fixed_bit_field.
15976 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
15977 to extract_split_bit_field and extract_fixed_bit_field_1.
15978 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
15979 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
15980 on the target mode.
15981 (extract_split_bit_field): Add an op0_mode argument. Update call
15982 to extract_fixed_bit_field.
15984 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15985 Alan Hayward <alan.hayward@arm.com>
15986 David Sherwood <david.sherwood@arm.com>
15988 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
15989 * explow.c (hard_function_value): Likewise.
15990 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
15991 convert_to_mode call outside the loop.
15992 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
15993 for the mode iterator. Require the mode specified by max_pieces
15995 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
15997 (copy_blkmode_to_reg): Likewise.
15998 (set_storage_via_setmem): Likewise.
15999 * optabs.c (prepare_cmp_insn): Likewise.
16000 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16001 * stor-layout.c (finish_bitfield_representative): Likewise.
16003 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16004 Alan Hayward <alan.hayward@arm.com>
16005 David Sherwood <david.sherwood@arm.com>
16007 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
16008 * expr.c (convert_move): Use them.
16009 (convert_modes): Likewise.
16010 (store_expr_with_bounds): Likewise.
16012 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16013 Alan Hayward <alan.hayward@arm.com>
16014 David Sherwood <david.sherwood@arm.com>
16016 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
16017 parameter for the mode of "x". Remove the "known_x", "known_mode"
16018 and "known_ret" arguments. Change the type of the mode argument
16019 to scalar_int_mode.
16020 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
16021 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
16022 (reg_num_sign_bit_copies_for_combine): Likewise.
16023 * rtlanal.c (nonzero_bits1): Likewise.
16024 (num_sign_bit_copies1): Likewise.
16025 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
16026 (reg_num_sign_bit_copies_general): Likewise.
16027 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
16028 (reg_nonzero_bits_general): Likewise.
16030 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16031 Alan Hayward <alan.hayward@arm.com>
16032 David Sherwood <david.sherwood@arm.com>
16034 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
16035 than in subroutines. Return 1 for non-integer modes.
16036 (cached_num_sign_bit_copies): Change the type of the mode parameter
16037 to scalar_int_mode.
16038 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
16039 classes. Handle CONST_INT_P first and then check whether X also
16040 has a scalar integer mode. Check the same thing for inner registers
16041 of a SUBREG and for values that are being extended or truncated.
16043 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16044 Alan Hayward <alan.hayward@arm.com>
16045 David Sherwood <david.sherwood@arm.com>
16047 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
16048 in subroutines. Return the mode mask for non-integer modes.
16049 (cached_nonzero_bits): Change the type of the mode parameter
16050 to scalar_int_mode.
16051 (nonzero_bits1): Likewise. Remove early exit for other mode
16052 classes. Handle CONST_INT_P first and then check whether X
16053 also has a scalar integer mode.
16055 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16056 Alan Hayward <alan.hayward@arm.com>
16057 David Sherwood <david.sherwood@arm.com>
16059 * expr.c (widest_int_mode_for_size): Make the comment match the code.
16060 Return a scalar_int_mode and assert that the size is greater than
16062 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
16063 (op_by_pieces_d::op_by_pieces_d): Likewise.
16064 (op_by_pieces_d::run): Likewise.
16065 (can_store_by_pieces): 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 * combine.c (extract_left_shift): Add a mode argument and update
16073 (make_compound_operation_int): Change the type of the mode parameter
16074 to scalar_int_mode and update the call to extract_left_shift.
16076 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16077 Alan Hayward <alan.hayward@arm.com>
16078 David Sherwood <david.sherwood@arm.com>
16080 * combine.c (simplify_and_const_int): Change the type of the mode
16081 parameter to scalar_int_mode.
16082 (simplify_and_const_int_1): Likewise. Update recursive call.
16084 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16085 Alan Hayward <alan.hayward@arm.com>
16086 David Sherwood <david.sherwood@arm.com>
16088 * combine.c (simplify_compare_const): Check that the mode is a
16089 scalar_int_mode (rather than VOIDmode) before testing its
16091 (simplify_comparison): Move COMPARISON_P handling out of the
16092 loop and restrict the latter part of the loop to scalar_int_modes.
16093 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
16094 and when considering SUBREG_REGs. Use is_int_mode instead of
16095 checking GET_MODE_CLASS against MODE_INT.
16097 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16098 Alan Hayward <alan.hayward@arm.com>
16099 David Sherwood <david.sherwood@arm.com>
16101 * combine.c (try_widen_shift_mode): Move check for equal modes to...
16102 (simplify_shift_const_1): ...here. Use scalar_int_mode for
16103 shift_unit_mode and for modes involved in scalar shifts.
16105 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16106 Alan Hayward <alan.hayward@arm.com>
16107 David Sherwood <david.sherwood@arm.com>
16109 * combine.c (force_int_to_mode): New function, split out from...
16110 (force_to_mode): ...here. Keep xmode up-to-date and use it
16111 instead of GET_MODE (x).
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 * optabs-query.h (extraction_insn::struct_mode): Change type to
16118 opt_scalar_int_mode and update comment.
16119 (extraction_insn::field_mode): Change type to scalar_int_mode.
16120 (extraction_insn::pos_mode): Likewise.
16121 * combine.c (make_extraction): Update accordingly.
16122 * optabs-query.c (get_traditional_extraction_insn): Likewise.
16123 (get_optab_extraction_insn): Likewise.
16124 * recog.c (simplify_while_replacing): Likewise.
16125 * expmed.c (narrow_bit_field_mem): Change the type of the mode
16126 parameter to opt_scalar_int_mode.
16128 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16129 Alan Hayward <alan.hayward@arm.com>
16130 David Sherwood <david.sherwood@arm.com>
16132 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
16133 to a scalar_int_mode instead of a machine_mode.
16134 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
16135 (get_best_mode): Return a boolean and use a pointer argument to store
16136 the selected mode. Replace the limit mode parameter with a bit limit.
16137 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
16138 for the values returned by bit_field_mode_iterator::next_mode.
16139 (store_bit_field): Update call to get_best_mode.
16140 (store_fixed_bit_field): Likewise.
16141 (extract_fixed_bit_field): Likewise.
16142 * expr.c (optimize_bitfield_assignment_op): Likewise.
16143 * fold-const.c (optimize_bit_field_compare): Likewise.
16144 (fold_truth_andor_1): Likewise.
16145 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
16146 Update for new type of m_mode.
16147 (get_best_mode): As above.
16149 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16150 Alan Hayward <alan.hayward@arm.com>
16151 David Sherwood <david.sherwood@arm.com>
16153 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
16154 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
16155 (store_bit_field): Check is_a <scalar_int_mode> before calling
16156 strict_volatile_bitfield_p.
16157 (extract_bit_field): Likewise.
16159 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16160 Alan Hayward <alan.hayward@arm.com>
16161 David Sherwood <david.sherwood@arm.com>
16163 * target.def (cstore_mode): Return a scalar_int_mode.
16164 * doc/tm.texi: Regenerate.
16165 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
16166 * targhooks.h (default_cstore_mode): Likewise.
16167 * targhooks.c (default_cstore_mode): Likewise, using a forced
16169 * expmed.c (emit_cstore): Expect the target of the cstore to be
16172 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16173 Alan Hayward <alan.hayward@arm.com>
16174 David Sherwood <david.sherwood@arm.com>
16176 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
16178 (niter_desc): Likewise mode.
16179 (iv_analyze): Add a mode parameter.
16181 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
16182 and change its type to scalar_int_mode.
16183 * loop-iv.c: Update commentary at head of file.
16184 (iv_constant): Pass the mode paraeter before the rtx it describes
16185 and change its type to scalar_int_mode. Remove VOIDmode handling.
16186 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
16187 (iv_extend): Likewise.
16188 (shorten_into_mode): Likewise.
16189 (iv_add): Use scalar_int_mode.
16190 (iv_mult): Likewise.
16191 (iv_shift): Likewise.
16192 (canonicalize_iv_subregs): Likewise.
16193 (get_biv_step_1): Pass the outer_mode parameter before the rtx
16194 it describes and change its mode to scalar_int_mode. Also change
16195 the type of the returned inner_mode to scalar_int_mode.
16196 (get_biv_step): Likewise, turning outer_mode from a pointer
16197 into a direct parameter. Update call to get_biv_step_1.
16198 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
16199 iv_constant and get_biv_step.
16200 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
16201 and change its type to scalar_int_mode. Don't initialise iv->mode
16202 to VOIDmode and remove later checks for its still being VOIDmode.
16203 Update calls to iv_analyze_op and iv_analyze_expr. Check
16204 is_a <scalar_int_mode> when changing the mode under consideration.
16205 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
16206 Update call to iv_analyze_expr.
16207 (iv_analyze_op): Add a mode parameter. Reject subregs whose
16208 inner register is not also a scalar_int_mode. Update call to
16210 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
16211 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
16212 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
16213 separate mode class checks. Update calls to iv_analyze. Remove
16214 fix-up of VOIDmodes after iv_analyze_biv.
16215 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
16216 don't have a scalar_int_mode. Update call to biv_p.
16218 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16219 Alan Hayward <alan.hayward@arm.com>
16220 David Sherwood <david.sherwood@arm.com>
16222 * cfgexpand.c (convert_debug_memory_address): Use
16223 as_a <scalar_int_mode>.
16224 * combine.c (expand_compound_operation): Likewise.
16225 (make_extraction): Likewise.
16226 (change_zero_ext): Likewise.
16227 (simplify_comparison): Likewise.
16228 * cse.c (cse_insn): Likewise.
16229 * dwarf2out.c (minmax_loc_descriptor): Likewise.
16230 (mem_loc_descriptor): Likewise.
16231 (loc_descriptor): Likewise.
16232 * expmed.c (init_expmed_one_mode): Likewise.
16233 (synth_mult): Likewise.
16234 (emit_store_flag_1): Likewise.
16235 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
16236 of a comparison with size.
16237 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
16238 (reduce_to_bit_field_precision): Likewise.
16239 * function.c (expand_function_end): Likewise.
16240 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
16241 * loop-doloop.c (doloop_modify): Likewise.
16242 * optabs.c (expand_binop): Likewise.
16243 (expand_unop): Likewise.
16244 (expand_copysign_absneg): Likewise.
16245 (prepare_cmp_insn): Likewise.
16246 (maybe_legitimize_operand): Likewise.
16247 * recog.c (const_scalar_int_operand): Likewise.
16248 * rtlanal.c (get_address_mode): Likewise.
16249 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16250 (simplify_cond_clz_ctz): Likewise.
16251 * tree-nested.c (get_nl_goto_field): Likewise.
16252 * tree.c (build_vector_type_for_mode): Likewise.
16253 * var-tracking.c (use_narrower_mode): Likewise.
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 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
16260 * builtins.c (expand_builtin_signbit): Use it.
16261 * cfgexpand.c (expand_debug_expr): Likewise.
16262 * dojump.c (do_jump): Likewise.
16263 (do_compare_and_jump): Likewise.
16264 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
16265 * expmed.c (make_tree): Likewise.
16266 * expr.c (expand_expr_real_2): Likewise.
16267 (expand_expr_real_1): Likewise.
16268 (try_casesi): Likewise.
16269 * fold-const-call.c (fold_const_call_ss): Likewise.
16270 * fold-const.c (unextend): Likewise.
16271 (extract_muldiv_1): Likewise.
16272 (fold_single_bit_test): Likewise.
16273 (native_encode_int): Likewise.
16274 (native_encode_string): Likewise.
16275 (native_interpret_int): Likewise.
16276 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
16277 * internal-fn.c (expand_addsub_overflow): Likewise.
16278 (expand_neg_overflow): Likewise.
16279 (expand_mul_overflow): Likewise.
16280 (expand_arith_overflow): Likewise.
16281 * match.pd: Likewise.
16282 * stor-layout.c (layout_type): Likewise.
16283 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
16284 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
16285 * tree-ssanames.c (get_range_info): Likewise.
16286 * tree-switch-conversion.c (array_value_type) Likewise.
16287 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
16288 (vect_recog_divmod_pattern): Likewise.
16289 (vect_recog_mixed_size_cond_pattern): Likewise.
16290 * tree-vrp.c (extract_range_basic): Likewise.
16291 (simplify_float_conversion_using_ranges): Likewise.
16292 * tree.c (int_fits_type_p): Likewise.
16293 * ubsan.c (instrument_bool_enum_load): Likewise.
16294 * varasm.c (mergeable_string_section): Likewise.
16295 (narrowing_initializer_constant_valid_p): Likewise.
16296 (output_constant): Likewise.
16298 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16299 Alan Hayward <alan.hayward@arm.com>
16300 David Sherwood <david.sherwood@arm.com>
16302 * machmode.h (NARROWEST_INT_MODE): New macro.
16303 * expr.c (alignment_for_piecewise_move): Use it instead of
16304 GET_CLASS_NARROWEST_MODE (MODE_INT).
16305 (push_block): Likewise.
16306 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
16308 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16310 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16311 Alan Hayward <alan.hayward@arm.com>
16312 David Sherwood <david.sherwood@arm.com>
16314 * postreload.c (move2add_valid_value_p): Change the type of the
16315 mode parameter to scalar_int_mode.
16316 (move2add_use_add2_insn): Add a mode parameter and use it instead
16318 (move2add_use_add3_insn): Likewise.
16319 (reload_cse_move2add): Update accordingly.
16321 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16322 Alan Hayward <alan.hayward@arm.com>
16323 David Sherwood <david.sherwood@arm.com>
16325 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
16327 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
16328 * optabs.c (expand_unop): Likewise.
16330 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16331 Alan Hayward <alan.hayward@arm.com>
16332 David Sherwood <david.sherwood@arm.com>
16334 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
16335 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
16336 (popcount_loc_descriptor): Likewise.
16337 (bswap_loc_descriptor): Likewise.
16338 (rotate_loc_descriptor): Likewise.
16339 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
16340 calling the functions above.
16342 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16343 Alan Hayward <alan.hayward@arm.com>
16344 David Sherwood <david.sherwood@arm.com>
16346 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
16348 (try_combine): Likewise.
16349 (simplify_if_then_else): Likewise.
16350 * cse.c (cse_insn): Likewise.
16351 * dwarf2out.c (mem_loc_descriptor): Likewise.
16352 * emit-rtl.c (gen_lowpart_common): Likewise.
16353 * simplify-rtx.c (simplify_truncation): Likewise.
16354 (simplify_binary_operation_1): Likewise.
16355 (simplify_const_relational_operation): Likewise.
16356 (simplify_ternary_operation): Likewise.
16357 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
16359 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16360 Alan Hayward <alan.hayward@arm.com>
16361 David Sherwood <david.sherwood@arm.com>
16363 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
16364 * reload.c (push_reload): Likewise.
16365 (find_reloads): 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 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
16372 (make_compound_operation_int): Likewise.
16373 (change_zero_ext): Likewise.
16374 * expr.c (convert_move): Likewise.
16375 (convert_modes): Likewise.
16376 * fwprop.c (forward_propagate_subreg): Likewise.
16377 * loop-iv.c (get_biv_step_1): Likewise.
16378 * optabs.c (widen_operand): Likewise.
16379 * postreload.c (move2add_valid_value_p): Likewise.
16380 * recog.c (simplify_while_replacing): Likewise.
16381 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16382 (simplify_binary_operation_1): Likewise. Remove redundant
16383 mode equality check.
16385 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16386 Alan Hayward <alan.hayward@arm.com>
16387 David Sherwood <david.sherwood@arm.com>
16389 * combine.c (combine_simplify_rtx): Add checks for
16390 is_a <scalar_int_mode>.
16391 (simplify_if_then_else): Likewise.
16392 (make_field_assignment): Likewise.
16393 (simplify_comparison): Likewise.
16394 * ifcvt.c (noce_try_bitop): Likewise.
16395 * loop-invariant.c (canonicalize_address_mult): Likewise.
16396 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16398 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16399 Alan Hayward <alan.hayward@arm.com>
16400 David Sherwood <david.sherwood@arm.com>
16402 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
16403 is_a <scalar_int_mode> instead of != BLKmode.
16405 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16406 Alan Hayward <alan.hayward@arm.com>
16407 David Sherwood <david.sherwood@arm.com>
16409 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
16410 instead of != VOIDmode.
16411 * combine.c (if_then_else_cond): Likewise.
16412 (change_zero_ext): Likewise.
16413 * dwarf2out.c (mem_loc_descriptor): Likewise.
16414 (loc_descriptor): Likewise.
16415 * rtlanal.c (canonicalize_condition): Likewise.
16416 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
16418 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16419 Alan Hayward <alan.hayward@arm.com>
16420 David Sherwood <david.sherwood@arm.com>
16422 * simplify-rtx.c (simplify_binary_operation_1): Use
16423 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
16425 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16426 Alan Hayward <alan.hayward@arm.com>
16427 David Sherwood <david.sherwood@arm.com>
16429 * wide-int.h (int_traits<unsigned char>) New class.
16430 (int_traits<unsigned short>) Likewise.
16431 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
16432 Use GET_MODE_UNIT_PRECISION and remove redundant test for
16434 * combine.c (set_nonzero_bits_and_sign_copies): Use
16435 is_a <scalar_int_mode>.
16436 (find_split_point): Likewise.
16437 (combine_simplify_rtx): Likewise.
16438 (simplify_logical): Likewise.
16439 (expand_compound_operation): Likewise.
16440 (expand_field_assignment): Likewise.
16441 (make_compound_operation): Likewise.
16442 (extended_count): Likewise.
16443 (change_zero_ext): Likewise.
16444 (simplify_comparison): Likewise.
16445 * dwarf2out.c (scompare_loc_descriptor): Likewise.
16446 (ucompare_loc_descriptor): Likewise.
16447 (minmax_loc_descriptor): Likewise.
16448 (mem_loc_descriptor): Likewise.
16449 (loc_descriptor): Likewise.
16450 * expmed.c (init_expmed_one_mode): Likewise.
16451 * lra-constraints.c (lra_constraint_offset): Likewise.
16452 * optabs.c (prepare_libcall_arg): Likewise.
16453 * postreload.c (move2add_note_store): Likewise.
16454 * reload.c (operands_match_p): Likewise.
16455 * rtl.h (load_extend_op): Likewise.
16456 * rtlhooks.c (gen_lowpart_general): Likewise.
16457 * simplify-rtx.c (simplify_truncation): Likewise.
16458 (simplify_unary_operation_1): Likewise.
16459 (simplify_binary_operation_1): Likewise.
16460 (simplify_const_binary_operation): Likewise.
16461 (simplify_const_relational_operation): Likewise.
16462 (simplify_subreg): Likewise.
16463 * stor-layout.c (bitwise_mode_for_mode): Likewise.
16464 * var-tracking.c (adjust_mems): Likewise.
16465 (prepare_call_arguments): Likewise.
16467 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16468 Alan Hayward <alan.hayward@arm.com>
16469 David Sherwood <david.sherwood@arm.com>
16471 * machmode.h (is_int_mode): New fuction.
16472 * combine.c (find_split_point): Use it.
16473 (combine_simplify_rtx): Likewise.
16474 (simplify_if_then_else): Likewise.
16475 (simplify_set): Likewise.
16476 (simplify_shift_const_1): Likewise.
16477 (simplify_comparison): Likewise.
16478 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
16479 * cse.c (notreg_cost): Likewise.
16480 (cse_insn): Likewise.
16481 * cselib.c (cselib_lookup_1): Likewise.
16482 * dojump.c (do_jump_1): Likewise.
16483 (do_compare_rtx_and_jump): Likewise.
16484 * dse.c (get_call_args): Likewise.
16485 * dwarf2out.c (rtl_for_decl_init): Likewise.
16486 (native_encode_initializer): Likewise.
16487 * expmed.c (emit_store_flag_1): Likewise.
16488 (emit_store_flag): Likewise.
16489 * expr.c (convert_modes): Likewise.
16490 (store_field): Likewise.
16491 (expand_expr_real_1): Likewise.
16492 * fold-const.c (fold_read_from_constant_string): Likewise.
16493 * gimple-ssa-sprintf.c (get_format_string): Likewise.
16494 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
16495 * optabs.c (expand_binop): Likewise.
16496 (expand_unop): Likewise.
16497 (expand_abs_nojump): Likewise.
16498 (expand_one_cmpl_abs_nojump): Likewise.
16499 * simplify-rtx.c (mode_signbit_p): Likewise.
16500 (val_signbit_p): Likewise.
16501 (val_signbit_known_set_p): Likewise.
16502 (val_signbit_known_clear_p): Likewise.
16503 (simplify_relational_operation_1): Likewise.
16504 * tree.c (vector_type_mode): Likewise.
16506 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16507 Alan Hayward <alan.hayward@arm.com>
16508 David Sherwood <david.sherwood@arm.com>
16510 * machmode.h (smallest_mode_for_size): Fix formatting.
16511 (smallest_int_mode_for_size): New function.
16512 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
16513 instead of smallest_mode_for_size.
16514 * combine.c (make_extraction): Likewise.
16515 * config/arc/arc.c (arc_expand_movmem): Likewise.
16516 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
16517 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
16518 * config/s390/s390.c (s390_expand_insv): Likewise.
16519 * config/sparc/sparc.c (assign_int_registers): Likewise.
16520 * config/spu/spu.c (spu_function_value): Likewise.
16521 (spu_function_arg): Likewise.
16522 * coverage.c (get_gcov_type): Likewise.
16523 (get_gcov_unsigned_t): Likewise.
16524 * dse.c (find_shift_sequence): Likewise.
16525 * expmed.c (store_bit_field_1): Likewise.
16526 * expr.c (convert_move): Likewise.
16527 (store_field): Likewise.
16528 * internal-fn.c (expand_arith_overflow): Likewise.
16529 * optabs-query.c (get_best_extraction_insn): Likewise.
16530 * optabs.c (expand_twoval_binop_libfunc): Likewise.
16531 * stor-layout.c (layout_type): Likewise.
16532 (initialize_sizetypes): Likewise.
16533 * targhooks.c (default_get_mask_mode): Likewise.
16534 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
16536 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16537 Alan Hayward <alan.hayward@arm.com>
16538 David Sherwood <david.sherwood@arm.com>
16540 * machmode.h (opt_mode::else_blk): New function.
16541 (int_mode_for_mode): Declare.
16542 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
16543 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
16545 * cfgexpand.c (expand_debug_expr): Likewise.
16546 * combine.c (gen_lowpart_or_truncate): Likewise.
16547 (gen_lowpart_for_combine): Likewise.
16548 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
16549 * config/avr/avr.c (avr_to_int_mode): Likewise.
16550 (avr_out_plus_1): Likewise.
16551 (avr_out_plus): Likewise.
16552 (avr_out_round): Likewise.
16553 * config/i386/i386.c (ix86_split_to_parts): Likewise.
16554 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
16555 (s390_expand_vcond): Likewise.
16556 * config/spu/spu.c (spu_split_immediate): Likewise.
16557 (spu_expand_mov): Likewise.
16558 * dse.c (get_stored_val): Likewise.
16559 * expmed.c (store_bit_field_1): Likewise.
16560 (convert_extracted_bit_field): Use int_mode_for_mode instead of
16562 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
16563 (extract_low_bits): Likewise.
16564 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
16565 handling rather than repeating the check.
16566 (emit_group_store): Likewise.
16567 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
16568 * optabs.c (expand_absneg_bit): Likewise.
16569 (expand_copysign_absneg): Likewise.
16570 (expand_copysign_bit): Likewise.
16571 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
16572 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
16573 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
16574 * var-tracking.c (prepare_call_arguments): Likewise.
16575 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
16576 int_mode_for_mode instead of mode_for_size.
16577 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
16579 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16580 Alan Hayward <alan.hayward@arm.com>
16581 David Sherwood <david.sherwood@arm.com>
16583 * machmode.h (int_mode_for_size): New function.
16584 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
16585 instead of mode_for_size.
16586 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
16588 * combine.c (expand_field_assignment): Use int_mode_for_size
16589 instead of mode_for_size.
16590 (make_extraction): Likewise.
16591 (simplify_shift_const_1): Likewise.
16592 (simplify_comparison): Likewise.
16593 * dojump.c (do_jump): Likewise.
16594 * dwarf2out.c (mem_loc_descriptor): Likewise.
16595 * emit-rtl.c (init_derived_machine_modes): Likewise.
16596 * expmed.c (flip_storage_order): Likewise.
16597 (convert_extracted_bit_field): Likewise.
16598 * expr.c (copy_blkmode_from_reg): Likewise.
16599 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
16600 * internal-fn.c (expand_mul_overflow): Likewise.
16601 * lower-subreg.c (simple_move): Likewise.
16602 * optabs-libfuncs.c (init_optabs): Likewise.
16603 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16604 * tree.c (vector_type_mode): Likewise.
16605 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
16606 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
16607 * tree-vect-generic.c (expand_vector_parallel): Likewise.
16608 * tree-vect-stmts.c (vectorizable_load): Likewise.
16609 (vectorizable_store): Likewise.
16611 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16612 Alan Hayward <alan.hayward@arm.com>
16613 David Sherwood <david.sherwood@arm.com>
16615 * coretypes.h (pod_mode): New type.
16616 (scalar_int_mode_pod): New typedef.
16617 * machmode.h (pod_mode): New class.
16618 (int_n_data_t::m): Change type to scalar_int_mode_pod.
16619 * genmodes.c (emit_mode_int_n): Update accordingly.
16620 * lower-subreg.h (target_lower_subreg): Change type to
16621 scalar_int_mode_pod.
16622 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
16623 scalar_int_mode_pod.
16625 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16626 Alan Hayward <alan.hayward@arm.com>
16627 David Sherwood <david.sherwood@arm.com>
16629 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
16630 machine_mode to scalar_int_mode.
16631 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
16632 (rs6000_option_override_internal): Remove cast to int.
16633 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
16634 machine_mode to scalar_int_mode.
16635 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
16636 (rs6000_option_override_internal): Remove cast to int.
16637 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
16638 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
16640 * config/s390/s390.c (s390_expand_builtin): Likewise.
16641 * coretypes.h (scalar_int_mode): New type.
16642 (opt_scalar_int_mode): New typedef.
16643 * machmode.h (scalar_int_mode): New class.
16644 (scalar_int_mode::includes_p): New function.
16645 (byte_mode): Change type to scalar_int_mode.
16646 (word_mode): Likewise.
16647 (ptr_mode): Likewise.
16648 * emit-rtl.c (byte_mode): Likewise.
16649 (word_mode): Likewise.
16650 (ptr_mode): Likewise.
16651 (init_derived_machine_modes): Update accordingly.
16652 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
16653 and MODE_PARTIAL_INT.
16654 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
16655 opt_scalar_int_mode.
16657 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16658 Alan Hayward <alan.hayward@arm.com>
16659 David Sherwood <david.sherwood@arm.com>
16661 * target.def (libgcc_floating_mode_supported_p): Take a
16663 * doc/tm.texi: Regenerate.
16664 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
16666 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16667 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
16670 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16671 Alan Hayward <alan.hayward@arm.com>
16672 David Sherwood <david.sherwood@arm.com>
16674 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
16675 * doc/tm.texi: Regenerate.
16676 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
16677 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
16678 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
16679 * targhooks.h (default_floatn_mode): Likewise.
16680 * targhooks.c (default_floatn_mode): Likewise.
16681 * tree.c (build_common_tree_nodes): Update accordingly.
16683 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16684 Alan Hayward <alan.hayward@arm.com>
16685 David Sherwood <david.sherwood@arm.com>
16687 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
16688 (mode_iterator::iterate_p): Likewise.
16689 (mode_iterator::get_wider): Likewise.
16690 * expr.c (init_expr_target): Use opt_scalar_float_mode.
16692 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16693 Alan Hayward <alan.hayward@arm.com>
16694 David Sherwood <david.sherwood@arm.com>
16696 * coretypes.h (opt_scalar_float_mode): New typedef.
16697 * machmode.h (float_mode_for_size): New function.
16698 * emit-rtl.c (double_mode): Delete.
16699 (init_emit_once): Use float_mode_for_size.
16700 * stor-layout.c (layout_type): Likewise.
16701 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
16703 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16704 Alan Hayward <alan.hayward@arm.com>
16705 David Sherwood <david.sherwood@arm.com>
16707 * output.h (assemble_real): Take a scalar_float_mode.
16708 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
16709 * config/arm/arm.md (consttable_4): Likewise.
16710 (consttable_8): Likewise.
16711 (consttable_16): Likewise.
16712 * config/mips/mips.md (consttable_float): Likewise.
16713 * config/s390/s390.c (s390_output_pool_entry): Likewise.
16714 * varasm.c (assemble_real): Take a scalar_float_mode.
16715 (output_constant_pool_2): Update accordingly.
16716 (output_constant): Likewise.
16718 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16719 Alan Hayward <alan.hayward@arm.com>
16720 David Sherwood <david.sherwood@arm.com>
16722 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
16723 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
16724 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
16725 (native_encode_real): Likewise.
16726 (native_interpret_real): Likewise.
16727 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
16728 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16730 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16731 Alan Hayward <alan.hayward@arm.com>
16732 David Sherwood <david.sherwood@arm.com>
16734 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
16735 <scalar_float_mode>. Simplify.
16736 (gen_extend_conv_libfunc): Likewise.
16738 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16739 Alan Hayward <alan.hayward@arm.com>
16740 David Sherwood <david.sherwood@arm.com>
16742 * coretypes.h (scalar_float_mode): New type.
16743 * machmode.h (mode_traits::from_int): Use machine_mode if
16744 USE_ENUM_MODES is defined.
16745 (is_a): New function.
16747 (dyn_cast): Likewise.
16748 (scalar_float_mode): New class.
16749 (scalar_float_mode::includes_p): New function.
16750 (is_float_mode): Likewise.
16751 * gdbhooks.py (MachineModePrinter): New class.
16752 (build_pretty_printer): Use it for scalar_float_mode.
16753 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
16754 (format_helper::format_helper): Turn into a template.
16755 * genmodes.c (get_mode_class): New function.
16756 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
16757 or machine_mode if none.
16758 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
16759 as_a <scalar_float_mode>.
16760 * dwarf2out.c (mem_loc_descriptor): Likewise.
16761 (insert_float): Likewise.
16762 (add_const_value_attribute): Likewise.
16763 * simplify-rtx.c (simplify_immed_subreg): Likewise.
16764 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
16765 (expand_unop): Update accordingly.
16766 (expand_abs_nojump): Likewise.
16767 (expand_copysign_absneg): Take a scalar_float_mode.
16768 (expand_copysign_bit): Likewise.
16769 (expand_copysign): Update accordingly.
16771 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16772 Alan Hayward <alan.hayward@arm.com>
16773 David Sherwood <david.sherwood@arm.com>
16775 * coretypes.h (opt_mode): New class.
16776 * machmode.h (opt_mode): Likewise.
16777 (opt_mode::else_void): New function.
16778 (opt_mode::require): Likewise.
16779 (opt_mode::exists): Likewise.
16780 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
16781 (GET_MODE_2XWIDER_MODE): Likewise.
16782 (mode_iterator::get_wider): Update accordingly.
16783 (mode_iterator::get_2xwider): Likewise.
16784 (mode_iterator::get_known_wider): Likewise, turning into a template.
16785 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
16786 forcing a wider mode to exist.
16787 * config/cr16/cr16.h (LONG_REG_P): Likewise.
16788 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16789 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
16790 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
16791 * lower-subreg.c (init_lower_subreg): Likewise.
16792 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
16793 on the final iteration.
16794 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
16795 a wider mode exists before asking for a move pattern.
16796 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
16797 forcing a wider mode to exist.
16798 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
16799 returning false if no such mode exists.
16800 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
16801 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
16802 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
16803 Avoid checking for a MODE_INT if we already know the mode is not a
16805 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
16806 forcing a wider mode to exist.
16807 (expmed_mult_highpart_optab): Likewise.
16808 (expmed_mult_highpart): Likewise.
16809 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
16811 * lto-streamer-in.c (lto_input_mode_table): Likewise.
16812 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
16813 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16814 * internal-fn.c (expand_mul_overflow): Update use of
16815 GET_MODE_2XWIDER_MODE.
16816 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16817 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
16818 GET_MODE_WIDER_MODE.
16819 (convert_plusminus_to_widen): Likewise.
16820 * tree-switch-conversion.c (array_value_type): Likewise.
16821 * var-tracking.c (emit_note_insn_var_location): Likewise.
16822 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16823 Return false inside rather than outside the loop if no wider mode
16825 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
16826 and GET_MODE_2XWIDER_MODE
16827 (can_compare_p): Use else_void.
16828 * gdbhooks.py (OptMachineModePrinter): New class.
16829 (build_pretty_printer): Use it for opt_mode.
16831 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16832 Alan Hayward <alan.hayward@arm.com>
16833 David Sherwood <david.sherwood@arm.com>
16835 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
16836 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
16838 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16839 Alan Hayward <alan.hayward@arm.com>
16840 David Sherwood <david.sherwood@arm.com>
16842 * machmode.h (mode_traits): New structure.
16843 (get_narrowest_mode): New function.
16844 (mode_iterator::start): Likewise.
16845 (mode_iterator::iterate_p): Likewise.
16846 (mode_iterator::get_wider): Likewise.
16847 (mode_iterator::get_known_wider): Likewise.
16848 (mode_iterator::get_2xwider): Likewise.
16849 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
16850 (FOR_EACH_MODE): Likewise.
16851 (FOR_EACH_MODE_FROM): Likewise.
16852 (FOR_EACH_MODE_UNTIL): Likewise.
16853 (FOR_EACH_WIDER_MODE): Likewise.
16854 (FOR_EACH_2XWIDER_MODE): Likewise.
16855 * builtins.c (expand_builtin_strlen): Use new mode iterators.
16856 * combine.c (simplify_comparison): Likewise
16857 * config/i386/i386.c (type_natural_mode): Likewise.
16858 * cse.c (cse_insn): Likewise.
16859 * dse.c (find_shift_sequence): Likewise.
16860 * emit-rtl.c (init_derived_machine_modes): Likewise.
16861 (init_emit_once): Likewise.
16862 * explow.c (hard_function_value): Likewise.
16863 * expmed.c (extract_fixed_bit_field_1): Likewise.
16864 (extract_bit_field_1): Likewise.
16865 (expand_divmod): Likewise.
16866 (emit_store_flag_1): Likewise.
16867 * expr.c (init_expr_target): Likewise.
16868 (convert_move): Likewise.
16869 (alignment_for_piecewise_move): Likewise.
16870 (widest_int_mode_for_size): Likewise.
16871 (emit_block_move_via_movmem): Likewise.
16872 (copy_blkmode_to_reg): Likewise.
16873 (set_storage_via_setmem): Likewise.
16874 (compress_float_constant): Likewise.
16875 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16876 * optabs-query.c (get_best_extraction_insn): Likewise.
16877 * optabs.c (expand_binop): Likewise.
16878 (expand_twoval_unop): Likewise.
16879 (expand_twoval_binop): Likewise.
16880 (widen_leading): Likewise.
16881 (widen_bswap): Likewise.
16882 (expand_parity): Likewise.
16883 (expand_unop): Likewise.
16884 (prepare_cmp_insn): Likewise.
16885 (prepare_float_lib_cmp): Likewise.
16886 (expand_float): Likewise.
16887 (expand_fix): Likewise.
16888 (expand_sfix_optab): Likewise.
16889 * postreload.c (move2add_use_add2_insn): Likewise.
16890 * reg-stack.c (reg_to_stack): Likewise.
16891 * reginfo.c (choose_hard_reg_mode): Likewise.
16892 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16893 * stor-layout.c (mode_for_size): Likewise.
16894 (smallest_mode_for_size): Likewise.
16895 (mode_for_vector): Likewise.
16896 (finish_bitfield_representative): Likewise.
16897 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
16898 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
16899 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16900 * var-tracking.c (prepare_call_arguments): Likewise.
16902 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16903 Alan Hayward <alan.hayward@arm.com>
16904 David Sherwood <david.sherwood@arm.com>
16906 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
16907 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
16908 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
16909 * machmode.h (mode_size): Move earlier in file.
16910 (mode_precision): Likewise.
16911 (mode_inner): Likewise.
16912 (mode_nunits): Likewise.
16913 (mode_unit_size): Likewise.
16914 (unit_unit_precision): Likewise.
16915 (mode_wider): Likewise.
16916 (mode_2xwider): Likewise.
16917 (machine_mode): New class.
16918 (mode_to_bytes): New function.
16919 (mode_to_bits): Likewise.
16920 (mode_to_precision): Likewise.
16921 (mode_to_inner): Likewise.
16922 (mode_to_unit_size): Likewise.
16923 (mode_to_unit_precision): Likewise.
16924 (mode_to_nunits): Likewise.
16925 (GET_MODE_SIZE): Use mode_to_bytes.
16926 (GET_MODE_BITSIZE): Use mode_to_bits.
16927 (GET_MODE_PRECISION): Use mode_to_precision.
16928 (GET_MODE_INNER): Use mode_to_inner.
16929 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
16930 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
16931 (GET_MODE_NUNITS): Use mode_to_nunits.
16932 * system.h (ALWAYS_INLINE): New macro.
16933 * config/powerpcspe/powerpcspe-c.c
16934 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
16935 int for arg1_mode and arg2_mode.
16937 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16938 Alan Hayward <alan.hayward@arm.com>
16939 David Sherwood <david.sherwood@arm.com>
16941 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16942 Prefix mode names with E_ in case statements.
16943 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16944 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
16945 (aarch64_split_simd_move): Likewise.
16946 (aarch64_gen_storewb_pair): Likewise.
16947 (aarch64_gen_loadwb_pair): Likewise.
16948 (aarch64_gen_store_pair): Likewise.
16949 (aarch64_gen_load_pair): Likewise.
16950 (aarch64_get_condition_code_1): Likewise.
16951 (aarch64_constant_pool_reload_icode): Likewise.
16952 (get_rsqrte_type): Likewise.
16953 (get_rsqrts_type): Likewise.
16954 (get_recpe_type): Likewise.
16955 (get_recps_type): Likewise.
16956 (aarch64_gimplify_va_arg_expr): Likewise.
16957 (aarch64_simd_container_mode): Likewise.
16958 (aarch64_emit_load_exclusive): Likewise.
16959 (aarch64_emit_store_exclusive): Likewise.
16960 (aarch64_expand_compare_and_swap): Likewise.
16961 (aarch64_gen_atomic_cas): Likewise.
16962 (aarch64_emit_bic): Likewise.
16963 (aarch64_emit_atomic_swap): Likewise.
16964 (aarch64_emit_atomic_load_op): Likewise.
16965 (aarch64_evpc_trn): Likewise.
16966 (aarch64_evpc_uzp): Likewise.
16967 (aarch64_evpc_zip): Likewise.
16968 (aarch64_evpc_ext): Likewise.
16969 (aarch64_evpc_rev): Likewise.
16970 (aarch64_evpc_dup): Likewise.
16971 (aarch64_gen_ccmp_first): Likewise.
16972 (aarch64_gen_ccmp_next): Likewise.
16973 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
16974 (alpha_emit_xfloating_libcall): Likewise.
16975 (emit_insxl): Likewise.
16976 (alpha_arg_type): Likewise.
16977 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
16978 (arc_preferred_simd_mode): Likewise.
16979 (arc_secondary_reload): Likewise.
16980 (get_arc_condition_code): Likewise.
16981 (arc_print_operand): Likewise.
16982 (arc_legitimate_constant_p): Likewise.
16983 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16984 * config/arc/arc.md (casesi_load): Likewise.
16985 (casesi_compact_jump): Likewise.
16986 * config/arc/predicates.md (proper_comparison_operator): Likewise.
16987 (cc_use_register): Likewise.
16988 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16989 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
16990 (arm_init_iwmmxt_builtins): Likewise.
16991 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
16992 (neon_expand_vector_init): Likewise.
16993 (arm_attr_length_move_neon): Likewise.
16994 (maybe_get_arm_condition_code): Likewise.
16995 (arm_emit_vector_const): Likewise.
16996 (arm_preferred_simd_mode): Likewise.
16997 (arm_output_iwmmxt_tinsr): Likewise.
16998 (thumb1_output_casesi): Likewise.
16999 (thumb2_output_casesi): Likewise.
17000 (arm_emit_load_exclusive): Likewise.
17001 (arm_emit_store_exclusive): Likewise.
17002 (arm_expand_compare_and_swap): Likewise.
17003 (arm_evpc_neon_vuzp): Likewise.
17004 (arm_evpc_neon_vzip): Likewise.
17005 (arm_evpc_neon_vrev): Likewise.
17006 (arm_evpc_neon_vtrn): Likewise.
17007 (arm_evpc_neon_vext): Likewise.
17008 (arm_validize_comparison): Likewise.
17009 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
17010 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
17011 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
17012 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
17013 (c6x_preferred_simd_mode): Likewise.
17014 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
17015 (epiphany_rtx_costs): Likewise.
17016 * config/epiphany/predicates.md (proper_comparison_operator):
17018 * config/frv/frv.c (condexec_memory_operand): Likewise.
17019 (frv_emit_move): Likewise.
17020 (output_move_single): Likewise.
17021 (output_condmove_single): Likewise.
17022 (frv_hard_regno_mode_ok): Likewise.
17023 (frv_matching_accg_mode): Likewise.
17024 * config/h8300/h8300.c (split_adds_subs): Likewise.
17025 (h8300_rtx_costs): Likewise.
17026 (h8300_print_operand): Likewise.
17027 (compute_mov_length): Likewise.
17028 (output_logical_op): Likewise.
17029 (compute_logical_op_length): Likewise.
17030 (compute_logical_op_cc): Likewise.
17031 (h8300_shift_needs_scratch_p): Likewise.
17032 (output_a_shift): Likewise.
17033 (compute_a_shift_length): Likewise.
17034 (compute_a_shift_cc): Likewise.
17035 (expand_a_rotate): Likewise.
17036 (output_a_rotate): Likewise.
17037 * config/i386/i386.c (classify_argument): Likewise.
17038 (function_arg_advance_32): Likewise.
17039 (function_arg_32): Likewise.
17040 (function_arg_64): Likewise.
17041 (function_value_64): Likewise.
17042 (ix86_gimplify_va_arg): Likewise.
17043 (ix86_legitimate_constant_p): Likewise.
17044 (put_condition_code): Likewise.
17045 (split_double_mode): Likewise.
17046 (ix86_avx256_split_vector_move_misalign): Likewise.
17047 (ix86_expand_vector_logical_operator): Likewise.
17048 (ix86_split_idivmod): Likewise.
17049 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
17050 (ix86_build_const_vector): Likewise.
17051 (ix86_build_signbit_mask): Likewise.
17052 (ix86_match_ccmode): Likewise.
17053 (ix86_cc_modes_compatible): Likewise.
17054 (ix86_expand_branch): Likewise.
17055 (ix86_expand_sse_cmp): Likewise.
17056 (ix86_expand_sse_movcc): Likewise.
17057 (ix86_expand_int_sse_cmp): Likewise.
17058 (ix86_expand_vec_perm_vpermi2): Likewise.
17059 (ix86_expand_vec_perm): Likewise.
17060 (ix86_expand_sse_unpack): Likewise.
17061 (ix86_expand_int_addcc): Likewise.
17062 (ix86_split_to_parts): Likewise.
17063 (ix86_vectorize_builtin_gather): Likewise.
17064 (ix86_vectorize_builtin_scatter): Likewise.
17065 (avx_vpermilp_parallel): Likewise.
17066 (inline_memory_move_cost): Likewise.
17067 (ix86_tieable_integer_mode_p): Likewise.
17068 (x86_maybe_negate_const_int): Likewise.
17069 (ix86_expand_vector_init_duplicate): Likewise.
17070 (ix86_expand_vector_init_one_nonzero): Likewise.
17071 (ix86_expand_vector_init_one_var): Likewise.
17072 (ix86_expand_vector_init_concat): Likewise.
17073 (ix86_expand_vector_init_interleave): Likewise.
17074 (ix86_expand_vector_init_general): Likewise.
17075 (ix86_expand_vector_set): Likewise.
17076 (ix86_expand_vector_extract): Likewise.
17077 (emit_reduc_half): Likewise.
17078 (ix86_emit_i387_round): Likewise.
17079 (ix86_mangle_type): Likewise.
17080 (ix86_expand_round_sse4): Likewise.
17081 (expand_vec_perm_blend): Likewise.
17082 (canonicalize_vector_int_perm): Likewise.
17083 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
17084 (expand_vec_perm_1): Likewise.
17085 (expand_vec_perm_interleave3): Likewise.
17086 (expand_vec_perm_even_odd_pack): Likewise.
17087 (expand_vec_perm_even_odd_1): Likewise.
17088 (expand_vec_perm_broadcast_1): Likewise.
17089 (ix86_vectorize_vec_perm_const_ok): Likewise.
17090 (ix86_expand_vecop_qihi): Likewise.
17091 (ix86_expand_mul_widen_hilo): Likewise.
17092 (ix86_expand_sse2_abs): Likewise.
17093 (ix86_expand_pextr): Likewise.
17094 (ix86_expand_pinsr): Likewise.
17095 (ix86_preferred_simd_mode): Likewise.
17096 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
17097 * config/i386/sse.md (*andnot<mode>3): Likewise.
17098 (<mask_codefor><code><mode>3<mask_name>): Likewise.
17099 (*<code><mode>3): Likewise.
17100 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
17101 (ia64_expand_atomic_op): Likewise.
17102 (ia64_arg_type): Likewise.
17103 (ia64_mode_to_int): Likewise.
17104 (ia64_scalar_mode_supported_p): Likewise.
17105 (ia64_vector_mode_supported_p): Likewise.
17106 (expand_vec_perm_broadcast): Likewise.
17107 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
17108 (iq2000_function_arg_advance): Likewise.
17109 (iq2000_function_arg): Likewise.
17110 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
17111 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
17112 (m68k_libcall_value): Likewise.
17113 (m68k_function_value): Likewise.
17114 (sched_attr_op_type): Likewise.
17115 * config/mcore/mcore.c (mcore_output_move): Likewise.
17116 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
17118 (microblaze_function_arg): Likewise.
17119 * config/mips/mips.c (mips16_build_call_stub): Likewise.
17120 (mips_print_operand): Likewise.
17121 (mips_mode_ok_for_mov_fmt_p): Likewise.
17122 (mips_vector_mode_supported_p): Likewise.
17123 (mips_preferred_simd_mode): Likewise.
17124 (mips_expand_vpc_loongson_even_odd): Likewise.
17125 (mips_expand_vec_unpack): Likewise.
17126 (mips_expand_vi_broadcast): Likewise.
17127 (mips_expand_vector_init): Likewise.
17128 (mips_expand_vec_reduc): Likewise.
17129 (mips_expand_msa_cmp): Likewise.
17130 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
17131 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
17132 (cc_flags_for_mode): Likewise.
17133 * config/msp430/msp430.c (msp430_print_operand): Likewise.
17134 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
17135 (nds32_output_casesi_pc_relative): Likewise.
17136 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17137 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
17138 (nvptx_gen_unpack): Likewise.
17139 (nvptx_gen_pack): Likewise.
17140 (nvptx_gen_shuffle): Likewise.
17141 (nvptx_gen_wcast): Likewise.
17142 (nvptx_preferred_simd_mode): Likewise.
17143 * config/pa/pa.c (pa_secondary_reload): Likewise.
17144 * config/pa/predicates.md (base14_operand): Likewise.
17145 * config/powerpcspe/powerpcspe-c.c
17146 (altivec_resolve_overloaded_builtin): Likewise.
17147 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
17149 (rs6000_preferred_simd_mode): Likewise.
17150 (output_vec_const_move): Likewise.
17151 (rs6000_expand_vector_extract): Likewise.
17152 (rs6000_split_vec_extract_var): Likewise.
17153 (reg_offset_addressing_ok_p): Likewise.
17154 (rs6000_legitimate_offset_address_p): Likewise.
17155 (rs6000_legitimize_address): Likewise.
17156 (rs6000_emit_set_const): Likewise.
17157 (rs6000_const_vec): Likewise.
17158 (rs6000_emit_move): Likewise.
17159 (spe_build_register_parallel): Likewise.
17160 (rs6000_darwin64_record_arg_recurse): Likewise.
17161 (swap_selector_for_mode): Likewise.
17162 (spe_init_builtins): Likewise.
17163 (paired_init_builtins): Likewise.
17164 (altivec_init_builtins): Likewise.
17165 (do_load_for_compare): Likewise.
17166 (rs6000_generate_compare): Likewise.
17167 (rs6000_expand_float128_convert): Likewise.
17168 (emit_load_locked): Likewise.
17169 (emit_store_conditional): Likewise.
17170 (rs6000_output_function_epilogue): Likewise.
17171 (rs6000_handle_altivec_attribute): Likewise.
17172 (rs6000_function_value): Likewise.
17173 (emit_fusion_gpr_load): Likewise.
17174 (emit_fusion_p9_load): Likewise.
17175 (emit_fusion_p9_store): Likewise.
17176 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
17177 (fusion_gpr_mem_load): Likewise.
17178 (fusion_addis_mem_combo_load): Likewise.
17179 (fusion_addis_mem_combo_store): Likewise.
17180 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
17181 (fusion_gpr_mem_load): Likewise.
17182 (fusion_addis_mem_combo_load): Likewise.
17183 (fusion_addis_mem_combo_store): Likewise.
17184 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17186 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
17187 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
17188 (rs6000_preferred_simd_mode): Likewise.
17189 (output_vec_const_move): Likewise.
17190 (rs6000_expand_vector_extract): Likewise.
17191 (rs6000_split_vec_extract_var): Likewise.
17192 (reg_offset_addressing_ok_p): Likewise.
17193 (rs6000_legitimate_offset_address_p): Likewise.
17194 (rs6000_legitimize_address): Likewise.
17195 (rs6000_emit_set_const): Likewise.
17196 (rs6000_const_vec): Likewise.
17197 (rs6000_emit_move): Likewise.
17198 (rs6000_darwin64_record_arg_recurse): Likewise.
17199 (swap_selector_for_mode): Likewise.
17200 (paired_init_builtins): Likewise.
17201 (altivec_init_builtins): Likewise.
17202 (rs6000_expand_float128_convert): Likewise.
17203 (emit_load_locked): Likewise.
17204 (emit_store_conditional): Likewise.
17205 (rs6000_output_function_epilogue): Likewise.
17206 (rs6000_handle_altivec_attribute): Likewise.
17207 (rs6000_function_value): Likewise.
17208 (emit_fusion_gpr_load): Likewise.
17209 (emit_fusion_p9_load): Likewise.
17210 (emit_fusion_p9_store): Likewise.
17211 * config/rx/rx.c (rx_gen_move_template): Likewise.
17212 (flags_from_mode): Likewise.
17213 * config/s390/predicates.md (s390_alc_comparison): Likewise.
17214 (s390_slb_comparison): Likewise.
17215 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
17216 (s390_vector_mode_supported_p): Likewise.
17217 (s390_cc_modes_compatible): Likewise.
17218 (s390_match_ccmode_set): Likewise.
17219 (s390_canonicalize_comparison): Likewise.
17220 (s390_emit_compare_and_swap): Likewise.
17221 (s390_branch_condition_mask): Likewise.
17222 (s390_rtx_costs): Likewise.
17223 (s390_secondary_reload): Likewise.
17224 (__SECONDARY_RELOAD_CASE): Likewise.
17225 (s390_expand_cs): Likewise.
17226 (s390_preferred_simd_mode): Likewise.
17227 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
17228 * config/sh/sh.c (sh_print_operand): Likewise.
17229 (dump_table): Likewise.
17230 (sh_secondary_reload): Likewise.
17231 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17232 * config/sh/sh.md (casesi_worker_1): Likewise.
17233 (casesi_worker_2): Likewise.
17234 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
17235 (fcc_comparison_operator): Likewise.
17236 * config/sparc/sparc.c (sparc_expand_move): Likewise.
17237 (emit_soft_tfmode_cvt): Likewise.
17238 (sparc_preferred_simd_mode): Likewise.
17239 (output_cbranch): Likewise.
17240 (sparc_print_operand): Likewise.
17241 (sparc_expand_vec_perm_bmask): Likewise.
17242 (vector_init_bshuffle): Likewise.
17243 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17244 (spu_vector_mode_supported_p): Likewise.
17245 (spu_expand_insv): Likewise.
17246 (spu_emit_branch_or_set): Likewise.
17247 (spu_handle_vector_attribute): Likewise.
17248 (spu_builtin_splats): Likewise.
17249 (spu_builtin_extract): Likewise.
17250 (spu_builtin_promote): Likewise.
17251 (spu_expand_sign_extend): Likewise.
17252 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17253 (tilegx_simd_int): Likewise.
17254 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
17255 (tilepro_simd_int): Likewise.
17256 * config/v850/v850.c (const_double_split): Likewise.
17257 (v850_print_operand): Likewise.
17258 (ep_memory_offset): Likewise.
17259 * config/vax/vax.c (vax_rtx_costs): Likewise.
17260 (vax_output_int_move): Likewise.
17261 (vax_output_int_add): Likewise.
17262 (vax_output_int_subtract): Likewise.
17263 * config/visium/predicates.md (visium_branch_operator): Likewise.
17264 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
17265 (visium_print_operand_address): Likewise.
17266 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17267 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
17268 (xtensa_expand_conditional_branch): Likewise.
17269 (xtensa_copy_incoming_a7): Likewise.
17270 (xtensa_output_literal): Likewise.
17271 * dfp.c (decimal_real_maxval): Likewise.
17272 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
17274 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17275 Alan Hayward <alan.hayward@arm.com>
17276 David Sherwood <david.sherwood@arm.com>
17278 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
17279 (mode_nunits_inline): Likewise.
17280 (mode_inner_inline): Likewise.
17281 (mode_unit_size_inline): Likewise.
17282 (mode_unit_precision_inline): Likewise.
17283 (emit_insn_modes_h): Likewise. Also emit a #define of the
17285 (emit_mode_wider): Add an E_ prefix to mode names.
17286 (emit_mode_complex): Likewise.
17287 (emit_mode_inner): Likewise.
17288 (emit_mode_adjustments): Likewise.
17289 (emit_mode_int_n): Likewise.
17290 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
17291 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
17292 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
17293 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
17294 (CRC32_BUILTIN, ENTRY): Likewise.
17295 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
17296 (aarch64_pop_regs): Likewise.
17297 (aarch64_process_components): Likewise.
17298 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
17299 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
17300 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
17301 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
17302 * config/arm/arm.c (arm_init_libfuncs): Likewise.
17303 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
17305 * config/i386/i386-builtin.def (pcmpestr): Likewise.
17306 (pcmpistr): Likewise.
17307 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
17308 * config/mmix/mmix.c (mmix_output_condition): Likewise.
17309 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17311 * config/rl78/rl78.c (mduc_regs): Likewise.
17312 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17313 (htm_expand_builtin): Likewise.
17314 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
17315 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
17316 * config/xtensa/xtensa.c (print_operand): Likewise.
17317 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
17318 (NUM_MODE_VECTOR_INT): Likewise.
17319 * genoutput.c (null_operand): Likewise.
17320 (output_operand_data): Likewise.
17321 * genrecog.c (print_parameter_value): Likewise.
17322 * lra.c (debug_operand_data): Likewise.
17324 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17326 * dwarf2out.c (scompare_loc_descriptor_wide)
17327 (scompare_loc_descriptor_narrow): New functions, split out from...
17328 (scompare_loc_descriptor): ...here.
17329 * expmed.c (emit_store_flag_int): New function, split out from...
17330 (emit_store_flag): ...here.
17332 2017-08-30 Richard Biener <rguenther@suse.de>
17334 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
17335 (dwarf2out_early_finish): Move setting of AT_pubnames from
17336 early debug output to early finish.
17338 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17340 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
17341 and -mdata-region to the assembler.
17343 2017-08-30 Richard Biener <rguenther@suse.de>
17345 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
17347 (gen_subprogram_die): Add DW_AT_object_pointer only early.
17348 (dwarf2out_early_global_decl): Only generate a DIE for the
17349 abstract origin if it doesn't already exist or is a declaration DIE.
17350 (resolve_addr): Do not add the linkage name twice when
17351 generating a stub DIE for the DW_TAG_GNU_call_site target.
17353 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17355 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17356 Use machine_mode rather than int for arg1_mode.
17358 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
17361 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
17362 that the second argument of the built-in functions to unpack
17363 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
17364 switch statement instead a lot of if statements.
17365 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
17366 Allow 64-bit values to be in Altivec registers as well as
17367 traditional floating point registers.
17368 (pack<mode>, FMOVE128_VSX iterator): Likewise.
17370 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
17372 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
17373 MAX_REGS_PER_ADDRESS == 1.
17375 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
17377 * config/i386/i386.opt (flag_fentry): Do not init to -1.
17378 * config/i386/i386.c (ix86_option_override_internal): Simplify
17379 setting of opts->x_flag_entry.
17381 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17382 Jakub Jelinek <jakub@redhat.com>
17383 Richard Biener <rguenther@suse.de>
17385 PR tree-optimization/81503
17386 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
17387 folded constant fits in the target type; reorder tests for clarity.
17389 2017-08-29 Martin Liska <mliska@suse.cz>
17391 * passes.def: Include pass_lower_switch.
17392 * stmt.c (dump_case_nodes): Remove and move to
17393 tree-switch-conversion.
17394 (case_values_threshold): Likewise.
17395 (expand_switch_as_decision_tree_p): Likewise.
17396 (emit_case_decision_tree): Likewise.
17397 (expand_case): Likewise.
17398 (balance_case_nodes): Likewise.
17399 (node_has_low_bound): Likewise.
17400 (node_has_high_bound): Likewise.
17401 (node_is_bounded): Likewise.
17402 (emit_case_nodes): Likewise.
17403 (struct simple_case_node): New struct.
17404 (add_case_node): Remove.
17405 (emit_case_dispatch_table): Use vector instead of case_list.
17406 (reset_out_edges_aux): Remove.
17407 (compute_cases_per_edge): Likewise.
17408 (expand_case): Build list of simple_case_node.
17409 (expand_sjlj_dispatch_table): Use it.
17410 * tree-switch-conversion.c (struct case_node): Moved from
17411 stmt.c and adjusted.
17412 (emit_case_nodes): Likewise.
17413 (node_has_low_bound): Likewise.
17414 (node_has_high_bound): Likewise.
17415 (node_is_bounded): Likewise.
17416 (case_values_threshold): Likewise.
17417 (reset_out_edges_aux): Likewise.
17418 (compute_cases_per_edge): Likewise.
17419 (add_case_node): Likewise.
17420 (dump_case_nodes): Likewise.
17421 (balance_case_nodes): Likewise.
17422 (expand_switch_as_decision_tree_p): Likewise.
17423 (emit_jump): Likewise.
17424 (emit_case_decision_tree): Likewise.
17425 (try_switch_expansion): Likewise.
17426 (do_jump_if_equal): Likewise.
17427 (emit_cmp_and_jump_insns): Likewise.
17428 (fix_phi_operands_for_edge): New function.
17429 (record_phi_operand_mapping): Likewise.
17430 (class pass_lower_switch): New pass.
17431 (pass_lower_switch::execute): New function.
17432 (make_pass_lower_switch): Likewise.
17433 (conditional_probability):
17434 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
17435 * tree-pass.h: Add make_pass_lower_switch.
17437 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17440 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
17443 2017-08-29 Richard Biener <rguenther@suse.de>
17445 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
17446 we do not add a DW_AT_inline attribute twice.
17447 (gen_subprogram_die): Remove code setting DW_AT_inline on
17448 DECL_ABSTRACT_P nodes.
17450 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
17452 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
17453 calls to internal functions.
17454 (gimplify_modify_expr): Likewise.
17455 * tree-call-cdce.c (use_internal_fn): Likewise.
17456 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
17457 (convert_to_divmod): Set the nothrow flag.
17458 * tree-if-conv.c (predicate_mem_writes): Likewise.
17459 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
17460 (vectorizable_call): Likewise.
17461 (vectorizable_store): Likewise.
17462 (vectorizable_load): Likewise.
17463 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
17464 (vect_recog_mask_conversion_pattern): Likewise.
17466 2017-08-29 Martin Liska <mliska@suse.cz>
17469 * gcc.c (driver_handle_option): Add new argument.
17470 * opts-common.c (handle_option): Pass
17471 target_option_override_hook.
17472 * opts-global.c (lang_handle_option): Add new option.
17473 (set_default_handlers): Add new argument.
17474 (decode_options): Likewise.
17475 * opts.c (target_handle_option): Likewise.
17476 (common_handle_option): Call target_option_override_hook.
17477 * opts.h (struct cl_option_handler_func): Add hook for
17478 target option override.
17479 (struct cl_option_handlers): Likewise.
17480 (set_default_handlers): Add new argument.
17481 (decode_options): Likewise.
17482 (common_handle_option): Likewise.
17483 (target_handle_option): Likewise.
17484 * toplev.c (toplev::main): Pass targetm.target_option.override
17487 2017-08-29 Richard Biener <rguenther@suse.de>
17488 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
17490 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
17491 life to the active subtree.
17493 2017-08-28 Jeff Law <law@redhat.com>
17495 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
17496 derive_equivalences.
17497 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
17498 Code moved into....
17499 (edge_info::derive_equivalences): New private member function
17501 * tree-ssa-dom.c (class edge_info): Changed from a struct
17502 to a class. Add ctor/dtor, methods and data members.
17503 (edge_info::edge_info): Renamed from allocate_edge_info.
17504 Initialize additional members.
17505 (edge_info::~edge_info): New.
17506 (free_dom_edge_info): Delete the edge info.
17507 (record_edge_info): Use new class & associated member functions.
17508 Tighten forms for testing for edge equivalences.
17509 (record_temporary_equivalences): Iterate over the simple
17510 equivalences rather than assuming there's only one per edge.
17511 (cprop_into_successor_phis): Iterate over the simple
17512 equivalences rather than assuming there's only one per edge.
17513 (optimize_stmt): Use operand_equal_p rather than pointer
17514 equality for mini-DSE code.
17516 2017-08-28 Nathan Sidwell <nathan@acm.org>
17518 * gcc.c (execute): Fold SIGPIPE handling into switch
17519 statement. Adjust internal error message.
17521 2017-08-28 Richard Biener <rguenther@suse.de>
17524 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
17525 Do nothing for removed DIEs.
17527 2017-08-28 Richard Biener <rguenther@suse.de>
17529 PR tree-optimization/81977
17530 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
17533 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
17536 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
17537 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
17538 using targetm.gen_mem_thread_fence.
17540 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
17543 * config/i386/i386.md (*<btsc><mode>): Change operand 2
17544 predicate to register_operand. Reorder operands.
17545 (*btr<mode>): Ditto.
17546 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
17547 (*btr<mode>_mask): Ditto.
17549 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
17551 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
17552 * config/rs6000/xmmintrin.h: New file.
17553 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
17555 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17558 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
17559 parameter and_insn and return it.
17560 (recombine_lvx_pattern): Insert a copy to ensure availability of
17561 the base register of the copied masking operation at the point of
17562 the instruction replacement.
17563 (recombine_stvx_pattern): Likewise.
17565 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
17567 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
17568 undocumented switches.
17569 (-mpower9-dform-vector): Likewise.
17570 (-mpower9-dform): Likewise.
17571 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
17572 comments to delete references to -mpower9-dform* switches.
17573 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
17574 Delete reference to -mpower9-dform* switches, test for
17575 -mpower9-vector instead.
17576 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
17577 (OTHER_P9_VECTOR_MASKS): Likewise.
17578 (POWERPC_MASKS): Likewise.
17579 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
17580 tests against -mpower9-dform* to -mpower9-vector. Delete code
17581 that checked for -mpower9-dform* consistancy with other options.
17582 Add test for -mpower9-misc to enable other power9 switches.
17583 (rs6000_init_hard_regno_mode_ok): Likewise.
17584 (rs6000_option_override_internal): Likewise.
17585 (rs6000_emit_prologue): Likewise.
17586 (rs6000_emit_epilogue): Likewise.
17587 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
17588 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
17589 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
17591 (emit_fusion_p9_store): Likewise.
17592 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
17593 resetting these macros if the assembler does not support ISA 3.0
17595 (TARGET_P9_DFORM_VECTOR): Likewise.
17596 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
17597 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
17599 2017-08-25 Alan Modra <amodra@gmail.com>
17601 PR rtl-optimization/81747
17602 * cse.c (cse_extended_basic_block): Don't attempt to record
17603 equivalences for degenerate conditional branches that branch
17604 to their fall-through.
17606 2017-08-24 Martin Sebor <msebor@redhat.com>
17608 PR middle-end/81908
17609 * gimple-fold.c (size_must_be_zero_p): New function.
17610 (gimple_fold_builtin_memory_op): Call it.
17612 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
17614 * config/rs6000/mm_malloc.h: New file.
17616 2017-08-24 Bin Cheng <bin.cheng@arm.com>
17618 PR tree-optimization/81913
17619 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
17620 analysis when either IVs in condition can wrap.
17622 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
17624 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
17625 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
17627 2017-08-24 Richard Biener <rguenther@suse.de>
17630 * targhooks.c (default_target_can_inline_p): Properly
17631 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
17632 is present and always compare.
17633 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
17634 infer -mfpmath=sse from TARGET_SSE_P.
17635 (ix86_can_inline_p): Properly use target_option_default_node when
17636 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
17638 2017-08-24 Richard Biener <rguenther@suse.de>
17641 * dwarf2out.c (output_die): Handle flag_generate_offload like
17643 (output_comp_unit): Likewise.
17644 (gen_array_type_die): Likewise.
17645 (dwarf2out_early_finish): Likewise.
17646 (note_variable_value_in_expr): Likewise.
17647 (dwarf2out_finish): Likewise. Adjust assert.
17648 * cgraphunit.c (symbol_table::compile): Move setting of
17649 flag_generate_offload earlier ...
17650 (symbol_table::finalize_compilation_unit): ... here, before
17651 early debug finalization.
17653 2017-08-24 Richard Biener <rguenther@suse.de>
17655 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
17656 and ipa-fnsummary.h.
17657 (ix86_can_inline_p): When ix86_fpmath flags do not match
17658 check whether the callee uses FP math at all.
17660 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
17662 PR middle-end/81931
17663 * tree-ssanames.c (get_nonzero_bits): Use element_precision
17664 instead of TYPE_PRECISION.
17666 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
17667 Alan Hayward <alan.hayward@arm.com>
17668 David Sherwood <david.sherwood@arm.com>
17670 * combine.c (make_extraction): Use subreg_offset_from_lsb.
17672 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
17674 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
17676 (ix86_frame::stack_realign_allocate): New field.
17677 (struct machine_frame_state): Modify comments.
17678 (machine_frame_state::sp_realigned_fp_end): New field.
17679 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
17680 layout calculation.
17681 (sp_valid_at): Add assertion to assure no attempt to access invalid
17682 offset of a realigned stack.
17683 (fp_valid_at): Likewise.
17684 (choose_baseaddr): Modify comments.
17685 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
17686 ix86_expand_prologue.
17687 (ix86_expand_prologue): Modify stack realignment and allocation.
17688 (ix86_expand_epilogue): Modify comments.
17689 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
17690 avx2_runtime, avx512f, and avx512f_runtime.
17692 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
17694 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
17695 (mstackrealign): Do not init to -1.
17696 * config/i386/i386.c (ix86_option_override_internal):
17697 Check opts_set, not opts when setting default value of
17698 opts->x_ix86_force_align_arg_pointer.
17700 2017-08-23 Richard Biener <rguenther@suse.de>
17702 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
17703 lang_hooks.decl_printable_name.
17704 * print-rtl-function.c (print_rtx_function): Likewise.
17705 * tree-pretty-print.c (dump_function_header): Likewise.
17707 2017-08-23 Richard Biener <rguenther@suse.de>
17710 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
17711 -g0 at compile-time.
17713 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17715 PR middle-end/19706
17716 * doc/sourcebuild.texi (Other hardware attributes):
17719 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17721 PR middle-end/19706
17722 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
17723 Fix single-use check.
17725 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
17727 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
17729 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17731 * doc/install.texi: Modify to add more details on running selected
17734 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17736 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
17737 is combined with -mabi=ms.
17738 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
17741 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17743 PR tree-optimization/81488
17744 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
17745 and cached_basis fields.
17746 (MAX_SPREAD): New constant.
17747 (alloc_cand_and_find_basis): Initialize new fields.
17748 (clear_visited): New function.
17749 (create_phi_basis_1): Rename from create_phi_basis, set visited
17750 and cached_basis fields.
17751 (create_phi_basis): New wrapper function.
17752 (phi_add_costs_1): Rename from phi_add_costs, add spread
17753 parameter, set visited field, short-circuit when limits reached.
17754 (phi_add_costs): New wrapper function.
17755 (record_phi_increments_1): Rename from record_phi_increments, set
17757 (record_phi_increments): New wrapper function.
17758 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
17759 (phi_incr_cost): New wrapper function.
17760 (all_phi_incrs_profitable_1): Rename from
17761 all_phi_incrs_profitable, set visited field.
17762 (all_phi_incrs_profitable): New wrapper function.
17764 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17765 Alan Hayward <alan.hayward@arm.com>
17766 David Sherwood <david.sherwood@arm.com>
17768 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
17769 that takes the outer and inner modes.
17770 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
17771 comparison as the canonical test for a paradoxical subreg.
17772 * combine.c (simplify_set): Use paradoxical_subreg_p.
17773 (make_extraction): Likewise.
17774 (force_to_mode): Likewise.
17775 (rtx_equal_for_field_assignment_p): Likewise.
17776 (gen_lowpart_for_combine): Likewise.
17777 (simplify_comparison): Likewise.
17778 * cse.c (equiv_constant): Likewise.
17779 * expmed.c (store_bit_field_1): Likewise.
17780 * final.c (alter_subreg): Likewise.
17781 * fwprop.c (propagate_rtx): Likewise.
17782 (forward_propagate_subreg): Likewise.
17783 * ira-conflicts.c (ira_build_conflicts): Likewise.
17784 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
17785 * lra-constraints.c (curr_insn_transform): Likewise.
17786 (split_reg): Likewise.
17787 * lra-eliminations.c (move_plus_up): Likewise.
17788 (lra_eliminate_regs_1): Likewise.
17789 * recog.c (general_operand): Likewise.
17790 * ree.c (combine_reaching_defs): Likewise.
17791 * reload.c (push_reload): Likewise.
17792 (find_reloads): Likewise.
17793 * reload1.c (elimination_effects): Likewise.
17794 (compute_reload_subreg_offset): Likewise.
17795 (choose_reload_regs): Likewise.
17796 * rtlanal.c (subreg_lsb_1): Likewise.
17797 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17798 (simplify_subreg): Likewise.
17799 * var-tracking.c (track_loc_p): Likewise.
17800 * emit-rtl.c (byte_lowpart_offset): Likewise.
17801 (paradoxical_subreg_p): Delete out-of-line definition.
17803 2017-08-22 Jeff Law <law@redhat.com>
17805 PR tree-optimization/81741
17806 PR tree-optimization/71947
17807 * tree-ssa-dom.c: Include tree-inline.h.
17808 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
17809 equivalences if one is more expensive to compute than the other.
17810 * tree-ssa-scopedtables.h (class const_or_copies): Make
17811 record_const_or_copy_raw method private.
17812 (class avail_exprs_stack): New method simplify_binary_operation.
17813 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
17814 avail_exprs_stack::simplify_binary_operation as needed.
17815 (avail_exprs_stack::simplify_binary_operation): New function.
17817 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
17819 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
17820 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
17821 (DOT_SYMBOLS): Likewise.
17822 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
17823 (RELOCATABLE_NEEDS_FIXUP): Likewise.
17824 (RS6000_ABI_NAME): Likewise.
17825 (TARGET_CMODEL): Likewise.
17826 (TOC_SECTION_ASM_OP): Likewise.
17827 (SET_CMODEL): New macro.
17828 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
17830 2017-08-22 Richard Biener <rguenther@suse.de>
17832 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
17833 to allow for free-lang-data replacements similar to verify_type_variant.
17835 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17836 Alan Hayward <alan.hayward@arm.com>
17837 David Sherwood <david.sherwood@arm.com>
17839 * config/aarch64/aarch64.md (casesi): Use DImode rather than
17840 VOIDmode for the LABEL_REF.
17842 2017-08-22 Richard Biener <rguenther@suse.de>
17844 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
17846 2017-08-22 Richard Biener <rguenther@suse.de>
17848 * common.opt (feliminate-dwarf2-dups): Ignore.
17849 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
17850 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
17851 same_die_p_wrap, compute_section_prefix,
17852 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
17853 (comdat_symbol_id, comdat_symbol_number): Likewise.
17854 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
17856 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
17857 (output_die): Mark unreachable path unreachable.
17858 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
17859 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
17860 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
17861 (dwarf2out_early_finish): Likewise.
17863 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
17865 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
17867 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
17870 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
17871 not VAR_P. Filter attribute warnings with OPT_Wattributes.
17872 (avr_attribute_table) <io, io_low, address>: Initialize
17873 .decl_required with true.
17875 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
17877 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
17878 undocumented debugging options.
17879 (-mvsx-scalar-double): Likewise.
17880 (-mallow-df-permute): Likewise.
17881 (-mvectorize-builtins): Likewise.
17882 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17883 (rs6000_builtin_vectorized_function): Likewise.
17884 (rs6000_builtin_md_vectorized_function): Likewise.
17885 (rs6000_opt_vars): Likewise.
17887 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
17890 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
17891 (*btrq_imm): Rename from *btrq.
17892 (*btcq_imm): Rename from *btcq.
17893 (btsc): New code attribute.
17894 (*<btsc><mode>): New insn pattern.
17895 (*btr<mode>): Ditto.
17896 (*<btsc><mode>_mask): New insn_and_split pattern.
17897 (*btr<mode>_mask): Ditto.
17899 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17900 Alan Hayward <alan.hayward@arm.com>
17901 David Sherwood <david.sherwood@arm.com>
17903 * function.c (pad_below): Simplify padding calculation.
17905 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17906 Alan Hayward <alan.hayward@arm.com>
17907 David Sherwood <david.sherwood@arm.com>
17909 * target.def (function_prologue): Remove frame size argument.
17910 (function_epilogue): Likewise.
17911 * doc/tm.texi: Regenerate.
17912 * final.c (final_start_function): Update call to function_prologue.
17913 (final_end_function): Update call to function_epilogue.
17914 (default_function_pro_epilogue): Remove frame size argument.
17915 * output.h (default_function_pro_epilogue): Likewise.
17916 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
17917 (arm_output_function_prologue): Likewise.
17918 * config/frv/frv.c (frv_function_prologue): Likewise.
17919 (frv_function_epilogue): Likewise.
17920 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
17921 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
17922 (ia64_output_function_epilogue): Likewise.
17923 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
17924 (m32r_output_function_epilogue): Likewise.
17925 * config/microblaze/microblaze.c (microblaze_function_prologue)
17926 (microblaze_function_epilogue): Likewise.
17927 * config/mips/mips.c (mips_output_function_prologue): Likewise.
17928 (mips_output_function_epilogue): Likewise.
17929 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
17930 (mmix_target_asm_function_epilogue): Likewise.
17931 * config/msp430/msp430.c (msp430_start_function): Likewise.
17932 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
17933 (nds32_asm_function_epilogue): Likewise.
17934 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
17935 * config/pa/pa.c (pa_output_function_prologue): Likewise.
17936 (pa_output_function_epilogue): Likewise.
17937 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
17938 (rs6000_output_function_epilogue): Likewise.
17939 * config/rl78/rl78.c (rl78_start_function): Likewise.
17940 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
17941 (rs6000_output_function_epilogue): Likewise.
17942 * config/rx/rx.c (rx_output_function_prologue): Likewise.
17943 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
17944 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
17945 (sparc_asm_function_epilogue): Likewise.
17947 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17949 * tree.h (type_has_mode_precision_p): New function.
17950 * convert.c (convert_to_integer_1): Use it.
17951 * expr.c (expand_expr_real_2): Likewise.
17952 (expand_expr_real_1): Likewise.
17953 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
17954 * match.pd: Likewise.
17955 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
17956 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
17957 * tree-tailcall.c (process_assignment): Likewise.
17958 * tree-vect-loop.c (vectorizable_reduction): Likewise.
17959 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
17960 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
17961 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17962 (vectorizable_assignment): Likewise.
17963 (vectorizable_shift): Likewise.
17964 (vectorizable_operation): Likewise.
17965 * tree-vrp.c (register_edge_assert_for_2): Likewise.
17967 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
17969 * match.pd: Add pow (C, x) simplification.
17971 2017-08-21 Richard Biener <rguenther@suse.de>
17973 PR tree-optimization/81900
17974 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
17975 for blocks with abnormal predecessors.
17976 (compute_antic): Do not set visited flag prematurely.
17978 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
17981 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
17983 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17985 * stor-layout.h (vector_type_mode): Move to...
17986 * tree.h (vector_type_mode): ...here.
17987 * stor-layout.c (vector_type_mode): Move to...
17988 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
17990 2017-08-21 Richard Biener <rguenther@suse.de>
17992 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
17993 register_external_die hooks.
17994 (debug_false_tree_charstarstar_uhwistar): Declare.
17995 (debug_nothing_tree_charstar_uhwi): Likewise.
17996 * debug.c (do_nothing_debug_hooks): Adjust.
17997 (debug_false_tree_charstarstar_uhwistar): New do nothing.
17998 (debug_nothing_tree_charstar_uhwi): Likewise.
17999 * dbxout.c (dbx_debug_hooks): Adjust.
18000 (xcoff_debug_hooks): Likewise.
18001 * sdbout.c (sdb_debug_hooks): Likewise.
18002 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
18003 * dwarf2out.c (macinfo_label_base): New global.
18004 (dwarf2out_register_external_die): New function for the
18005 register_external_die hook.
18006 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
18007 (dwarf2_debug_hooks): Use them.
18008 (dwarf2_lineno_debug_hooks): Adjust.
18009 (struct die_struct): Add with_offset flag.
18010 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
18011 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
18012 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
18013 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
18014 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
18015 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
18016 defining section names for the early LTO debug variants.
18017 (reset_indirect_string): New helper.
18018 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
18019 (print_dw_val): Add support for offsetted symbol references.
18020 (get_ultimate_context): Split out from is_cxx.
18021 (is_cxx): Use get_ultimate_context.
18022 (is_fortran): Add decl overload.
18023 (compute_comp_unit_symbol): Split out worker from
18024 compute_section_prefix.
18025 (compute_section_prefix): Call compute_comp_unit_symbol and
18026 set comdat_type_p here.
18027 (output_die): Skip DIE symbol output for the LTO added one.
18028 Handle DIE symbol references with offset.
18029 (output_comp_unit): Guard section name mangling properly.
18030 For LTO debug sections emit a symbol at the section beginning
18031 which we use to refer to its DIEs.
18032 (add_abstract_origin_attribute): For DIEs registered via
18033 dwarf2out_register_external_die directly refer to the early
18034 DIE rather than indirectly through the shadow one we created.
18035 Remove obsolete call to dwarf2out_abstract_function for
18036 non-function/block origins.
18037 (gen_array_type_die): When generating early LTO debug do
18038 not emit DW_AT_string_length.
18039 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
18040 late when in LTO. As suggested place a gcc_unreachable for
18041 the DECL_ABSTRACT_P case.
18042 (gen_subprogram_die): Avoid another specification DIE
18043 for early built declarations/definitions for the late LTO case.
18044 (gen_variable_die): Add type references for late duplicated VLA dies
18046 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
18047 we have the abstract instance already.
18048 (process_scope_var): Adjust decl DIE contexts in LTO which
18049 first puts them in limbo.
18050 (gen_decl_die): Do not generate type DIEs late apart from
18051 types for VLAs or for decls we do not yet have a DIE. Do not
18052 call dwarf2out_abstract_function late.
18053 (dwarf2out_early_global_decl): Make sure to create DIEs
18054 for abstract instances of a decl first.
18055 (dwarf2out_late_global_decl): Adjust comment.
18056 (output_macinfo_op): With multiple macro sections use
18057 macinfo_label_base to distinguish labels.
18058 (output_macinfo): Likewise. Update macinfo_label_base.
18059 Pass in the line info label.
18060 (note_variable_value_in_expr): When generating LTO resolve
18061 all variable values here by generating DIEs as needed.
18062 (init_sections_and_labels): Add early LTO debug flag parameter
18063 and generate different sections and names if set. Add generation
18064 counter for the labels so we can have multiple of them.
18065 (reset_dies): Helper to allow DIEs to be output multiple times.
18066 (dwarf2out_finish): When outputting DIEs to the fat part of an
18067 LTO object first reset DIEs.
18068 (dwarf2out_early_finish): Output early DIEs when generating LTO.
18069 (modified_type_die): Check for decl_ultimate_origin being self
18071 (gen_type_die_with_usage): Likewise.
18072 (gen_typedef_die): Allow decl_ultimate_origin being self.
18073 (set_decl_abstract_flags): Remove.
18074 (set_block_abstract_flags): Likewise.
18075 (dwarf2out_abstract_function): Treat the early generated DIEs
18076 as the abstract copy and only add DW_AT_inline and
18077 DW_AT_artificial here and call set_decl_origin_self.
18078 If the DIE has an abstract origin don't do anything.
18079 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
18080 if we have none yet (Go fails to build one, PR78628).
18081 (variably_modified_type_p): Prevent endless recursion for Ada
18082 cyclic pointer types.
18083 * lto-streamer-in.c: Include debug.h.
18084 (dref_queue): New global.
18085 (lto_read_tree_1): Stream in DIE references.
18086 (lto_input_tree): Register DIE references.
18087 (input_function): Stream DECL_DEBUG_ARGS.
18088 * lto-streamer-out.c: Include debug.h.
18089 (lto_write_tree_1): Output DIE references.
18090 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
18091 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
18092 (output_function): Stream DECL_DEBUG_ARGS.
18093 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
18094 Stream DECL_ABSTRACT_ORIGIN.
18095 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
18096 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
18097 DECL_CONTEXT for file-scope decls.
18098 * lto-streamer.h (struct dref_entry): Declare.
18099 (dref_queue): Likewise.
18100 * cfgexpand.c (pass_expand::execute): Do not call the
18101 outlining_inline_function hook here.
18102 * lto-wrapper.c (debug_obj): New global.
18103 (tool_cleanup): Unlink it if required.
18104 (debug_objcopy): New function.
18105 (run_gcc): Handle early debug sections in the IL files by
18106 extracting them to separate files, partially linkin them and
18107 feeding the result back as result to the linker.
18108 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
18109 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
18110 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
18111 sections into a separate segment.
18112 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
18114 (darwin_asm_dwarf_section): Likewise.
18115 (darwin_asm_output_dwarf_offset): Likewise.
18116 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
18118 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18119 Alan Hayward <alan.hayward@arm.com>
18120 David Sherwood <david.sherwood@arm.com>
18122 * read-md.h (md_reader::record_potential_iterator_use): Replace
18123 pointer argument with an rtx and an index.
18124 * read-rtl.c (iterator_group::apply_iterator): Likewise.
18125 (apply_mode_iterator): Likewise.
18126 (apply_code_iterator): Likewise.
18127 (apply_int_iterator): Likewise.
18128 (apply_subst_iterator): Likewise.
18129 (record_iterator_use): Likewise.
18130 (record_attribute_use): Likewise.
18131 (md_reader::record_potential_iterator_use): Likewise. Update calls
18132 to record_iterator_use and apply_iterator.
18133 (iterator_use): Replace ptr with x and index.
18134 (attribute_use): Likewise.
18135 (apply_attribute_uses): Update calls to apply_iterator.
18136 (apply_iterators): Likewise. Update initialization of iterator_use.
18137 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
18138 and record_potential_iterator_use.
18139 (rtx_reader::read_rtx_operand): Likewise.
18141 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18142 Alan Hayward <alan.hayward@arm.com>
18143 David Sherwood <david.sherwood@arm.com>
18145 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
18148 2017-08-21 Richard Biener <rguenther@suse.de>
18150 PR middle-end/81884
18151 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
18152 at struct end conservatively when comparing common bases.
18154 2017-08-21 Richard Biener <rguenther@suse.de>
18156 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
18157 (mem_ref_in_stmt): Remove.
18158 (determine_max_movement): Use ref index to get at the reference.
18159 (invariantness_dom_walker::before_dom_children): Deal with
18160 lim data already initialized.
18161 (gather_mem_refs_stmt): Initialize lim data and record ref index.
18163 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
18165 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
18166 (TARGET_ISA_ROUND): Ditto.
18167 (TARGET_ROUND): Ditto.
18168 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
18169 * config/i386/i386.md: Ditto.
18170 * config/i386/sse.md: Ditto.
18171 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
18172 with OPTION_MASK_ISA_SSE4_1.
18174 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
18177 * doc/extend.texi (x86 Built-in Functions): Correct the name of
18178 __builtin_ia32_lzcnt_u16.
18180 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
18183 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
18184 (rs6000_set_current_function): Rewrite function to use it.
18186 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
18189 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
18190 and TYPE_WARN_IF_NOT_ALIGN.
18191 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
18192 (handle_warn_if_not_align): New.
18193 (place_union_field): Call handle_warn_if_not_align.
18194 (place_field): Call handle_warn_if_not_align.
18195 Copy TYPE_WARN_IF_NOT_ALIGN.
18196 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
18197 (layout_type): Likewise.
18198 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
18200 (tree_decl_common): Add warn_if_not_align.
18201 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
18202 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
18203 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
18204 (DECL_WARN_IF_NOT_ALIGN): Likewise.
18205 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
18206 * doc/extend.texi: Document warn_if_not_aligned attribute.
18207 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
18209 2017-08-17 Martin Liska <mliska@suse.cz>
18212 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
18213 (get_data_dependence): Use it as pointer type.
18214 (distribute_loop): Likewise.
18216 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18218 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
18219 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
18220 (p8_vmrgow_<mode>_direct): New define_insn.
18221 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
18222 handle endianness for vmrgew and vmrgow permute patterns.
18224 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
18226 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
18227 * config/rs6000/rs6000-cpus.def: Remove comment.
18228 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
18229 (POWERPC_MASKS): Likewise.
18230 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
18231 use of TARGET_VSX_TIMODE.
18232 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
18233 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
18234 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
18235 (rs6000_option_override_internal): Remove dead code.
18236 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
18237 (rs6000_legitimize_reload_address): Likewise.
18238 (rs6000_legitimate_address_p): Likewise.
18239 (rs6000_opt_masks): Delete "vsx-timode".
18240 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
18241 from function comment.
18242 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
18243 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
18244 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
18246 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
18247 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
18248 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
18249 (VSX_TI): Likewise.
18251 (define_peephole2): Likewise.
18253 2017-08-17 Martin Sebor <msebor@redhat.com>
18256 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
18257 past the end of an array.
18258 (test_pp_format): Add test cases.
18260 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
18262 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
18263 missing ECF_NOTHROW flags.
18265 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
18268 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
18269 operands residing in integer registers.
18270 (*vsx_le_perm_load_<mode>): Likewise.
18271 (*vsx_le_perm_store_<mode>): Likewise.
18272 (define_peephole2): Add peepholes to optimize the above.
18274 2017-08-17 Marek Polacek <polacek@redhat.com>
18276 PR middle-end/81814
18277 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
18278 to mimic what shorten_compare did. Change the return type to bool.
18279 (fold_cond_expr_with_comparison): Update call to
18280 operand_equal_for_comparison_p.
18281 (fold_ternary_loc): Likewise.
18283 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
18285 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
18287 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
18289 2017-08-17 Richard Biener <rguenther@suse.de>
18291 * tree-ssa-structalias.c (solve_graph): When propagating
18292 to successors update the graphs succ edges and avoid duplicate work.
18294 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
18297 * config/i386/i386.c (ix86_option_override_internal): Save target
18298 specific options after ix86_stack_protector_guard_reg was changed.
18300 2017-08-17 Richard Biener <rguenther@suse.de>
18302 PR tree-optimization/81827
18303 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
18304 (new_var_info): Initialize it conservatively.
18305 (get_call_vi): Mark register vars.
18306 (new_scalar_tmp_constraint_exp): Likewise.
18307 (handle_rhs_call): Likewise.
18308 (handle_const_call): Likewise.
18309 (create_function_info_for): Likewise.
18310 (solve_constraints): Sort varinfos to separate register from
18311 non-register vars to pack points-to solution bitmaps during
18314 2017-08-17 Marek Polacek <polacek@redhat.com>
18316 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
18318 2017-08-17 Richard Biener <rguenther@suse.de>
18320 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
18321 to true when overflow is undefined and we saturated the result.
18323 2017-08-17 Alan Modra <amodra@gmail.com>
18326 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
18327 Don't use store multiple if only one reg needs saving.
18328 (interesting_frame_related_regno): New function.
18329 (rs6000_frame_related): Don't emit frame info for regs that
18331 (rs6000_emit_epilogue): Likewise.
18333 2017-08-16 Nathan Sidwell <nathan@acm.org>
18335 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
18336 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
18337 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
18338 * tree.c (free_lang_data_in_type): Use else-if chain. Always
18339 clear TYPE_LANG_1. Remove obsolete member-function stripping.
18340 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
18341 (verify_type): Adjust for TYPE_BINFO move.
18342 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
18343 process TYPE_BINFO directly.
18344 (hash_tree): Likewise.
18345 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
18347 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
18350 2017-08-16 David Malcolm <dmalcolm@redhat.com>
18352 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
18354 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
18357 * config/i386/i386.md (*anddi_1_btr): Change predicates of
18358 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
18359 Add ix86_binary_operator_ok to insn constraint.
18360 (*iordi_1_bts): Ditto.
18361 (*xordi_1_btc): Ditto.
18362 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
18363 Update corresponding peephole2 pattern.
18367 2017-08-16 Bin Cheng <bin.cheng@arm.com>
18369 PR tree-optimization/81832
18370 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
18371 copy loop header which has IFN_LOOP_DIST_ALIAS call.
18373 2017-08-16 Marek Polacek <polacek@redhat.com>
18376 * fold-const.c (fold_indirect_ref_1): Restore original behavior
18377 regarding size_zero_node.
18379 2017-08-16 Martin Liska <mliska@suse.cz>
18382 * config.gcc: Respect previously set extra_objs in case
18385 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
18387 PR tree-optimization/81835
18388 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
18389 the phi SSA_NAME. Check that the condition in a COND_EXPR does
18390 not depend on the phi.
18392 2017-08-16 Alan Modra <amodra@gmail.com>
18394 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
18397 2017-08-16 Alan Modra <amodra@gmail.com>
18399 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
18400 (save_reg_p): ..into this. Update all callers.
18401 (first_reg_to_save): Simplify.
18403 2017-08-16 Alan Modra <amodra@gmail.com>
18405 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
18408 2017-08-15 Joseph Myers <joseph@codesourcery.com>
18412 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
18413 constant count if that count is less than 32.
18415 2017-08-15 Nathan Sidwell <nathan@acm.org>
18417 * gcc.c (execute): Emit friendlier message if inferior is killed
18418 by an external cause.
18420 2017-08-15 Richard Biener <rguenther@suse.de>
18422 PR tree-optimization/81790
18423 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
18424 CONSTRUCTORs from simplifying and VN.
18426 2017-08-14 Martin Sebor <msebor@redhat.com>
18428 * builtin-attrs.def: Add comments.
18430 2017-08-14 Martin Sebor <msebor@redhat.com>
18433 * doc/extend.texi (attribute nonstring): Document new attribute.
18435 2017-08-14 Martin Sebor <msebor@redhat.com>
18438 * tree-diagnostic.c (default_tree_printer): Handle %G.
18439 * gimple-pretty-print.h (percent_G_format): Declare new function.
18440 * gimple-pretty-print.c (percent_G_format): Define.
18441 * tree-pretty-print.c (percent_K_format): Add argument.
18443 2017-08-14 Martin Sebor <msebor@redhat.com>
18445 PR translation/79998
18446 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
18447 Remove a stray space.
18449 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
18452 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
18453 (*iordi_1_bts): Ditto.
18454 (*xordi_1_btc): Ditto.
18456 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18459 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
18460 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18462 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
18463 quoted strings, and make more translator-friendly.
18464 (darwin_rs6000_override_options): Likewise.
18465 (rs6000_option_override_internal): Likewise.
18466 (rs6000_return_in_memory): Fix overlong line.
18467 (init_cmulative_args): Use quoted strings, and make more
18468 translator-friendly.
18469 (rs6000_pass_by_reference): Fix overlong line.
18470 (def_builtin): Use quoted strings.
18471 (altivec_expand_predicate_builtin): Use quoted strings, and make
18472 more translator-friendly.
18473 (htm_expand_builtin): Use quoted strings.
18474 (cpu_expand_builtin): Use quoted strings, and make more
18475 translator-friendly.
18476 (altivec_expand_builtin): Likewise.
18477 (paired_expand_predicate_builtin): Likewise.
18478 (rs6000_invalid_builtin): Likewise.
18479 (builtin_function_type): Use quoted strings.
18480 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
18481 more translator-friendly.
18482 (rs6000_trampoline_init): Likewise.
18483 (rs6000_handle_altivec_attribute): Likewise.
18484 (rs6000_inner_target_options): Use quoted strings.
18485 (rs6000_disable_incompatible_switches): Likewise.
18486 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
18487 strings, and make more translator-friendly.
18488 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
18490 2017-08-14 Bin Cheng <bin.cheng@arm.com>
18492 PR tree-optimization/81799
18493 * tree-loop-distribution.c (version_loop_by_alias_check): Force
18494 cond_expr to simple gimple operand.
18496 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
18498 PR middle-end/46932
18499 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
18501 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
18505 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
18506 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
18507 TARGET_GASISR_PROLOGUES.
18508 * config/avr/avr.c (avr_option_override): Same.
18509 (avr_pass_pre_proep::execute): Same.
18511 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
18514 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
18515 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
18517 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
18519 * config/i386/i386.md (*load_tp_<mode>): Redefine as
18520 define_insn_and_split. Split to a memory load from 0 in
18521 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
18522 using PTR mode iterator.
18523 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
18524 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
18525 (*add_tp_<mode>): Redefine as define_insn_and_split.
18526 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
18527 address space. Merge with *add_tp_x32 using PTR mode iterator.
18528 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
18529 Split to an add with a memory load from 0 in
18530 DEFAULT_TLS_SEG_REG address space.
18532 2017-08-12 Andrew Pinski <apinski@cavium.com>
18534 * config/aarch64/aarch64-option-extensions.def (rdma):
18535 Fix feature string to what Linux prints out in /proc/cpuinfo.
18537 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
18540 * dwarf2out.c (modified_type_die): For C typedef types that have
18541 an ultimate origin, process the ultimate origin instead of the
18543 (gen_typedef_die): Assert that input DECLs have no ultimate
18545 (gen_type_die_with_usage): For typedef variants that have an
18546 ultimate origin, just call gen_decl_die on the original DECL.
18547 (process_scope_var): Avoid creating DIEs for local typedefs and
18548 concrete static variables.
18550 2017-08-12 Alan Modra <amodra@gmail.com>
18554 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
18555 match gnu-user.h startfile.
18556 (ENDFILE_LINUX_SPEC): Similarly.
18558 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
18561 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
18563 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
18565 2017-08-11 Tamar Christina <tamar.christina@arm.com>
18566 * config/aarch64/aarch64.md (mov<mode>): Change.
18567 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
18568 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
18569 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
18571 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
18573 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
18574 for storage order barriers.
18576 2017-08-11 Martin Liska <mliska@suse.cz>
18579 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
18580 variables of void type.
18582 2017-08-11 Martin Liska <mliska@suse.cz>
18584 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
18585 TARGET_SUPPORTS_ALIASES.
18586 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
18587 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
18588 (optimize_weakref): Likewise.
18589 * symtab.c (symtab_node::noninterposable_alias): Likewise.
18590 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
18591 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
18593 2017-08-11 Martin Liska <mliska@suse.cz>
18596 * config/i386/i386.c (make_resolver_func): Do complete
18597 refactoring of the function.
18599 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
18602 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
18603 * config/i386/i386.c (ix86_stack_protect_guard): Use
18604 ix86_stack_protect_guard_symbol_str to generate varible declaration.
18605 * doc/invoke.texi (x86 Options): Document
18606 -mstack-protector-guard-symbol= option.
18608 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
18610 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
18611 * config/i386/i386.c (ix86_split_stack_guard): New function.
18612 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
18613 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
18614 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
18615 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
18616 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
18617 (split_stack_space_check): Call ix86_split_stack_guard.
18619 2017-08-10 Martin Sebor <msebor@redhat.com>
18621 * print-tree.c (print_node): Print location using the established
18623 Replace spaces with colons.
18624 (debug_raw, debug): Ditto.
18626 2017-08-10 Martin Sebor <msebor@redhat.com>
18629 * pretty-print.c (pp_format): Correct the handling of %s precision.
18631 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
18634 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18636 (ix86_finalize_stack_frame_flags): This. Also clear
18637 frame_pointer_needed if -fno-omit-frame-pointer is used without
18639 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18640 with ix86_finalize_stack_frame_flags.
18641 (ix86_expand_epilogue): Likewise.
18642 (ix86_expand_split_stack_prologue): Likewise.
18643 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
18645 2017-08-10 Martin Liska <mliska@suse.cz>
18648 * c-attribs.c (handle_target_attribute):
18649 Report warning for an empty string argument of target attribute.
18651 2017-08-09 Jakub Jelinek <jakub@redhat.com>
18654 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
18656 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
18657 or DECL_NONLOCAL labels.
18658 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
18659 or DECL_NONLOCAL labels here.
18661 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
18663 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
18664 to indicate when early gimple folding has been disabled.
18665 (rs6000_gimple_fold_builtin): Add debug content.
18666 (rs6000_invalid_builtin): Fix whitespace.
18667 (rs6000_expand_builtin): Fix whitespace.
18668 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
18670 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
18673 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
18674 SAVE_MULTIPLE if not all the registers that saves, should be saved.
18676 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
18678 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
18679 (qdf24xx): Likewise.
18680 * config/aarch64/aarch64.md: Include falkor.md.
18681 * config/aarch64/falkor.md: New.
18683 2017-08-09 Marek Polacek <polacek@redhat.com>
18686 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
18687 * diagnostic.c (emit_diagnostic): Add a comment.
18688 (emit_diagnostic_valist): New function.
18690 2017-08-09 Marek Polacek <polacek@redhat.com>
18693 * input.c (make_location): New overload.
18694 * input.h (make_location): Declare.
18696 2017-08-08 Alan Modra <amodra@gmail.com>
18697 H.J. Lu <hongjiu.lu@intel.com>
18700 * gcc.c (NO_PIE_SPEC): Delete.
18701 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
18703 (LINK_PIE_SPEC): ..to here.
18704 (LINK_COMMAND_SPEC): Support -no-pie.
18705 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
18706 chain of crtbegin*.o selection, update for PIE_SPEC changes and
18708 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
18709 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
18710 (ENDFILE_CRTEND_SPEC): Similarly.
18712 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
18715 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
18716 (mstack-protector-guard-offset=): Ditto.
18717 * config/i386/i386.c (ix86_option_override): Handle
18718 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
18720 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
18721 ix86_stack_protect_guard_offset variables.
18722 (TARGET_STACK_PROTECT_GUARD): Always define.
18723 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
18724 and -mstack-protector-guard-offset= options.
18726 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18728 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
18729 boundary case for the last candidate.
18731 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18733 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
18734 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
18736 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18738 PR middle-end/19706
18739 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
18740 * config/aarch64/aarch64-builtins.c
18741 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
18742 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
18743 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
18745 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18746 Andrew Pinski <pinskia@gmail.com>
18748 PR middle-end/19706
18749 * internal-fn.def (XORSIGN): New.
18750 * optabs.def (xorsign_optab): New.
18751 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
18752 (convert_expand_mult_copysign): New.
18753 (pass_optimize_widening_mul::execute): Call
18754 convert_expand_mult_copysign.
18756 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18758 PR tree-optimization/81354
18759 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
18760 Insert on edges rather than explicitly creating landing pads.
18761 (analyze_candidates_and_replace): Commit edge inserts.
18763 2017-08-08 Richard Biener <rguenther@suse.de>
18765 PR middle-end/81719
18766 * tree-ssa-loop-niter.c: Include tree-dfa.h.
18767 (expand_simple_operations): Also look through ADDR_EXPRs with
18768 MEM_REF bases treating them as POINTER_PLUS_EXPR.
18770 2017-08-08 Richard Biener <rguenther@suse.de>
18772 PR tree-optimization/81723
18773 * tree-vect-slp.c (struct bst_traits): New hash traits.
18774 (bst_fail): New global.
18775 (vect_build_slp_tree_2): New worker, split out from ...
18776 (vect_build_slp_tree): ... this now wrapping it with using
18777 bst_fail set to cache SLP tree build fails. Properly handle
18779 (vect_analyze_slp_instance): Allocate and free bst_fail.
18781 2017-08-08 Martin Liska <mliska@suse.cz>
18784 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
18785 LABEL_DECLs that can be from a different function.
18787 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18789 PR tree-optimization/81744
18790 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
18791 loop's number of iterations.
18793 2017-08-08 Martin Liska <mliska@suse.cz>
18795 * asan.c: Include header files.
18796 * attribs.c (build_decl_attribute_variant): New function moved
18798 (build_type_attribute_qual_variant): Likewise.
18799 (cmp_attrib_identifiers): Likewise.
18800 (simple_cst_list_equal): Likewise.
18801 (omp_declare_simd_clauses_equal): Likewise.
18802 (attribute_value_equal): Likewise.
18803 (comp_type_attributes): Likewise.
18804 (build_type_attribute_variant): Likewise.
18805 (lookup_ident_attribute): Likewise.
18806 (remove_attribute): Likewise.
18807 (merge_attributes): Likewise.
18808 (merge_type_attributes): Likewise.
18809 (merge_decl_attributes): Likewise.
18810 (merge_dllimport_decl_attributes): Likewise.
18811 (handle_dll_attribute): Likewise.
18812 (attribute_list_equal): Likewise.
18813 (attribute_list_contained): Likewise.
18814 * attribs.h (lookup_attribute): New function moved from tree.[ch].
18815 (lookup_attribute_by_prefix): Likewise.
18816 * bb-reorder.c: Include header files.
18817 * builtins.c: Likewise.
18818 * calls.c: Likewise.
18819 * cfgexpand.c: Likewise.
18820 * cgraph.c: Likewise.
18821 * cgraphunit.c: Likewise.
18822 * convert.c: Likewise.
18823 * dwarf2out.c: Likewise.
18824 * final.c: Likewise.
18825 * fold-const.c: Likewise.
18826 * function.c: Likewise.
18827 * gimple-expr.c: Likewise.
18828 * gimple-fold.c: Likewise.
18829 * gimple-pretty-print.c: Likewise.
18830 * gimple.c: Likewise.
18831 * gimplify.c: Likewise.
18832 * hsa-common.c: Likewise.
18833 * hsa-gen.c: Likewise.
18834 * internal-fn.c: Likewise.
18835 * ipa-chkp.c: Likewise.
18836 * ipa-cp.c: Likewise.
18837 * ipa-devirt.c: Likewise.
18838 * ipa-fnsummary.c: Likewise.
18839 * ipa-inline.c: Likewise.
18840 * ipa-visibility.c: Likewise.
18842 * lto-cgraph.c: Likewise.
18843 * omp-expand.c: Likewise.
18844 * omp-general.c: Likewise.
18845 * omp-low.c: Likewise.
18846 * omp-offload.c: Likewise.
18847 * omp-simd-clone.c: Likewise.
18848 * opts-global.c: Likewise.
18849 * passes.c: Likewise.
18850 * predict.c: Likewise.
18851 * sancov.c: Likewise.
18852 * sanopt.c: Likewise.
18853 * symtab.c: Likewise.
18854 * toplev.c: Likewise.
18855 * trans-mem.c: Likewise.
18856 * tree-chkp.c: Likewise.
18857 * tree-eh.c: Likewise.
18858 * tree-into-ssa.c: Likewise.
18859 * tree-object-size.c: Likewise.
18860 * tree-parloops.c: Likewise.
18861 * tree-profile.c: Likewise.
18862 * tree-ssa-ccp.c: Likewise.
18863 * tree-ssa-live.c: Likewise.
18864 * tree-ssa-loop.c: Likewise.
18865 * tree-ssa-sccvn.c: Likewise.
18866 * tree-ssa-structalias.c: Likewise.
18867 * tree-ssa.c: Likewise.
18868 * tree-streamer-in.c: Likewise.
18869 * tree-vectorizer.c: Likewise.
18870 * tree-vrp.c: Likewise.
18871 * tsan.c: Likewise.
18872 * ubsan.c: Likewise.
18873 * varasm.c: Likewise.
18874 * varpool.c: Likewise.
18875 * tree.c: Remove functions moved to attribs.[ch].
18876 * tree.h: Likewise.
18877 * config/aarch64/aarch64.c: Add attrs.h header file.
18878 * config/alpha/alpha.c: Likewise.
18879 * config/arc/arc.c: Likewise.
18880 * config/arm/arm.c: Likewise.
18881 * config/avr/avr.c: Likewise.
18882 * config/bfin/bfin.c: Likewise.
18883 * config/c6x/c6x.c: Likewise.
18884 * config/cr16/cr16.c: Likewise.
18885 * config/cris/cris.c: Likewise.
18886 * config/darwin.c: Likewise.
18887 * config/epiphany/epiphany.c: Likewise.
18888 * config/fr30/fr30.c: Likewise.
18889 * config/frv/frv.c: Likewise.
18890 * config/ft32/ft32.c: Likewise.
18891 * config/h8300/h8300.c: Likewise.
18892 * config/i386/winnt.c: Likewise.
18893 * config/ia64/ia64.c: Likewise.
18894 * config/iq2000/iq2000.c: Likewise.
18895 * config/lm32/lm32.c: Likewise.
18896 * config/m32c/m32c.c: Likewise.
18897 * config/m32r/m32r.c: Likewise.
18898 * config/m68k/m68k.c: Likewise.
18899 * config/mcore/mcore.c: Likewise.
18900 * config/microblaze/microblaze.c: Likewise.
18901 * config/mips/mips.c: Likewise.
18902 * config/mmix/mmix.c: Likewise.
18903 * config/mn10300/mn10300.c: Likewise.
18904 * config/moxie/moxie.c: Likewise.
18905 * config/msp430/msp430.c: Likewise.
18906 * config/nds32/nds32-isr.c: Likewise.
18907 * config/nds32/nds32.c: Likewise.
18908 * config/nios2/nios2.c: Likewise.
18909 * config/nvptx/nvptx.c: Likewise.
18910 * config/pa/pa.c: Likewise.
18911 * config/pdp11/pdp11.c: Likewise.
18912 * config/powerpcspe/powerpcspe.c: Likewise.
18913 * config/riscv/riscv.c: Likewise.
18914 * config/rl78/rl78.c: Likewise.
18915 * config/rx/rx.c: Likewise.
18916 * config/s390/s390.c: Likewise.
18917 * config/sh/sh.c: Likewise.
18918 * config/sol2.c: Likewise.
18919 * config/sparc/sparc.c: Likewise.
18920 * config/spu/spu.c: Likewise.
18921 * config/stormy16/stormy16.c: Likewise.
18922 * config/tilegx/tilegx.c: Likewise.
18923 * config/tilepro/tilepro.c: Likewise.
18924 * config/v850/v850.c: Likewise.
18925 * config/vax/vax.c: Likewise.
18926 * config/visium/visium.c: Likewise.
18927 * config/xtensa/xtensa.c: Likewise.
18929 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18932 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
18933 constraints since the -mupper-regs-* switches have been
18935 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
18936 into a vector from a double word element that was extracted from
18937 another vector, and eliminate extra XXPERMDI instructions.
18938 (vsx_concat_<mode>_2): Likewise.
18939 (vsx_concat_<mode>_3): Likewise.
18940 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
18941 concat to allow optimizing inserts from previous extracts.
18943 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
18945 * config/i386/i386.c (ix86_stack_protect_guard): Generate
18946 memory reference to a SSP offset in TLS address space.
18947 (ix86_print_operand) <case '@'>: Remove.
18948 (ix86_print_operand_punct_valid_p): Remove '@' code.
18949 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
18950 UNSPEC_SP_TLS_TEST.
18951 (stack_tls_protect_set_<mode>): Remove.
18952 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
18953 (stack_tls_protect_test_<mode>): Remove.
18954 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
18956 2017-08-07 Olivier Hainque <hainque@adacore.com>
18959 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
18961 2017-08-07 Douglas Rupp <rupp@adacore.com>
18963 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
18964 variable was referenced as multidir in command.
18966 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18969 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
18972 2017-08-07 Martin Liska <mliska@suse.cz>
18974 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
18975 * config/rl78/rl78.c: Add include of attribs.h.
18976 * config/sh/sh.c: Likewise.
18977 * config/v850/v850.c: Likewise.
18979 2017-08-07 Tom de Vries <tom@codesourcery.com>
18981 PR middle-end/78266
18982 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
18984 2017-08-07 Martin Liska <mliska@suse.cz>
18986 * config/mips/mips.c: Include attribs.h.
18988 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
18991 * doc/invoke.texi: Document change in behvaior for -Ofast for
18994 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
18996 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
18998 (aarch64_pop_regs): Likewise.
18999 (aarch64_gen_load_pair): Likewise.
19000 (aarch64_save_callee_saves): Likewise.
19001 (aarch64_restore_callee_saves): Likewise.
19003 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
19005 * config/i386/i386.c: Revert the last change.
19007 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
19010 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
19012 (ix86_finalize_stack_frame_flags): This. Also clear
19013 frame_pointer_needed if -fno-omit-frame-pointer is used without
19015 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
19016 with ix86_finalize_stack_frame_flags.
19017 (ix86_expand_epilogue): Likewise.
19018 (ix86_expand_split_stack_prologue): Likewise.
19020 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
19023 * config/i386/i386.c (get_builtin_code_for_version): Set priority
19024 to P_AES for Westmere.
19026 2017-08-07 Jonathan Yong <10walls@gmail.com>
19028 * config/i386/mingw.opt (fset-stack-executable): Removed.
19029 * config/i386/cygming.opt (fset-stack-executable): Moved
19031 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
19033 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
19035 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
19037 2017-08-07 Marek Polacek <polacek@redhat.com>
19039 PR middle-end/81737
19040 * fold-const.c (fold_indirect_ref_1): Check type_domain.
19042 2017-08-07 Martin Liska <mliska@suse.cz>
19044 * attribs.h (canonicalize_attr_name): New function.
19045 (cmp_attribs): Move from c-format.c and adjusted.
19046 (is_attribute_p): Moved from tree.h.
19047 * tree-inline.c: Add new includes.
19048 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
19049 (private_is_attribute_p): Remove.
19050 (private_lookup_attribute): Likewise.
19051 (private_lookup_attribute_by_prefix): Simplify.
19052 (remove_attribute): Use is_attribute_p.
19053 * tree.h: Remove removed declarations.
19055 2017-08-07 Jakub Jelinek <jakub@redhat.com>
19057 PR middle-end/81698
19058 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
19059 instead of computing it in the function. Formatting fix.
19060 (expand_case): Don't rely on default_edge being the first edge,
19061 clear it if removing it, pass default_edge to
19062 emit_case_dispatch_table.
19063 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
19066 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
19068 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
19069 insn in the function, emit NOP after the insn.
19071 2017-08-06 Tom de Vries <tom@codesourcery.com>
19073 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
19076 2017-08-06 Tom de Vries <tom@codesourcery.com>
19078 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
19081 2017-08-04 Yury Gribov <tetra2005@gmail.com>
19083 PR tree-optimization/57371
19084 * match.pd: New pattern.
19086 2017-08-04 Marek Polacek <polacek@redhat.com>
19088 PR middle-end/81695
19089 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
19090 perform the computation in offset_int.
19092 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19094 PR tree-optimization/81136
19095 * tree-vectorizer.h: Include tree-hash-traits.h.
19096 (vec_base_alignments): New typedef.
19097 (vec_info): Add a base_alignments field.
19098 (vect_record_base_alignments): Declare.
19099 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
19101 (DR_IS_CONDITIONAL_IN_STMT): New macro.
19102 (create_data_ref): Add an is_conditional_in_stmt argument.
19103 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
19104 the is_conditional_in_stmt field.
19105 (data_ref_loc): Add an is_conditional_in_stmt field.
19106 (get_references_in_stmt): Set the is_conditional_in_stmt field.
19107 (find_data_references_in_stmt): Update call to create_data_ref.
19108 (graphite_find_data_references_in_stmt): Likewise.
19109 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
19110 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
19111 (vect_record_base_alignment): New function.
19112 (vect_record_base_alignments): Likewise.
19113 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
19114 for nested statements even if we fail to compute a misalignment.
19115 Use pooled base alignments for unconditional references.
19116 (vect_find_same_alignment_drs): Compare base addresses instead
19118 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
19119 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
19121 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19123 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
19124 Add an explicit name for the enum. Use auto_vec for slp_instances
19125 and grouped_stores.
19126 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
19128 (_bb_vec_info): Add a constructor and destructor.
19129 (vinfo_for_stmt): Return NULL for uids of -1 as well.
19130 (destroy_loop_vec_info): Delete.
19131 (vect_destroy_datarefs): Likewise.
19132 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
19133 (vec_info::vec_info): New function.
19134 (vec_info::~vec_info): Likewise.
19135 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
19136 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
19137 destroy_loop_vec_info.
19138 * tree-vect-loop.c (new_loop_vec_info): Replace with...
19139 (_loop_vec_info::_loop_vec_info): ...this.
19140 (destroy_loop_vec_info): Replace with...
19141 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
19142 the stmt_vec_infos. Leave handling of vec_info information to its
19143 destructor. Remove explicit vector releases.
19144 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
19145 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
19146 * tree-vect-slp.c (new_bb_vec_info): Replace with...
19147 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
19148 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
19149 (destroy_bb_vec_info): Replace with...
19150 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
19151 information to its destructor.
19152 (vect_slp_analyze_bb_1): Use new and delete instead of
19153 new_bb_vec_info and destroy_bb_vec_info.
19154 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
19157 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19159 * tree-data-ref.h (subscript): Add access_fn field.
19160 (data_dependence_relation): Add could_be_independent_p.
19161 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
19162 (same_access_functions): Move to tree-data-ref.c.
19163 * tree-data-ref.c (ref_contains_union_access_p): New function.
19164 (access_fn_component_p): Likewise.
19165 (access_fn_components_comparable_p): Likewise.
19166 (dr_analyze_indices): Add a reference to access_fn_component_p.
19167 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
19169 (constant_access_functions): Likewise.
19170 (add_other_self_distances): Likewise.
19171 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
19172 (initialize_data_dependence_relation): Use XCNEW and remove
19173 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
19174 of access functions that have the same type. Allow the
19175 subsequence to end with different bases in some circumstances.
19176 Record the chosen access functions in SUB_ACCESS_FN.
19177 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
19178 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
19179 (subscript_dependence_tester_1): Likewise dra and drb.
19180 (build_classic_dist_vector): Update calls accordingly.
19181 (subscript_dependence_tester): Likewise.
19182 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
19183 DDR_COULD_BE_INDEPENDENT_P.
19184 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
19185 comp_alias_ddrs instead of may_alias_ddrs.
19186 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
19188 (vect_analyze_data_ref_dependence): Use it if
19189 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
19190 distance vectors if that fails.
19191 (dependence_distance_ge_vf): New function.
19192 (vect_prune_runtime_alias_test_list): Use it. Don't clear
19193 LOOP_VINFO_MAY_ALIAS_DDRS.
19195 2017-08-04 Richard Biener <rguenther@suse.de>
19197 PR middle-end/81705
19198 * fold-const.c (fold_binary_loc): Properly restrict
19199 minus_var0 && minus_var1 case when associating undefined overflow
19202 2017-08-04 Tom de Vries <tom@codesourcery.com>
19204 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
19206 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19208 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19209 Don't start diagnostic messages with a capital letter.
19210 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19212 (rs6000_invalid_builtin): Likewise.
19213 (rs6000_trampoline_init): Likewise.
19215 2017-08-03 Jakub Jelinek <jakub@redhat.com>
19218 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
19219 after setting changeable df flags.
19221 2017-08-03 Richard Biener <rguenther@suse.de>
19223 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
19224 up if the use is in USE - X.
19226 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
19228 * toplev.c (dumpfile.h): New include.
19229 (internal_error_reentered): New static function. Use it...
19230 (internal_error_function): ...here to handle reentered internal_error.
19232 2017-08-03 Richard Biener <rguenther@suse.de>
19234 PR middle-end/81148
19235 * fold-const.c (split_tree): Add minus_var and minus_con
19236 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
19237 here but always use minus_*.
19238 (associate_trees): Assert we never associate with MINUS_EXPR
19239 and NULL first operand. Do not recurse for PLUS_EXPR operands
19240 when associating as MINUS_EXPR either.
19241 (fold_binary_loc): Track minus_var and minus_con.
19243 2017-08-03 Tom de Vries <tom@codesourcery.com>
19246 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
19247 ACCEL_COMPILER, apply finish_options on
19248 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
19250 2017-08-03 Tom de Vries <tom@codesourcery.com>
19253 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
19254 function_entry_patch_area_size > 0.
19256 2017-08-03 Jakub Jelinek <jakub@redhat.com>
19259 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
19260 instead of 10??LU, perform unit multiplication in wide_int,
19261 don't change alloc_object_size_limit if the limit is larger
19264 PR tree-optimization/81655
19265 PR tree-optimization/81588
19266 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
19267 the case when ranges[i].low and high are 1 for unsigned type with
19270 PR middle-end/81052
19271 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
19272 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
19274 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19276 * tree-vrp.h: Add include guard.
19278 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
19281 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
19282 (ud2): New insn pattern.
19283 * config/i386/i386.c (ix86_expand_epilogue):
19284 For naked functions, generate ud2 instead of trap insn.
19286 2017-08-02 Marek Polacek <polacek@redhat.com>
19289 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
19291 2017-08-02 Tom de Vries <tom@codesourcery.com>
19292 Cesar Philippidis <cesar@codesourcery.com>
19294 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
19295 Add missing edge probabilities.
19297 2017-08-02 Tamar Christina <tamar.christina@arm.com>
19299 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
19300 Correct endianness.
19302 2017-08-02 Jakub Jelinek <jakub@redhat.com>
19304 PR middle-end/79499
19305 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
19306 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
19307 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
19308 prologue_seq sequences - if any.
19310 2017-08-02 Richard Biener <rguenther@suse.de>
19312 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
19313 via vectors if supported, integer extracts via punning if supported
19314 or otherwise vector extracts.
19316 2017-08-02 Richard Biener <rguenther@suse.de>
19318 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
19320 (bitmap_insert_into_set): ... this.
19322 2017-08-02 Richard Biener <rguenther@suse.de>
19324 PR tree-optimization/81633
19326 2015-08-17 Alan Hayward <alan.hayward@arm.com>
19328 PR tree-optimization/71752
19329 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
19331 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
19333 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
19334 (machine_function::call_ms2sysv_pad_out): Remove field.
19335 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
19336 (ix86_compute_frame_layout): Likewise.
19338 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
19341 * config/i386/i386.c (ix86_set_func_type): Disallow naked
19342 attribute with interrupt attribute.
19344 2017-08-01 Andrew Pinski <apinski@cavium.com>
19346 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
19347 BIT_INSERT_EXPR's operand 1
19348 to see if the types precision matches.
19350 2017-08-01 Martin Liska <mliska@suse.cz>
19352 PR middle-end/70140
19353 * builtins.c (expand_builtin_memcpy_args): Remove.
19354 (expand_builtin_memcpy): Call newly added function
19355 expand_builtin_memory_copy_args.
19356 (expand_builtin_memcpy_with_bounds): Likewise.
19357 (expand_builtin_mempcpy): Remove last argument.
19358 (expand_builtin_mempcpy_with_bounds): Likewise.
19359 (expand_builtin_memory_copy_args): New function created from
19360 expand_builtin_mempcpy_args with small modifications.
19361 (expand_builtin_mempcpy_args): Remove.
19362 (expand_builtin_stpcpy): Remove unused argument.
19363 (expand_builtin): Likewise.
19364 (expand_builtin_with_bounds): Likewise.
19366 2017-08-01 Martin Liska <mliska@suse.cz>
19369 Make mempcpy more optimal (PR middle-end/70140).
19371 2017-08-01 Jakub Jelinek <jakub@redhat.com>
19374 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
19375 __builtin_vec_cmpne verify both arguments are compatible vectors
19376 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
19377 verify arg1_type is a pointer or array type. For __builtin_vec_st,
19378 move computation of aligned to after checking the argument types.
19382 * config/rs6000/vsx.md (vextract_fp_from_shorth,
19383 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
19386 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
19387 Doug Rupp <rupp@adacore.com>
19388 Olivier Hainque <hainque@adacore.com>
19390 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
19391 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
19393 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
19394 for TARGET_OS_CPP_BUILTIN.
19395 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
19396 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
19398 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
19399 passing required abi options to the assembler for EABI configurations.
19400 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
19401 of .text.hot and .text.unlikely sections for kernel modules when
19402 using ARM style exceptions.
19403 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
19404 options. Add EXTRA_CC1_SPEC.
19405 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
19407 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
19409 (ARM_TARGET2_DWARF_FORMAT): Define.
19410 * config/arm/t-vxworks: Adjust multilib control to removal of the
19411 Diab command line options.
19413 2017-08-01 Martin Liska <mliska@suse.cz>
19415 PR gcov-profile/81561
19416 * gcov.c (unblock): Make unblocking safe as we need to preserve
19417 index correspondence of blocks and block_lists.
19419 2017-08-01 Richard Biener <rguenther@suse.de>
19421 PR tree-optimization/81181
19422 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
19423 (compute_antic): ... end of iteration here.
19425 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
19427 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
19428 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
19429 (ftree-slp-vectorize): Likewise.
19430 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
19431 can no longer be set independent of flag_tree_loop_vectorize.
19432 * omp-general.c (emp_max_vf): Likewise.
19433 * opts.c (enable_fdo_optimizations): Remove references to
19434 flag_tree_vectorize, these are now implicit.
19435 (common_handle_option): Remove handling for OPT_ftree_vectorize,
19436 and leave it for the options machinery.
19438 2017-08-01 Martin Liska <mliska@suse.cz>
19440 PR middle-end/70140
19441 * builtins.c (expand_builtin_memcpy_args): Remove.
19442 (expand_builtin_memcpy): Call newly added function
19443 expand_builtin_memory_copy_args.
19444 (expand_builtin_memcpy_with_bounds): Likewise.
19445 (expand_builtin_mempcpy): Remove last argument.
19446 (expand_builtin_mempcpy_with_bounds): Likewise.
19447 (expand_builtin_memory_copy_args): New function created from
19448 expand_builtin_mempcpy_args with small modifications.
19449 (expand_builtin_mempcpy_args): Remove.
19450 (expand_builtin_stpcpy): Remove unused argument.
19451 (expand_builtin): Likewise.
19452 (expand_builtin_with_bounds): Likewise.
19454 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
19457 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
19458 print "ds:" only for immediates in generic address space.
19460 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
19463 * config/i386/i386.c (ix86_funciton_naked): New prototype.
19464 (ix86_function_ok_for_sibcall): Return false for naked functions.
19466 2017-08-01 Richard Biener <rguenther@suse.de>
19468 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
19469 (compute_antic): Seed worklist with exit block predecessors.
19470 * cfganal.c (dfs_find_deadend): For a cycle return the source
19471 of the edge closing it.
19473 2017-08-01 Tamar Christina <tamar.christina@arm.com>
19475 * config/aarch64/aarch64.c
19476 (aarch64_can_const_movi_rtx_p): Move 0 check.
19478 2017-08-01 Bin Cheng <bin.cheng@arm.com>
19480 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
19481 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
19483 * match.pd: Ditto in address comparison pattern.
19485 2017-08-01 Bin Cheng <bin.cheng@arm.com>
19487 PR tree-optimization/81627
19488 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
19489 closed ssa form for store-store chain.
19491 2017-08-01 Bin Cheng <bin.cheng@arm.com>
19493 PR tree-optimization/81620
19494 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
19495 for store-store chain.
19497 2017-08-01 Jakub Jelinek <jakub@redhat.com>
19499 PR tree-optimization/81588
19500 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
19501 ranges[i].in_p, invert comparison code ccode. For >/>=,
19502 swap rhs1 and rhs2 and comparison code unconditionally,
19503 for </<= don't do that. Don't swap rhs1/rhs2 again if
19504 ranges[i].in_p, instead invert comparison code ccode if
19505 opcode or oe->rank is BIT_IOR_EXPR.
19508 * optabs.def (vec_extract_optab, vec_init_optab): Change from
19509 a direct optab to conversion optab.
19510 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
19511 with GET_MODE_INNER as last argument instead of optab_handler.
19512 * expmed.c (extract_bit_field_1): Likewise. Use vector from
19513 vector extraction if possible and optab is available.
19514 * expr.c (store_constructor): Use convert_optab_handler instead
19515 of optab_handler. Use vector initialization from smaller
19516 vectors if possible and optab is available.
19517 * tree-vect-stmts.c (vectorizable_load): Likewise.
19518 * doc/md.texi (vec_extract, vec_init): Document that the optabs
19519 now have two modes.
19520 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
19521 of vec_init from half-sized vectors with the same element mode.
19522 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
19523 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
19524 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
19525 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
19526 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
19527 after mode in gen_vec_extract* calls.
19528 (vec_extract<mode>): Renamed to ...
19529 (vec_extract<mode><ssescalarmodelower>): ... this.
19530 (vec_extract<mode><ssehalfvecmodelower>): New expander.
19531 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
19532 element mode after mode in gen_vec_init* calls.
19533 (VEC_INIT_HALF_MODE): New mode iterator.
19534 (vec_init<mode>): Renamed to ...
19535 (vec_init<mode><ssescalarmodelower>): ... this.
19536 (vec_init<mode><ssehalfvecmodelower>): New expander.
19537 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
19538 (vec_extractv2sfsf): ... this.
19539 (vec_initv2sf): Renamed to ...
19540 (vec_initv2sfsf): ... this.
19541 (vec_extractv2si): Renamed to ...
19542 (vec_extractv2sisi): ... this.
19543 (vec_initv2si): Renamed to ...
19544 (vec_initv2sisi): ... this.
19545 (vec_extractv4hi): Renamed to ...
19546 (vec_extractv4hihi): ... this.
19547 (vec_initv4hi): Renamed to ...
19548 (vec_initv4hihi): ... this.
19549 (vec_extractv8qi): Renamed to ...
19550 (vec_extractv8qiqi): ... this.
19551 (vec_initv8qi): Renamed to ...
19552 (vec_initv8qiqi): ... this.
19553 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
19554 (vec_init<mode>): Renamed to ...
19555 (vec_init<mode><VEC_base_l>): ... this.
19556 (vec_extract<mode>): Renamed to ...
19557 (vec_extract<mode><VEC_base_l>): ... this.
19558 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
19559 (vec_initv2sfsf): ... this.
19560 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
19561 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
19562 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
19563 element mode after mode in gen_vec_init* calls.
19564 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
19565 (vec_init<mode><Vel>): ... this.
19566 (vec_extract<mode>): Renamed to ...
19567 (vec_extract<mode><Vel>): ... this.
19568 * config/aarch64/iterators.md (Vel): New mode attribute.
19569 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
19570 Add element mode after mode in gen_vec_extract* calls.
19571 * config/s390/vector.md (non_vec_l): New mode attribute.
19572 (vec_extract<mode>): Renamed to ...
19573 (vec_extract<mode><non_vec_l>): ... this.
19574 (vec_init<mode>): Renamed to ...
19575 (vec_init<mode><non_vec_l>): ... this.
19576 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
19577 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
19579 * config/arm/iterators.md (V_elem_l): New mode attribute.
19580 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
19581 (vec_extract<mode><V_elem_l>): ... this.
19582 (vec_extractv2di): Renamed to ...
19583 (vec_extractv2didi): ... this.
19584 (vec_init<mode>): Renamed to ...
19585 (vec_init<mode><V_elem_l>): ... this.
19586 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
19587 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
19588 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
19589 Add element mode after gen_vec_extract* calls.
19590 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
19591 (vec_init<mode><unitmode>): ... this.
19592 (vec_extract<mode>): Renamed to ...
19593 (vec_extract<mode><unitmode>): ... this.
19594 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
19595 (vec_init<mode><unitmode>): ... this.
19596 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
19597 (vec_initv2sfsf): ... this.
19598 (vec_extractv2sf): Renamed to ...
19599 (vec_extractv2sfsf): ... this.
19600 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
19601 Add element mode after gen_vec_extract* calls.
19602 * config/mips/mips.md (unitmode): New mode iterator.
19603 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
19604 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
19605 * config/spu/spu.md (inner_l): New mode attribute.
19606 (vec_init<mode>): Renamed to ...
19607 (vec_init<mode><inner_l>): ... this.
19608 (vec_extract<mode>): Renamed to ...
19609 (vec_extract<mode><inner_l>): ... this.
19610 * config/sparc/sparc.md (veltmode): New mode iterator.
19611 (vec_init<VMALL:mode>): Renamed to ...
19612 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
19613 * config/ia64/vect.md (vec_initv2si): Renamed to ...
19614 (vec_initv2sisi): ... this.
19615 (vec_initv2sf): Renamed to ...
19616 (vec_initv2sfsf): ... this.
19617 (vec_extractv2sf): Renamed to ...
19618 (vec_extractv2sfsf): ... this.
19619 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
19620 (vec_init<mode>): Renamed to ...
19621 (vec_init<mode><VEC_base_l>): ... this.
19622 (vec_extract<mode>): Renamed to ...
19623 (vec_extract<mode><VEC_base_l>): ... this.
19624 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
19625 (vec_initv2sfsf): ... this.
19626 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
19627 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
19628 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
19629 gen_vec_init* calls.
19631 2017-08-01 Richard Biener <rguenther@suse.de>
19633 PR tree-optimization/81297
19634 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
19635 TREE_OVERFLOW from INTEGER_CSTs.
19637 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
19639 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
19641 2017-07-31 Carl Love <cel@us.ibm.com>
19643 * config/rs6000/rs6000-c: Add support for built-in functions
19644 vector signed char vec_xl_be (signed long long, signed char *);
19645 vector unsigned char vec_xl_be (signed long long, unsigned char *);
19646 vector signed int vec_xl_be (signed long long, signed int *);
19647 vector unsigned int vec_xl_be (signed long long, unsigned int *);
19648 vector signed long long vec_xl_be (signed long long, signed long long *);
19649 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
19650 vector signed short vec_xl_be (signed long long, signed short *);
19651 vector unsigned short vec_xl_be (signed long long, unsigned short *);
19652 vector double vec_xl_be (signed long long, double *);
19653 vector float vec_xl_be (signed long long, float *);
19654 * config/rs6000/altivec.h (vec_xl_be): Add #define.
19655 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
19656 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
19658 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
19659 (altivec_expand_builtin): Add switch statement to call
19660 altivec_expand_xl_be for each builtin.
19661 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
19662 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
19663 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
19664 __builtin_vsx_le_be_v16qi.
19665 * doc/extend.texi: Update the built-in documentation file for the
19666 new built-in functions.
19668 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19671 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
19673 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
19675 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19677 * config.gcc: Add z14.
19678 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
19679 CPU model numbers for z13s and z14.
19680 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
19682 * config/s390/s390-opts.h (enum processor_type): Rename
19683 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19684 * config/s390/s390.c (processor_table): Add field for CPU name to
19685 be passed to Binutils.
19686 (s390_asm_output_machine_for_arch): Use the new field in
19687 processor_table for Binutils.
19688 (s390_expand_builtin): Replace arch12 with z14.
19689 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19690 (s390_get_sched_attrmask): Likewise.
19691 (s390_get_unit_mask): Likewise.
19692 * config/s390/s390.opt: Add z14 to processor_type enum.
19694 2017-07-31 Martin Jambor <mjambor@suse.cz>
19697 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
19698 regardless of optimization level.
19700 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
19701 Martin Liska <mliska@suse.cz>
19703 * predict.def: Remove old comment and adjust probability.
19704 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
19705 PREDICT statements.
19707 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19710 * config/i386/i386.c (ix86_function_naked): New function.
19711 (ix86_can_use_return_insn_p): Return false for naked functions.
19712 (ix86_expand_prologue): Skip prologue for naked functions.
19713 (ix86_expand_epilogue): Skip epilogue for naked functions
19714 and emit trap instruction.
19715 (ix86_warn_func_return): New function.
19716 (ix86_attribute_table): Add "naked" attribute specification.
19717 (TARGET_WARN_FUNC_RETURN): Define.
19718 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
19720 2017-07-31 Martin Liska <mliska@suse.cz>
19722 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
19723 (dump_gimple_bb_header): Always dump BB info.
19724 (pp_cfg_jump): Do not append info about BB when dumping a jump.
19726 2017-07-31 Martin Liska <mliska@suse.cz>
19729 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
19730 also with current_function_decl non-null equality.
19732 2017-07-31 Jakub Jelinek <jakub@redhat.com>
19735 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
19736 change type to the element type, instead add eltype variable and
19737 use it where we are interested in the element type.
19739 PR tree-optimization/81603
19740 * ipa-polymorphic-call.c
19741 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
19742 offset arithmetic in offset_int, bail out if the resulting bit offset
19743 doesn't fit into shwi.
19745 2017-07-31 Martin Liska <mliska@suse.cz>
19747 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
19748 (gimplify_save_expr): Fix comment.
19750 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
19753 * config/i386/i386.c (ix86_function_arg): Update arguments for
19755 (ix86_compute_frame_layout): Set the initial stack offset to
19756 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
19757 INCOMING_FRAME_SP_OFFSET.
19758 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
19759 stack before exception handler returns.
19760 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
19761 the 'ERROR_CODE' for exception handler.
19763 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
19765 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
19766 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
19767 (ASM_OUTPUT_REG_POP): Ditto.
19768 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
19769 instead of asm_fprintf to output pure string.
19771 2017-07-29 Jakub Jelinek <jakub@redhat.com>
19773 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
19774 to imported_module_or_decl hook.
19775 (debug_nothing_tree_tree_tree_bool): Remove.
19776 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
19777 * debug.c (do_nothing_debug_hooks): Use
19778 debug_nothing_tree_tree_tree_bool_bool instead of
19779 debug_nothing_tree_tree_tree_bool.
19780 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19781 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
19782 * sdbout.c (sdb_debug_hooks): Likewise.
19783 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
19784 (gen_namespace_die): Add DW_AT_export_symbols attribute if
19786 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
19787 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
19788 attribute, don't add anything.
19790 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19792 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
19793 (fold_build2_stat_loc): Likewise.
19794 (fold_build3_stat_loc): Likewise.
19795 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
19796 (fold_build1_loc): Remove macro.
19797 (fold_build2_loc): Likewise.
19798 (fold_build3_loc): Likewise.
19800 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19802 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
19803 (gimple_build_debug_bind_source_stat): Likewise.
19804 * gimple.h (gimple_build_debug_bind): Remove macro.
19805 (gimple_build_debug_bind_source): Likewise.
19807 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19809 * bitmap.c (bitmap_alloc): Adjust.
19810 (bitmap_gc_alloc): Likewise.
19811 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
19813 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19815 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
19816 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
19817 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
19818 (bitmap_gc_alloc_stat): Likewise.
19819 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
19821 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19823 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
19824 * rtl.h (shallow_copy_rtx): Remove macro.
19826 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19828 * emit-rtl.c (gen_raw_REG): Adjust.
19829 * gengenrtl.c (gendef): Likewise.
19830 * rtl.c (rtx_alloc_stat): Remove _stat from name.
19831 * rtl.h (rtx_alloc): Remove macro.
19833 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19835 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
19836 (build_tree_list_stat): Likewise.
19837 * tree.h (build_tree_list): Remove macro.
19838 (build_tree_list_vec): Likewise.
19840 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19842 * tree.c (make_vector_stat): Remove _stat from name.
19843 (build_vector_stat): Likewise.
19844 * tree.h (make_vector_stat): Remove macro.
19845 (build_vector_stat): Likewise.
19847 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19849 * tree.h (build_var_debug_value): Remove prototype.
19851 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19853 * tree.c (tree_cons_stat): Remove _stat from name.
19854 * tree.h (tree_cons): Remove macro.
19856 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19858 * tree.c (build_vl_exp_stat): Remove _stat from name.
19859 * tree.h (build_vl_exp): Remove macro.
19861 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19863 * tree.c (build_decl_stat): Remove _stat from name.
19864 * tree.h (build_decl): Remove macro.
19866 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19868 * gimple.c (gimple_build_with_ops_stat): Adjust.
19869 (gimple_alloc_stat): Remove _stat from name.
19870 * gimple.h (gimple_alloc): Remove macro.
19872 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19874 * tree.c (make_tree_vec_stat): Remove _stat from name.
19875 (grow_tree_vec_stat): Likewise.
19876 * tree.h (make_tree_vec_stat): Adjust prototype.
19877 (grow_tree_vec_stat): Likewise.
19878 (make_tree_vec): Remove macro.
19879 (grow_tree_vec): Likewise.
19881 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19883 * fold-const.c (fold_build1_stat_loc): Adjust.
19884 (fold_build2_stat_loc): Likewise.
19885 (fold_build3_stat_loc): Likewise.
19886 * tree.c (build0_stat): Remove _stat from name.
19887 (build1_stat): Likewise.
19888 (build2_stat): Likewise.
19889 (build3_stat): Likewise.
19890 (build4_stat): Likewise.
19891 (build5_stat): Likewise.
19892 * tree.h (build1_loc): Remove macro, and rename _stat function
19894 (build2_loc): Likewise.
19895 (build3_loc): Likewise.
19896 (build4_loc): Likewise.
19897 (build5_loc): Likewise.
19899 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19901 * tree.c (make_int_cst_stat): Remove _stat from name.
19902 * tree.h (make_int_cst_stat): Adjust prototype.
19903 (make_int_cst): Remove macro.
19905 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19907 * tree.c (make_tre_binfo_stat): Remove _stat from name.
19908 * tree.h (make_tree_binfo_stat): Adjust prototype.
19909 (make_tree_binfo): Remove.
19911 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19913 * tree.c (copy_node_stat): Rename to copy_node.
19914 (build_distinct_type_copy): Adjust.
19915 * tree.h (copy_node_stat): Adjust prototype.
19916 (copy_node): Remove macro.
19918 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19920 * tree.c (make_node_stat): rename to make_node.
19921 (build_tree_list_stat): Adjust.
19922 (build0_stat): Likewise.
19923 (build2_stat): Likewise.
19924 (build3_stat): Likewise.
19925 (build4_stat): Likewise.
19926 (build5_stat): Likewise.
19927 (build_decl_stat): Likewise.
19928 * tree.h (make_node_stat): Adjust prototype.
19929 (make_node): remove macro.
19931 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
19933 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
19934 (PPC_FEATURE2_SCV): Likewise.
19935 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
19937 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19939 * config/aarch64/aarch64.c
19940 (aarch64_internal_mov_immediate): Add new special pattern.
19941 * config/aarch64/aarch64.md (*movdi_aarch64):
19942 Add reg/32bit const mov case.
19944 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19945 Richard Sandiford <richard.sandiford@linaro.org>
19947 * config/aarch64/aarch64.md (mov<mode>): Generalize.
19948 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
19949 Add integer and movi cases.
19950 (movi-split-hf-df-sf split, fp16): New.
19951 (enabled): Added TARGET_FP_F16INST.
19952 * config/aarch64/iterators.md (GPF_HF): New.
19953 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
19955 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19957 * config/aarch64/aarch64.c
19958 (aarch64_simd_container_mode): Add prototype.
19959 (aarch64_expand_mov_immediate): Add HI support.
19960 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
19961 (aarch64_can_const_movi_rtx_p): New.
19962 (aarch64_preferred_reload_class):
19963 Remove restrictions of using FP registers for certain SIMD operations.
19964 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
19965 (aarch64_valid_floating_const): Add integer move validation.
19966 (aarch64_simd_imm_scalar_p): Remove.
19967 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
19968 (aarch64_legitimate_constant_p): Expand list of supported cases.
19969 * config/aarch64/aarch64-protos.h
19970 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
19971 (aarch64_reinterpret_float_as_int): New.
19972 (aarch64_simd_imm_scalar_p): Remove.
19973 * config/aarch64/constraints.md (Uvi): New.
19974 (Dd): Split into Ds and new Dd.
19975 * config/aarch64/aarch64.md (*movsi_aarch64):
19977 (*movdi_aarch64): Add SIMD mov case.
19979 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19981 * tree-predcom.c: (struct chain): Handle store-store chain in which
19982 stores for elimination only store loop invariant values.
19983 (execute_pred_commoning_chain): Ditto.
19984 (prepare_initializers_chain_store_elim): Ditto.
19985 (prepare_finalizers): Ditto.
19986 (is_inv_store_elimination_chain): New function.
19987 (initialize_root_vars_store_elim_1): New function.
19989 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19991 * tree-predcom.c: Revise general description of the pass.
19992 (enum chain_type): New enum type for store elimination.
19993 (struct chain): New field supporting store elimination.
19994 (struct component): Ditto.
19995 (dump_chain): Dump store-stores chain.
19996 (release_chain): Release resources.
19997 (split_data_refs_to_components): Compute and create component
19998 contains only stores for elimination.
19999 (get_chain_last_ref_at): New function.
20000 (make_invariant_chain): Initialization.
20001 (make_rooted_chain): Specify chain type in parameter and record it.
20002 (add_looparound_copies): Skip for store-stores chain.
20003 (determine_roots_comp): Compute type of chain and pass it to
20005 (initialize_root_vars_store_elim_2): New function.
20006 (finalize_eliminated_stores): New function.
20007 (remove_stmt): Handle store for elimination.
20008 (execute_pred_commoning_chain): Execute predictive commoning on
20009 store-store chains.
20010 (determine_unroll_factor): Skip unroll for store-stores chain.
20011 (prepare_initializers_chain_store_elim): New function.
20012 (prepare_initializers_chain): Hanlde store-store chain.
20013 (prepare_finalizers_chain, prepare_finalizers): New function.
20014 (tree_predictive_commoning_loop): Return integer value indicating
20015 if loop is unrolled or lcssa form is corrupted.
20016 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
20018 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20020 * tree-predcom.c (initialize_root): Delete.
20021 (execute_pred_commoning_chain): Initialize root vars and replace
20022 reference of non-combined chain directly, rather than call above
20025 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20027 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
20028 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
20030 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20032 * tree-predcom.c (struct chain): New field init_seq.
20033 (release_chain): Release init_seq.
20034 (prepare_initializers_chain): Record intialization stmts in above
20036 (insert_init_seqs): New function.
20037 (tree_predictive_commoning_loop): Call insert_init_seqs.
20039 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20041 * tree-predcom.c (determine_roots_comp): Skip trivial components.
20043 2017-07-28 Richard Biener <rguenther@suse.de>
20045 * match.pd: Remove superfluous :c.
20046 * genmatch.c (simplify::id): Add member.
20047 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
20049 (current_id): New global.
20050 (dt_node::parent): Move from ...
20051 (dt_operand::parent): ... here. Add for_id member.
20052 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
20053 (decision_tree::find_node): Relax order requirement when
20054 merging DT_TRUE nodes to ones inbetween the current simplify
20055 and the one we try to merge with. Add diagnostic whenever
20056 we need to enforce pattern order by not merging.
20057 (decision_tree::insert): Set current_id.
20058 (decision_tree::print_node): Dump parent node and for_id.
20059 (parser::last_id): Add member.
20060 (parser::push_simplify): Assign unique id.
20061 (parser::parser): Initialize last_id.
20063 2017-07-28 Martin Liska <mliska@suse.cz>
20066 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
20067 gimple_build_debug_bind.
20069 2017-07-28 Richard Biener <rguenther@suse.de>
20071 PR tree-optimization/81502
20072 * match.pd: Add pattern combining BIT_INSERT_EXPR with
20074 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
20076 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
20077 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
20078 for BIT_FIELD_REF args.
20079 * fold-const.c (make_bit_field_ref): Likewise.
20080 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
20082 2017-07-28 Jakub Jelinek <jakub@redhat.com>
20085 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
20086 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
20087 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
20088 Or it into SANITIZER_UNDEFINED.
20089 * ubsan.c: Include gimple-fold.h and varasm.h.
20090 (ubsan_expand_ptr_ifn): New function.
20091 (instrument_pointer_overflow): New function.
20092 (maybe_instrument_pointer_overflow): New function.
20093 (instrument_object_size): Formatting fix.
20094 (pass_ubsan::execute): Call instrument_pointer_overflow
20095 and maybe_instrument_pointer_overflow.
20096 * internal-fn.c (expand_UBSAN_PTR): New function.
20097 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
20098 * sanitizer.def (__ubsan_handle_pointer_overflow,
20099 __ubsan_handle_pointer_overflow_abort): New builtins.
20100 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
20101 * internal-fn.def (UBSAN_PTR): New internal function.
20102 * opts.c (sanitizer_opts): Add pointer-overflow.
20103 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
20104 * fold-const.c (build_range_check): Compute pointer range check in
20105 integral type if pointer arithmetics would be needed. Formatting
20108 2017-07-28 Martin Liska <mliska@suse.cz>
20111 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
20112 parameters that are of a variable-length.
20114 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
20116 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
20118 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
20119 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
20120 (CRT_CALL_STATIC_FUNCTION): Likewise.
20121 (ASM_DEFAULT_SPEC): New define.
20122 (ASM_SPEC32): Likewise.
20123 (ASM_SPEC64): Likewise.
20124 (ASM_SPEC_COMMON): Likewise.
20125 (ASM_SPEC): Likewise.
20126 (INVALID_64BIT): Likewise.
20127 (LINK_OS_DEFAULT_SPEC): Likewise.
20128 (LINK_OS_SPEC32): Likewise.
20129 (LINK_OS_SPEC64): Likewise.
20130 (POWERPC_LINUX): Likewise.
20131 (PTRDIFF_TYPE): Likewise.
20132 (RESTORE_FP_PREFIX): Likewise.
20133 (RESTORE_FP_SUFFIX): Likewise.
20134 (SAVE_FP_PREFIX): Likewise.
20135 (SAVE_FP_SUFFIX): Likewise.
20136 (SIZE_TYPE): Likewise.
20137 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
20138 (TARGET_64BIT): Likewise.
20139 (TARGET_64BIT): Likewise.
20140 (TARGET_AIX): Likewise.
20141 (WCHAR_TYPE_SIZE): Likewise.
20142 (WCHAR_TYPE): Undefine.
20143 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
20144 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
20145 (CPP_OS_RTEMS_SPEC): Delete.
20146 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
20147 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
20149 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
20151 2017-07-28 Jakub Jelinek <jakub@redhat.com>
20153 PR tree-optimization/81578
20154 * tree-parloops.c (build_new_reduction): Bail out if
20155 reduction_code isn't one of the standard OpenMP reductions.
20156 Move the details printing after that decision.
20158 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20160 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
20161 related to reload_in_progress.
20162 (splat_input_operand): Likewise.
20163 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
20165 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
20167 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
20168 (TARGET_INSTANTIATE_DECLS): Likewise.
20169 (legitimate_indexed_address_p): Delete reload_in_progress code.
20170 (rs6000_debug_legitimate_address_p): Likewise.
20171 (rs6000_eliminate_indexed_memrefs): Likewise.
20172 (rs6000_emit_le_vsx_store): Likewise.
20173 (rs6000_emit_move_si_sf_subreg): Likewise.
20174 (rs6000_emit_move): Likewise.
20175 (register_to_reg_type): Likewise.
20176 (rs6000_pre_atomic_barrier): Likewise.
20177 (rs6000_machopic_legitimize_pic_address): Likewise.
20178 (rs6000_allocate_stack_temp): Likewise.
20179 (rs6000_address_for_fpconvert): Likewise.
20180 (rs6000_address_for_altivec): Likewise.
20181 (rs6000_secondary_memory_needed_rtx): Delete function.
20182 (rs6000_check_sdmode): Likewise.
20183 (rs6000_alloc_sdmode_stack_slot): Likewise.
20184 (rs6000_instantiate_decls): Likewise.
20185 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
20186 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
20187 Delete reload_in_progress.
20188 (*vec_reload_and_plus_<mptrsize>): Likewise.
20189 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
20190 (vsx_div_v2di): Likewise.
20191 (vsx_udiv_v2di): Likewise.
20193 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20195 * config/rs6000/rs6000.opt (mlra): Replace with stub.
20196 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
20197 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
20198 (rs6000_debug_reg_global): Delete print of LRA status.
20199 (rs6000_option_override_internal): Delete dead LRA related code.
20200 (rs6000_lra_p): Delete function.
20201 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
20203 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
20205 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
20206 * config/riscv/rtems.h: New file.
20208 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20209 Sudakshina Das <sudi.das@arm.com>
20211 * config/aarch64/aarch64.md
20212 (define_split for and<mode>3nr_compare): Move
20213 non aarch64_logical_operand to a register.
20214 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
20215 register immediate operand to a register.
20216 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
20218 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20220 PR middle-end/81564
20221 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
20223 2017-07-27 Richard Biener <rguenther@suse.de>
20225 PR tree-optimization/81573
20226 PR tree-optimization/81494
20227 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
20228 multi defuse cycle case.
20230 2017-07-27 Richard Biener <rguenther@suse.de>
20232 PR tree-optimization/81571
20233 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
20236 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
20238 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
20239 earlier and only if MASK_FPU is set. Adjust formatting.
20241 2017-07-27 Martin Liska <mliska@suse.cz>
20243 * opt-functions.awk: Add validation of value of Init.
20244 * optc-gen.awk: Pass new argument.
20246 2017-07-27 Martin Liska <mliska@suse.cz>
20248 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
20249 Fix wrong condition.
20251 2017-07-27 Martin Liska <mliska@suse.cz>
20253 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
20254 BBs and edges seen by autoFDO.
20256 2017-07-27 Richard Biener <rguenther@suse.de>
20258 PR tree-optimization/81502
20259 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
20260 with incompatible but same sized type.
20261 (execute_update_addresses_taken): Likewise.
20263 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
20265 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
20266 flag_tree_loop_vectorize rather than flag_tree_vectorize.
20268 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20271 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
20272 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
20273 Change s_operand to memory_operand.
20275 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
20277 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
20278 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
20279 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
20280 Emit instructions rather than returning an expression. Handle TFmode
20281 and KFmode by casting to TImode.
20282 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
20283 (rs6000_emit_le_vsx_store): Likewise.
20284 * config/rs6000/vsx.md (VSX_TI): New iterator.
20285 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
20286 (*vsx_le_undo_permute_<mode>): Likewise.
20287 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
20288 emit the split sequence.
20289 (*vsx_le_perm_store_<mode>): Likewise.
20291 2017-07-27 Jakub Jelinek <jakub@redhat.com>
20293 PR tree-optimization/81555
20294 PR tree-optimization/81556
20295 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
20296 if true, force CHANGED for the recursive invocation.
20297 (reassociate_bb): Remember original length of ops array, pass
20298 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
20300 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
20301 attributes for noipa attribute. For naked attribute use
20302 lookup_attribute first before lookup_attribute_spec.
20303 * final.c (rest_of_handle_final): Disable IPA RA for functions with
20305 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
20306 for functions with noipa attribute.
20307 (cgraph_externally_visible_p): Return true for functions with noipa
20309 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
20310 for functions with noipa attribute.
20311 * doc/extend.texi: Document noipa function attribute.
20312 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
20313 also for functions with noipa attribute.
20314 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
20316 2017-07-26 Andrew Pinski <apinski@cavium.com>
20318 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
20319 vec_unalign_load_cost and vec_unalign_store_cost.
20321 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
20323 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
20324 -mvsx-small-integer option.
20325 (ISA_3_0_MASKS_IEEE): Likewise.
20326 (OTHER_VSX_VECTOR_MASKS): Likewise.
20327 (POWERPC_MASKS): Likewise.
20328 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
20329 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
20330 code, only testing for DImode being allowed in non-VSX floating
20332 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
20333 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
20335 (rs6000_option_override_internal): Delete -mvsx-small-integer.
20336 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
20337 TARGET_P8_VECTOR test.
20338 (rs6000_secondary_reload_simple_move): Likewise.
20339 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
20340 since TARGET_P9_VECTOR was already tested.
20341 (rs6000_opt_masks): Remove -mvsx-small-integer.
20342 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
20343 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
20345 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
20346 test for TARGET_VEXTRACTUB was used, and that uses
20348 (p9 extract splitter): Likewise.
20349 (vsx_extract_<mode>_di_p9): Likewise.
20350 (vsx_extract_<mode>_store_p9): Likewise.
20351 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
20352 for TARGET_P9_VECTOR was used. Delete code that is now dead with
20353 the elimination of TARGET_VSX_SMALL_INTEGER.
20354 (vsx_extract_<mode>_p8): Likewise.
20355 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
20356 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
20357 (vsx_set_<mode>_p9): Likewise.
20358 (vsx_set_v4sf_p9): Likewise.
20359 (vsx_set_v4sf_p9_zero): Likewise.
20360 (vsx_insert_extract_v4sf_p9): Likewise.
20361 (vsx_insert_extract_v4sf_p9_2): Likewise.
20362 * config/rs6000/rs6000.md (sign extend splitter): Change
20363 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
20364 (floatsi<mode>2_lfiwax_mem): Likewise.
20365 (floatunssi<mode>2_lfiwzx_mem): Likewise.
20366 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
20367 since a test for TARGET_P9_VECTOR was used.
20368 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
20369 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
20370 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
20371 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
20372 TARGET_P8_VECTOR test.
20373 (fix_trunc<mode>si2_stfiwx): Likewise.
20374 (fix_trunc<mode>si2_internal): Likewise.
20375 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
20376 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
20378 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
20379 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
20380 TARGET_P8_VECTOR test.
20381 (fixuns_trunc<mode>si2_stfiwx): Likewise.
20382 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
20383 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
20385 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
20386 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
20387 since a test for TARGET_P9_VECTOR was used.
20388 (splitter for loading small constants): Likewise.
20390 2017-07-26 Andrew Pinski <apinski@cavium.com>
20392 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
20395 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
20398 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
20399 (fp_valid_at): Likewise.
20401 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
20403 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
20404 (qdf24xx_addrcost_table): Likewise.
20405 (cortexa57_tunings): Update to use generic_branch_cost.
20406 (cortexa72_tunings): Likewise.
20407 (cortexa73_tunings): Likewise.
20408 (qdf24xx_tunings): Likewise.
20410 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
20412 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
20413 (thunderx2t99_branch_cost): Likewise.
20414 (cortexa35_tunings): Update to use generic_branch_cost.
20415 (cortexa53_tunings): Likewise.
20416 (cortexa57_tunings): Likewise.
20417 (cortexa72_tunings): Likewise.
20418 (cortexa73_tunings): Likewise.
20419 (thunderx2t99_tunings): Likewise.
20421 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
20423 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
20424 (sparc_option_override): Honour MASK_FSMULD.
20425 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
20426 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
20427 * config/sparc/sparc.opt (mfsmuld): New option.
20428 * doc/invoke.texi (mfsmuld): Document option.
20430 2017-07-26 Marek Polacek <polacek@redhat.com>
20432 PR middle-end/70992
20433 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
20435 2017-07-26 Richard Biener <rguenther@suse.de>
20437 * gimple-match-head.c (do_valueize): Return OP if valueize
20439 (get_def): New helper to get at the def stmt of a SSA name
20440 if valueize allows.
20441 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
20442 do_valueize to get at the def stmt.
20443 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
20445 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
20447 PR middle-end/46932
20448 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
20450 2017-07-26 Martin Liska <mliska@suse.cz>
20453 * function.c (expand_function_start): Make expansion of
20454 nonlocal_goto_save_area after parm_birth_insn.
20456 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
20458 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
20459 from all CPU target flags enable members.
20461 2017-07-26 Richard Biener <rguenther@suse.de>
20463 * genmatch.c (dt_simplify::gen): Make iterator vars const.
20464 (decision_tree::gen): Make 'type' const.
20465 (write_predicate): Likewise.
20467 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
20469 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
20470 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
20471 (rs6000_option_override_internal): Likewise.
20472 (rs6000_expand_vector_set): Likewise.
20473 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
20474 (TARGET_UPPER_REGS_SF): Likewise.
20475 (TARGET_UPPER_REGS_DI): Likewise.
20476 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
20477 (TARGET_DIRECT_MOVE_64BIT): Likewise.
20478 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
20479 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
20480 (Splitters for DI constants in Altivec registers): Likewise.
20481 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
20482 (vsx_set_v4sf_p9): Likewise.
20483 (vsx_set_v4sf_p9_zero): Likewise.
20484 (vsx_insert_extract_v4sf_p9): Likewise.
20485 (vsx_insert_extract_v4sf_p9_2): Likewise.
20487 2017-07-25 Carl Love <cel@us.ibm.com>
20489 * doc/extend.texi: Update the built-in documentation file for the
20490 existing built-in functions
20491 vector signed char vec_cnttz (vector signed char);
20492 vector unsigned char vec_cnttz (vector unsigned char);
20493 vector signed short vec_cnttz (vector signed short);
20494 vector unsigned short vec_cnttz (vector unsigned short);
20495 vector signed int vec_cnttz (vector signed int);
20496 vector unsigned int vec_cnttz (vector unsigned int);
20497 vector signed long long vec_cnttz (vector signed long long);
20498 vector unsigned long long vec_cnttz (vector unsigned long long);
20500 2017-07-25 Andrew Pinski <apinski@cavium.com>
20502 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
20503 accesses where the use is for the first operand of a BIT_INSERT.
20505 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
20508 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
20509 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
20511 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
20513 * config/i386/gstabs.h: Delete.
20514 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
20516 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
20518 * config/i386/i386.c (ix86_decompose_address): Do not check for
20519 register RTX when looking at index_reg or base_reg.
20520 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
20522 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
20524 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
20525 to update EH info here.
20527 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20529 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
20531 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20533 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
20535 2017-07-25 Torsten Duwe <duwe@suse.de>
20537 * common.opt: Introduce -fpatchable-function-entry
20538 command line option, and its variables function_entry_patch_area_size
20539 and function_entry_patch_area_start.
20540 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
20541 including a two-value parser.
20542 * target.def (print_patchable_function_entry): New target hook.
20543 * targhooks.h (default_print_patchable_function_entry): New function.
20544 * targhooks.c (default_print_patchable_function_entry): Likewise.
20545 * toplev.c (process_options): Switch off IPA-RA if
20546 patchable function entries are being generated.
20547 * varasm.c (assemble_start_function): Look at the
20548 patchable-function-entry command line switch and current
20549 function attributes and maybe generate NOP instructions by
20550 calling the print_patchable_function_entry hook.
20551 * doc/extend.texi: Document patchable_function_entry attribute.
20552 * doc/invoke.texi: Document -fpatchable_function_entry
20553 command line option.
20554 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
20556 * doc/tm.texi: Re-generate.
20558 2017-07-25 Jakub Jelinek <jakub@redhat.com>
20561 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
20562 TARGET_AVX512DQ rather than TARGET_AVX512BW.
20564 2017-07-25 Tamar Christina <tamar.christina@arm.com>
20566 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
20568 2017-07-25 Richard Biener <rguenther@suse.de>
20570 PR tree-optimization/81455
20571 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
20572 not walk in cycles when looking for guards.
20574 2017-07-25 Richard Biener <rguenther@suse.de>
20576 PR tree-optimization/81529
20577 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
20578 when optimizing backedge uses.
20580 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
20582 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
20584 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
20585 to dl_section_ref. On AIX, append an expression to subtract
20586 the size of the section length to dl_section_ref.
20588 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
20590 * configure.ac: If any of the config.* scripts fail, exit 1.
20591 * configure: Regenerate.
20593 2017-07-25 Richard Biener <rguenther@suse.de>
20595 PR middle-end/81546
20596 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
20597 of immediate uses, be more verbose on errors.
20599 2017-07-25 Richard Biener <rguenther@suse.de>
20601 PR tree-optimization/81510
20602 * tree-vect-loop.c (vect_is_simple_reduction): When the
20603 reduction stmt is not inside the loop bail out.
20605 2017-07-25 Richard Biener <rguenther@suse.de>
20607 PR tree-optimization/81303
20608 * tree-vect-loop-manip.c (vect_loop_versioning): Build
20609 profitability check against LOOP_VINFO_NITERSM1.
20611 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20613 * domwalk.c (cmp_bb_postorder): Simplify.
20614 (sort_bbs_postorder): New function. Use it...
20615 (dom_walker::walk): ...here to optimize common cases.
20617 2017-07-25 Martin Liska <mliska@suse.cz>
20620 * ipa-visibility.c (function_and_variable_visibility): Make the
20621 redirection just on target that supports aliasing.
20622 Fix GNU coding style.
20624 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
20627 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
20629 * config/arm/rtems.h: Likewise.
20630 * config/bfin/rtems.h: Likewise.
20631 * config/i386/rtemself.h: Likewise.
20632 * config/lm32/rtems.h: Likewise.
20633 * config/m32c/rtems.h: Likewise.
20634 * config/m68k/rtemself.h: Likewise.
20635 * config/microblaze/rtems.h: Likewise.
20636 * config/mips/rtems.h: Likewise.
20637 * config/moxie/rtems.h: Likewise.
20638 * config/nios2/rtems.h: Likewise.
20639 * config/powerpcspe/rtems.h: Likewise.
20640 * config/rs6000/rtems.h: Likewise.
20641 * config/rtems.h: Likewise.
20642 * config/sh/rtems.h: Likewise.
20643 * config/sh/rtemself.h: Likewise.
20644 * config/sparc/rtemself.h: Likewise.
20646 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20649 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
20650 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
20651 * tree-ssa-structalias.c (alias_get_name): Same.
20653 2017-07-25 Bin Cheng <bin.cheng@arm.com>
20656 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
20657 instructions if no du chain is found.
20659 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20661 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
20663 2017-07-25 Richard Biener <rguenther@suse.de>
20665 PR middle-end/81505
20666 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
20669 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
20671 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
20672 upper-regs options.
20673 (ISA_2_7_MASKS_SERVER): Likewise.
20674 (ISA_3_0_MASKS_IEEE): Likewise.
20675 (OTHER_P8_VECTOR_MASKS): Likewise.
20676 (OTHER_VSX_VECTOR_MASKS): Likewise.
20677 (POWERPC_MASKS): Likewise.
20678 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
20679 duplicate list of options.
20680 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
20681 explicit -mupper-regs options.
20682 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
20683 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
20684 alias for -mupper-regs-df.
20685 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20686 (rs6000_init_hard_regno_mode_ok): Likewise.
20687 (rs6000_option_override_internal): Likewise.
20688 (rs6000_opt_masks): Likewise.
20689 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
20690 options in terms of whether -mvsx or -mpower8-vector was used.
20691 (TARGET_UPPER_REGS_DI): Likewise.
20692 (TARGET_UPPER_REGS_SF): Likewise.
20693 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
20694 -mupper-regs-* options.
20696 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20698 * passes.c (emergency_dump_function): Print some empty lines and a
20699 header before the RTL dump.
20701 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20703 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
20705 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
20708 * config/aarch64/aarch64.c (aarch64_classify_symbol):
20709 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
20711 2017-07-24 Carl Love <cel@us.ibm.com>
20713 * config/rs6000/rs6000-c.c: Add support for built-in functions
20714 vector float vec_extract_fp32_from_shorth (vector unsigned short);
20715 vector float vec_extract_fp32_from_shortl (vector unsigned short);
20716 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
20717 vec_extract_fp_from_shortl): Add defines for the two builtins.
20718 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
20719 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
20721 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
20722 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
20723 * doc/extend.texi: Update the built-in documentation file for the
20724 new built-in function.
20726 2017-07-24 Jakub Jelinek <jakub@redhat.com>
20729 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
20731 * doc/generic.texi: Likewise.
20732 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
20733 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
20735 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
20737 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
20738 (aarch64_mls_elt_merge<mode>): Likewise.
20740 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
20742 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
20743 having __cxa_atexit.
20745 2017-07-23 Michael Collison <michael.collison@arm.com>
20747 * config/arm/arm.c (arm_option_override): Deprecate
20748 use of -mstructure-size-boundary.
20749 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
20750 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
20752 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20755 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
20756 Reduce cost estimate for direct moves.
20758 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
20761 * config/i386/i386.c (ix86_option_override_internal): Disable
20762 BMI, BMI2 and TBM instructions for -m16.
20764 2017-07-21 Carl Love <cel@us.ibm.com>
20766 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20767 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20768 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20769 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20770 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20771 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20772 VMULOSW): New enum "unspec" values.
20773 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20774 altivec_vmulosw): New patterns.
20775 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20776 VMULOSW): Add definitions.
20778 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
20780 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
20781 (qdf24xx): Likewise.
20782 * config/aarch64/aarch64-options-extensions.def (rdma); New.
20783 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
20784 (AARCH64_FL_V8_1): Renumber.
20785 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
20786 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
20787 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
20788 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
20789 rdma to feature modifiers list.
20791 2017-07-21 Yury Gribov <tetra2005@gmail.com>
20793 PR middle-end/56727
20794 * ipa-visibility (function_and_variable_visibility): Convert
20795 recursive PLT call to direct call if appropriate.
20797 2017-07-21 Andrew Pinski <apinski@cavium.com>
20799 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
20800 operand 1 to see if the types precision matches.
20801 * fold-const.c (operand_equal_p): Likewise.
20803 2017-07-21 Richard Biener <rguenther@suse.de>
20805 PR tree-optimization/81303
20806 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
20807 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
20808 (vect_peeling_hash_get_lowest_cost): Adjust.
20809 (vect_enhance_data_refs_alignment): Likewise. Use
20810 vect_get_peeling_costs_all_drs to compute the penalty for no
20811 peeling to match up costs.
20813 2017-07-21 Richard Biener <rguenther@suse.de>
20815 PR tree-optimization/81500
20816 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
20817 we didn't identify a reduction path.
20819 2017-07-21 Tom de Vries <tom@codesourcery.com>
20820 Cesar Philippidis <cesar@codesourcery.com>
20822 PR gcov-profile/81442
20823 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
20826 2017-07-21 Tom de Vries <tom@codesourcery.com>
20829 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
20831 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
20832 nvptx_override_options_after_change.
20834 2017-07-21 Ulrich Drepper <drepper@redhat.com>
20836 * dwarf2out.c (output_file_names): Avoid double testing for
20837 dwarf_version >= 5.
20839 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
20841 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
20843 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20845 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
20847 (try_crossjump_to_edge): Do not punt on partitioned functions.
20849 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20851 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
20852 Put all BBs reachable only via paths crossing cold region to cold
20854 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
20856 2016-07-21 Richard Biener <rguenther@suse.de>
20858 PR tree-optimization/81303
20859 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
20860 into account prologue and epilogue iterations when raising
20861 min_profitable_iters to sth at least covering one vector iteration.
20863 2017-07-21 Tamar Christina <tamar.christina@arm.com>
20865 * config/arm/arm.c (arm_test_cpu_arch_dat):
20868 2017-07-20 Nathan Sidwell <nathan@acm.org>
20870 Remove TYPE_METHODS.
20871 * tree.h (TYPE_METHODS): Delete.
20872 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
20873 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
20874 (dbxout_type_methods): Scan TYPE_FIELDS.
20875 (dbxout_type): Don't check TYPE_METHODS here.
20876 * function.c (use_register_for_decl): Always ignore register for
20877 class types when not optimizing.
20878 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
20879 * tree.c (free_lang_data_in_type): Stitch out member functions and
20880 templates from TYPE_FIELDS.
20881 (build_distinct_type_copy, verify_type_variant,
20882 verify_type): Member fns are on TYPE_FIELDS.
20883 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
20884 * tree-pretty-print.c (dump_generic_node): Likewise.
20886 2017-07-20 Jakub Jelinek <jakub@redhat.com>
20889 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
20890 V2TImode and V4TImode.
20891 (ix86_expand_vector_extract): Likewise.
20892 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
20893 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
20894 (ssescalarmode): Handle V4TImode and V2TImode.
20895 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
20896 (*vec_extractv2ti, *vec_extractv4ti): New insns.
20897 (VEXTRACTI128_MODE): New mode iterator.
20898 (splitter for *vec_extractv?ti first element): New.
20899 (VEC_INIT_MODE): New mode iterator.
20900 (vec_init<mode>): Consolidate 3 expanders into one using
20901 VEC_INIT_MODE mode iterator.
20903 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20905 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
20906 non_spilled_static_chain_regno_p.
20908 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20910 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
20912 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
20914 * bb-reorder.c (connect_traces): Allow copying of blocks within
20917 2017-07-20 Richard Biener <rguenther@suse.de>
20919 * gimple.h (gimple_phi_result): Add gphi * overload.
20920 (gimple_phi_result_ptr): Likewise.
20921 (gimple_phi_arg): Likewise. Adjust index assert to only
20922 allow actual argument accesses rather than all slots available
20924 (gimple_phi_arg_def): Add gphi * overload.
20925 * tree-phinodes.c (make_phi_node): Initialize only actual
20927 (resize_phi_node): Clear memory not covered by old node,
20928 do not initialize excess argument slots.
20929 (reserve_phi_args_for_new_edge): Initialize new argument slot
20932 2017-07-20 Bin Cheng <bin.cheng@arm.com>
20934 PR tree-optimization/81388
20936 2016-07-21 Bin Cheng <bin.cheng@arm.com>
20938 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
20939 by removing computation of may_be_zero.
20941 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20942 Tom de Vries <tom@codesourcery.com>
20944 PR middle-end/81030
20945 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
20946 when gimple level profile disagrees with what RTL expander did.
20948 2017-07-20 Richard Biener <rguenther@suse.de>
20950 PR tree-optimization/61171
20951 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
20952 (vect_analyze_stmt): Add slp instance parameter.
20953 (vectorizable_reduction): Likewise.
20954 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
20955 (vect_is_simple_reduction): Deal with chains not detected
20956 as SLP reduction chain, specifically not properly associated
20957 chains containing a mix of plus/minus.
20958 (get_reduction_op): Remove.
20959 (get_initial_defs_for_reduction): Simplify, pass in whether
20960 this is a reduction chain, pass in the SLP node for the PHIs.
20961 (vect_create_epilog_for_reduction): Get the SLP instance as
20963 (vectorizable_reduction): Get the SLP instance as arg.
20964 During analysis remember the SLP node with the PHIs in the
20965 instance. Simplify getting at the vectorized reduction PHIs.
20966 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
20967 through SLP instance.
20968 (vect_slp_analyze_operations): Likewise.
20969 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
20970 (vect_transform_stmt): Likewise.
20972 2017-07-20 Tom de Vries <tom@codesourcery.com>
20974 PR tree-optimization/81489
20975 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
20976 read of phi arg location to before loop that modifies phi.
20978 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20980 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
20983 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20985 PR middle-end/81331
20986 * except.c (execute): Fix ordering issue.
20988 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20990 PR rtl-optimization/81423
20991 * combine.c (make_compound_operation_int): Don't try to optimize
20992 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
20994 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20996 PR rtl-optimization/81423
20997 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
20998 with a constant that is -1 in the truncated to mode.
21000 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
21002 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
21003 (determine_unlikely_bbs): ... here.
21004 * predict.h (propagate_unlikely_bbs_forward): Declare.
21005 * cfgexpand.c (pass_expand::execute): Use it.
21006 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
21008 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
21009 propagate_unlikely_bbs_forward.
21011 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
21013 PR middle-end/81331
21014 * except.c (maybe_add_nop_after_section_switch): New function.
21017 2017-07-19 Tom de Vries <tom@codesourcery.com>
21019 * gimple.h (gimple_phi_set_arg): Make assert more strict.
21021 2017-07-19 Tom de Vries <tom@codesourcery.com>
21023 * gimple.h (gimple_phi_arg): Make assert more strict.
21025 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
21027 * config.gcc (powerpc*-*-*): Add mmintrin.h.
21028 * config/rs6000/mmintrin.h: New file.
21029 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
21031 2017-07-19 Jakub Jelinek <jakub@redhat.com>
21033 PR tree-optimization/81346
21034 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
21036 2017-07-19 Tom de Vries <tom@codesourcery.com>
21038 * config/nvptx/nvptx.md (VECIM): Add V2DI.
21040 2017-07-19 Tom de Vries <tom@codesourcery.com>
21042 * config/nvptx/nvptx-modes.def: Add V2DImode.
21043 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
21044 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
21045 (nvptx_output_mov_insn): Handle lack of mov.b128.
21046 (nvptx_print_operand): Handle 'H' and 'L' codes.
21047 (nvptx_vector_mode_supported): Allow V2DImode.
21048 (nvptx_preferred_simd_mode): New function.
21049 (nvptx_data_alignment): New function.
21050 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
21051 nvptx_preferred_simd_mode.
21052 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
21054 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
21056 2017-07-19 Tom de Vries <tom@codesourcery.com>
21058 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
21059 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
21060 (nvptx_vector_mode_supported): New function. Allow V2SImode.
21061 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
21062 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
21063 (mov<VECIM>_insn): New define_insn.
21064 (define_expand "mov<VECIM>): New define_expand.
21066 2017-07-19 Tom de Vries <tom@codesourcery.com>
21068 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
21070 2017-07-19 Jakub Jelinek <jakub@redhat.com>
21072 PR tree-optimization/81346
21073 * fold-const.h (fold_div_compare, range_check_type): Declare.
21074 * fold-const.c (range_check_type): New function.
21075 (build_range_check): Use range_check_type.
21076 (fold_div_compare): No longer static, rewritten into
21077 a match.pd helper function.
21078 (fold_comparison): Don't call fold_div_compare here.
21079 * match.pd (X / C1 op C2): New optimization using fold_div_compare
21080 as helper function.
21082 2017-07-19 Nathan Sidwell <nathan@acm.org>
21084 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
21085 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
21086 * tree.c (find_decls_types_r, verify_type): Use
21087 TYPE_{MIN,MAX}_VALUE_RAW.
21088 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
21089 (hash_tree): Likewise.
21090 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
21092 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
21095 2017-07-18 Tom de Vries <tom@codesourcery.com>
21097 PR middle-end/81464
21098 * omp-expand.c (expand_omp_for_static_chunk): Handle
21099 equal-argument loop exit phi.
21101 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
21104 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
21105 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
21106 operand 2 predicate.
21107 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
21108 operand 2 predicate.
21109 (ror,rol -> rorx splitters): Use const_int_operand as
21110 operand 2 predicate.
21112 2017-06-18 Richard Biener <rguenther@suse.de>
21114 PR tree-optimization/81410
21115 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
21116 the gap in the ! slp_perm SLP case after each group.
21118 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
21120 PR middle-end/81463
21121 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
21124 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
21126 PR middle-end/81462
21127 * predict.c (set_even_probabilities): Cleanup; do not affect
21128 probabilities that are already known.
21129 (combine_predictions_for_bb): Call even when count is set.
21131 2017-07-18 Nathan Sidwell <nathan@acm.org>
21133 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
21136 2017-07-18 Bin Cheng <bin.cheng@arm.com>
21139 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
21140 optimization for loop niter analysis.
21142 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
21145 * config/avr/avr.c (avr_optimize_casesi): Don't use
21146 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
21148 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
21150 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
21151 body_cost_vec from _vect_peel_extended_info.
21152 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
21153 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
21156 2017-07-18 Bin Cheng <bin.cheng@arm.com>
21158 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
21160 2017-07-18 Richard Biener <rguenther@suse.de>
21162 PR tree-optimization/80620
21163 PR tree-optimization/81403
21164 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
21165 info when re-using a VN table entry.
21167 2017-07-18 Richard Biener <rguenther@suse.de>
21169 PR tree-optimization/81418
21170 * tree-vect-loop.c (vectorizable_reduction): Properly compute
21171 vectype_in. Verify that with lane-reducing reduction operations
21172 we have a single def-use cycle.
21174 2017-07-17 Carl Love <cel@us.ibm.com>
21176 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
21178 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21179 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21180 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21181 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21182 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21183 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21184 VMULOSW): New enum "unspec" values.
21185 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
21186 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
21187 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21188 altivec_vmulosw): New patterns.
21189 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21190 VMULOSW): Add definitions.
21192 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
21194 * config/alpha/alpha.c: Include predict.h.
21196 2017-07-17 Yury Gribov <tetra2005@gmail.com>
21198 * tree-vrp.c (compare_assert_loc): Fix comparison function
21199 to return predictable results.
21201 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21203 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
21205 (subdi3): Likewise.
21206 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
21207 * doc/invoke.texi (mexpand-adddi): Update text.
21209 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21211 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
21212 that also clobbers the CC register. The old expand code is moved
21214 (*arc_clzsi2): ... here.
21215 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
21216 the CC register. The old expand code is moved to ...
21217 (arc_ctzsi2): ... here.
21219 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21221 * config/arc/arc.opt (mindexed-loads): Use initial value
21222 TARGET_INDEXED_LOADS_DEFAULT.
21223 (mauto-modify-reg): Use initial value
21224 TARGET_AUTO_MODIFY_REG_DEFAULT.
21225 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
21226 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
21227 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
21228 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
21230 2017-07-17 Martin Liska <mliska@suse.cz>
21233 * opts.c (finish_options): Do not allow -fgnu-tm
21234 w/ -fsanitize={kernel-,}address. Say sorry.
21236 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21239 * tree-loop-distribution.c (classify_partition): Only assert on
21240 numer of iterations.
21241 (merge_dep_scc_partitions): Delete prameter. Update function call.
21242 (distribute_loop): Remove code handling loop with unknown niters.
21243 (pass_loop_distribution::execute): Skip loop with unknown niters.
21245 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21248 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
21249 function sort_partitions_by_post_order.
21251 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21253 PR tree-optimization/81374
21254 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
21255 the max index of basic blocks, rather than number of basic blocks.
21257 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21259 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
21261 (arc_legitimate_pic_operand_p): Likewise.
21262 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
21264 (arc_needs_pcl_p): Likewise.
21265 (arc_legitimate_pc_offset_p): Likewise.
21266 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
21267 function is also used in constrains.md.
21268 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
21269 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
21270 PLUS. Only return true/false in known cases, otherwise assert.
21271 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
21272 is already called in arc_legitimate_constant_p.
21273 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
21275 (LEGITIMATE_PIC_OPERAND_P): Use
21276 arc_raw_symbolic_reference_mentioned_p function.
21277 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
21282 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21283 Andrew Burgess <andrew.burgess@embecosm.com>
21285 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
21286 (arc_return_address_register): New function.
21287 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
21288 (arc_handle_fndecl_attribute): Add naked attribute.
21289 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21290 (TARGET_WARN_FUNC_RETURN): Likewise.
21291 (arc_allocate_stack_slots_for_args): New function.
21292 (arc_warn_func_return): Likewise.
21293 (machine_function): Change type fn_type.
21294 (arc_compute_function_type): Consider new naked function type,
21295 change function return type.
21296 (arc_must_save_register): Adapt to handle new
21297 arc_compute_function_type's return type.
21298 (arc_expand_prologue): Likewise.
21299 (arc_expand_epilogue): Likewise.
21300 (arc_return_address_regs): Delete.
21301 (arc_return_address_register): New function.
21302 (arc_epilogue_uses): Use above function.
21303 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
21304 (arc_function_type): Change encoding, add naked type.
21305 (ARC_INTERRUPT_P): Change to handle the new encoding.
21306 (ARC_FAST_INTERRUPT_P): Likewise.
21307 (ARC_NORMAL_P): Define.
21308 (ARC_NAKED_P): Likewise.
21309 (arc_compute_function_type): Delete prototype.
21310 * config/arc/arc.md (in_ret_delay_slot): Use
21311 arc_return_address_register function.
21312 (simple_return): Likewise.
21313 (p_return_i): Likewise.
21315 2017-07-17 Jakub Jelinek <jakub@redhat.com>
21317 PR tree-optimization/81428
21318 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
21319 can't be built for those types.
21321 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
21323 Remove stuff dead since r239246.
21325 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
21326 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
21327 (avr_inform_devices): Remove dead stuff.
21329 2017-07-17 Tamar Christina <tamar.christina@arm.com>
21331 * config/arm/arm_neon.h: Fix softp typo.
21333 2017-07-17 Jakub Jelinek <jakub@redhat.com>
21335 PR tree-optimization/81365
21336 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
21337 aggregate moves onto bb predecessor edges, make sure there are no
21338 loads that could alias the lhs in between the start of bb and the
21341 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
21344 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
21345 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
21346 [LSHIFTRT, outer_code = TRUNCATE]: Same.
21348 2017-07-17 Jakub Jelinek <jakub@redhat.com>
21350 PR tree-optimization/81396
21351 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
21352 (init_symbolic_number): Initialize it to 1.
21353 (perform_symbolic_merge): Add n_ops from both operands into the new
21355 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
21356 without base_addr as useless if they need more than one operation.
21357 (bswap_replace): Handle !bswap case for NULL base_addr.
21359 2017-07-17 Tom de Vries <tom@codesourcery.com>
21362 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
21365 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
21367 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
21368 conditional builtin define __FIX_LEON3FT_B2BST.
21370 2017-07-17 Daniel Cederman <cederman@gaisler.com>
21372 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
21373 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
21376 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
21378 PR rtl-optimization/81424
21379 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
21380 to remove potential trapping from operands if -fnon-call-exceptions.
21382 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21384 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
21385 profile_proability for scalling.
21386 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
21388 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21390 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
21392 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21394 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
21395 fixpoint arithmetics.
21397 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21399 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
21400 fixpoint arithmetics.
21402 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21404 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
21405 fixpoint arithmetics.
21407 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21409 * profile-count.h (profile_probability::from_reg_br_prob_note,
21410 profile_probability::to_reg_br_prob_note): New functions.
21411 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
21412 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
21413 * predict.c (probability_reliable_p): Update.
21414 (edge_probability_reliable_p): Update.
21415 (br_prob_note_reliable_p): Update.
21416 (invert_br_probabilities): Update.
21417 (add_reg_br_prob_note): New function.
21418 (combine_predictions_for_insn): Update.
21419 * asan.c (asan_clear_shadow): Update.
21420 * cfgbuild.c (compute_outgoing_frequencies): Update.
21421 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21422 (update_br_prob_note): Update.
21423 (rtl_verify_edges): Update.
21424 (purge_dead_edges): Update.
21425 (fixup_reorder_chain): Update.
21426 * emit-rtl.c (try_split): Update.
21427 * ifcvt.c (cond_exec_process_insns): Update.
21428 (cond_exec_process_if_block): Update.
21429 (dead_or_predicable): Update.
21430 * internal-fn.c (expand_addsub_overflow): Update.
21431 (expand_neg_overflow): Update.
21432 (expand_mul_overflow): Update.
21433 * loop-doloop.c (doloop_modify): Update.
21434 * loop-unroll.c (compare_and_jump_seq): Update.
21435 * optabs.c (emit_cmp_and_jump_insn_1): Update.
21436 * predict.h: Update.
21437 * reorg.c (mostly_true_jump): Update.
21439 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
21440 * config/alpha/alpha.c (emit_unlikely_jump): Update.
21441 * config/arc/arc.c: (emit_unlikely_jump): Update.
21442 * config/arm/arm.c: (emit_unlikely_jump): Update.
21443 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
21444 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
21445 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
21446 (ix86_print_operand): Update.
21447 (ix86_split_fp_branch): Update.
21448 (predict_jump): Update.
21449 * config/ia64/ia64.c (ia64_print_operand): Update.
21450 * config/mmix/mmix.c (mmix_print_operand): Update.
21451 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
21452 (rs6000_expand_split_stack_prologue): Update.
21453 * config/rs6000/rs6000.c: Update.
21454 * config/s390/s390.c (s390_expand_vec_strlen): Update.
21455 (s390_expand_vec_movstr): Update.
21456 (s390_expand_cs_tdsi): Update.
21457 (s390_expand_split_stack_prologue): Update.
21458 * config/sh/sh.c (sh_print_operand): Update.
21459 (expand_cbranchsi4): Update.
21460 (expand_cbranchdi4): Update.
21461 * config/sparc/sparc.c (output_v9branch): Update.
21462 * config/spu/spu.c (get_branch_target): Update.
21463 (ea_load_store_inline): Update.
21464 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
21465 * config/tilepro/tilepro.c: Update.
21467 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
21469 * gimplify.c (mostly_copy_tree_r): Revert latest change.
21470 (gimplify_save_expr): Likewise.
21472 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21474 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
21476 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21478 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
21480 * timevar.def (TV_IPA_FNSUMMARY): Define.
21482 2017-07-16 Daniel Cederman <cederman@gaisler.com>
21484 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
21485 to back store errata sensitive sequence from being generated.
21486 (sqrtdf2_fix): Likewise.
21488 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21490 * tree-ssa-threadupdate.c (compute_path_counts,
21491 update_joiner_offpath_counts): Use profile_probability.
21493 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
21496 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21498 * config/arm/arm-c.c (arm_cpu_builtins): Define
21499 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
21501 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
21503 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21504 array entries to represent __ieee128 versions of the
21505 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
21506 scalar_extract_sig, and scalar_insert_exp built-in functions.
21507 (altivec_resolve_overloaded_builtin): Add special case handling
21508 for the __builtin_scalar_insert_exp function, as represented by
21509 the P9V_BUILTIN_VEC_VSIEDP constant.
21510 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
21511 exponent support for __ieee128 argument.
21512 (VSESQP): Add scalar extract signature support for __ieee128
21514 (VSTDCNQP): Add scalar test negative support for __ieee128
21516 (VSIEQP): Add scalar insert exponent support for __int128 argument
21517 with __ieee128 result.
21518 (VSIEQPF): Add scalar insert exponent support for __ieee128
21519 argument with __ieee128 result.
21520 (VSTDCQP): Add scalar test data class support for __ieee128
21522 (VSTDCNQP): Add overload support for scalar test negative with
21523 __ieee128 argument.
21524 (VSTDCQP): Add overload support for scalar test data class
21525 __ieee128 argument.
21526 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
21527 UNSPEC_VSX_SXSIGDP.
21528 (UNSPEC_VSX_SIEXPQP): New constant.
21529 (xsxexpqp): New insn for VSX scalar extract exponent quad
21531 (xsxsigqp): New insn for VSX scalar extract significand quad
21533 (xsiexpqpf): New insn for VSX scalar insert exponent quad
21534 precision with floating point argument.
21535 (xststdcqp): New expand for VSX scalar test data class quad
21537 (xststdcnegqp): New expand for VSX scalar test negative quad
21539 (xststdcqp): New insn to match expansions for VSX scalar test data
21540 class quad precision and VSX scalar test negative quad precision.
21541 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
21542 special case operand checking to enforce that second operand of
21543 VSX scalar test data class with quad precision argument is a 7-bit
21545 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21546 prototypes and descriptions of __ieee128 versions of
21547 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
21548 scalar_test_data_class, and scalar_test_neg built-in functions.
21550 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21552 PR tree-optimization/81162
21553 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
21554 replace a negate with an add.
21556 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
21558 * doc/invoke.texi (arm/-mcpu): Document +crypto.
21560 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21562 * config/arm/arm-c.c (arm_cpu_builtins): Define
21563 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
21565 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21567 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
21568 (armv8-r): Set ARM Cortex-R52 as default CPU.
21569 * config/arm/arm-tables.opt: Regenerate.
21570 * config/arm/arm-tune.md: Regenerate.
21571 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
21573 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
21574 extension for -mcpu=cortex-r52.
21576 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21578 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
21579 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
21580 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
21581 (fp-armv8): Define it as FP_ARMv8 only.
21582 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
21583 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
21585 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
21586 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
21587 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
21588 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
21589 than TARGET_FPU_ARMV8.
21590 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
21591 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
21592 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
21594 * config/arm/neon.md (neon_vrint): Likewise.
21595 (neon_vcvt): Likewise.
21596 (neon_<fmaxmin_op><mode>): Likewise.
21597 (<fmaxmin><mode>3): Likewise.
21598 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
21599 * config/arm/predicates.md (arm_cond_move_operator): Check against
21600 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
21602 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
21604 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
21605 to top of function.
21607 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21609 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
21610 loop in comment with memset.
21612 2017-07-14 Martin Liska <mliska@suse.cz>
21614 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
21615 * dwarf2out.c (is_java): Remove the function.
21616 (output_pubname): Remove usage of the function.
21617 (lower_bound_default): Remove usage of DW_LANG_Java.
21618 (gen_compile_unit_die): Likewise.
21619 * gcc.c: Remove compiler defaults for .java and .zip files.
21620 * gimple-expr.c (remove_suffix): Change as there's no longer
21621 extension than 4-letter one.
21622 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
21623 (gimplify_save_expr): Likewise.
21624 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
21625 as it's possible even for other languages than Java.
21626 * langhooks.h (struct lang_hooks): Remove Java from a comment.
21627 * lto-opts.c (lto_write_options): Remove reference to Java.
21628 * opts.c (strip_off_ending): Update file extension handling.
21629 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
21630 * tree-eh.c (lower_resx): Likewise.
21631 * tree.c (free_lang_data_in_type): Remove dead code.
21632 (find_decls_types_r): Likewise.
21633 (build_common_builtin_nodes): Remove Java from a comment.
21634 (verify_type): Remove dead code.
21635 * varasm.c (assemble_external): Remove Java from a comment.
21637 2017-07-14 Martin Liska <mliska@suse.cz>
21639 * opts.c (finish_options): Add quotes.
21640 (common_handle_option): Likewise.
21642 2017-07-14 Martin Liska <mliska@suse.cz>
21644 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
21645 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
21646 Remove N_SO_PASCAL.
21647 * dwarf2out.c (lower_bound_default): Do not handle
21649 (gen_compile_unit_die): Likewise.
21650 * gcc.c: Remove default extension binding for GNU Pascal.
21651 * stmt.c: Remove Pascal language from a comment.
21652 * xcoffout.c: Likewise.
21654 2017-07-13 David Malcolm <dmalcolm@redhat.com>
21657 * diagnostic-show-locus.c (fixit_cmp): New function.
21658 (layout::layout): Sort m_fixit_hints.
21659 (column_range::column_range): Assert that the values are valid.
21660 (struct char_span): New struct.
21661 (correction::overwrite): New method.
21662 (struct source_line): New struct.
21663 (line_corrections::add_hint): Add assertions. Reimplement memcpy
21664 calls in terms of classes source_line and char_span, and
21665 correction::overwrite.
21666 (selftest::test_overlapped_fixit_printing_2): New function.
21667 (selftest::diagnostic_show_locus_c_tests): Call it.
21669 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
21671 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
21672 early if there is no lhs.
21674 2017-07-13 Martin Liska <mliska@suse.cz>
21676 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
21677 (gen_reference_type_die): Likewise.
21678 * stor-layout.c: Remove Pascal-related comment.
21680 2017-07-13 Martin Liska <mliska@suse.cz>
21682 * opts.c (finish_options): Add quotes to error messages.
21683 (parse_sanitizer_options): Likewise.
21685 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
21687 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
21689 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
21691 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
21693 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
21695 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
21697 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
21699 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
21702 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
21703 provides the hardware capability bits, define the macro
21704 __BUILTIN_CPU_SUPPORTS__.
21705 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
21706 if GLIBC does not provide the hardware capability bits. Add a
21707 gcc_unreachable call if the built-in cpu function is neither
21708 __builtin_cpu_is nor __builtin_cpu_supports.
21709 (rs6000_get_function_versions_dispatcher): Change the warning
21710 that an old GLIBC is used which does not export the capability
21711 bits to be an error.
21712 * doc/extend.texi (target_clones attribute): Document the
21713 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
21714 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
21715 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
21716 the macros defined by GCC if the newer GLIBC is available.
21718 2017-07-12 Jeff Law <law@redhat.com>
21720 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
21721 remaining includes slightly.
21722 * config/riscv/riscv-builtins.c: Include profile-count.h.
21724 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21727 * config/avr/avr.c (avr_set_current_function): In diagnostic
21728 messages: Quote keywords and (parts of) identifiers.
21729 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
21732 2017-07-12 Carl Love <cel@us.ibm.com>
21734 * config/rs6000/rs6000-c.c: Add support for built-in functions
21735 vector bool char vec_revb (vector bool char);
21736 vector bool short vec_revb (vector short char);
21737 vector bool int vec_revb (vector bool int);
21738 vector bool long long vec_revb (vector bool long long);
21739 * doc/extend.texi: Update the built-in documentation file for the
21740 new built-in functions.
21742 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21744 * config/s390/s390.md: Remove movcc splitter.
21746 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21748 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
21749 load/store on condition.
21751 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21754 * config/avr/avr.c (avr_encode_section_info)
21755 [progmem && !TREE_READONLY]: Error if progmem object needs
21758 2017-07-11 Michael Collison <michael.collison@arm.com>
21760 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
21763 2017-07-11 Carl Love <cel@us.ibm.com>
21765 * config/rs6000/rs6000-c.c: Add support for builtins
21766 vector unsigned int vec_parity_lsbb (vector signed int);
21767 vector unsigned int vec_parity_lsbb (vector unsigned int);
21768 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
21769 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
21770 vector unsigned long long vec_parity_lsbb (vector signed long long);
21771 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
21772 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
21773 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
21774 * doc/extend.texi: Update the built-in documentation file for the
21775 new built-in functions.
21777 2017-07-11 David Malcolm <dmalcolm@redhat.com>
21779 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
21780 (layout::m_primary_loc): New field.
21781 (layout::layout): Initialize new field. Move location filtering
21782 logic from here to...
21783 (layout::maybe_add_location_range): ...this new method. Add
21784 support for filtering to just the lines already specified by other
21786 (layout::will_show_line_p): New method.
21787 (gcc_rich_location::add_location_if_nearby): New method.
21788 (selftest::test_add_location_if_nearby): New test function.
21789 (selftest::diagnostic_show_locus_c_tests): Call it.
21790 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
21793 2017-07-11 Tom de Vries <tom@codesourcery.com>
21795 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
21796 (bb_first_real_insn): New function.
21797 (nvptx_single): Add extra initialization of broadcasted condition
21800 2017-07-11 Nathan Sidwell <nathan@acm.org>
21802 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
21804 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
21806 * doc/extend.texi (AVR Function Attributes): Remove weblink to
21807 Binutils doc as TEXI will mess them up.
21808 * doc/invoke.texi (AVR Options): Same here.
21810 2017-07-11 Daniel Cederman <cederman@gaisler.com>
21812 * config/sparc/sparc.opt (mfix-ut700): New option.
21813 (mfix-gr712rc): Likewise.
21814 (sparc_fix_b2bst): New variable.
21815 * doc/invoke.texi (SPARC options): Document them.
21816 (ARM options): Fix warnings.
21817 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
21818 instructions to prevent sequences that can trigger the store-store
21819 errata for certain LEON3FT processors.
21820 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
21821 (sparc_option_override): Set sparc_fix_b2bst appropriately.
21822 * config/sparc/sparc.md (fix_b2bst): New attribute.
21823 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
21825 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
21828 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
21830 (*rsqrtsf2_sse): Ditto.
21832 (div<mode>3): Macroize insn from divdf3 and divsf3
21833 using MODEF mode iterator.
21835 2017-07-10 Martin Sebor <msebor@redhat.com>
21837 PR tree-optimization/80397
21838 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
21839 instead of testing for equality to INTEGER_TYPE.
21841 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
21843 * config.gcc: Remove uclibc from arc target spec.
21845 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
21847 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
21849 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21852 * lto-wrapper.c (remove_option): New function.
21853 (merge_and_complain): Merge PIC/PIE options more realistically.
21855 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21857 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
21861 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
21862 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
21863 * config.in: Regenerate.
21864 * configure: Regenerate.
21865 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
21866 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
21867 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
21868 (TARGET_GASISR_PROLOGUES): ...target mask.
21869 * common/config/avr/avr-common.c
21870 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
21871 Set -mgas-isr-prologues.
21872 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
21873 INSERT_PASS_BEFORE for it.
21874 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
21875 * config/avr/avr.c (avr_option_override)
21876 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
21877 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
21878 (avr_attribute_table) <no_gccisr>: Add new function attribute.
21879 (avr_set_current_function) <is_no_gccisr>: Init machine field.
21880 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
21882 (make_avr_pass_pre_proep): New function.
21883 (emit_push_sfr) <treg>: Add argument to function and use it
21884 instead of TMP_REG.
21885 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
21886 and set machine->gasisr.yes.
21887 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
21888 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
21889 __gcc_isr.n_pushed to .L__stack_usage.
21890 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
21891 (avr_asm_final_postscan_insn): ...this new static function.
21892 * config/avr/avr.h (machine_function)
21893 <is_no_gccisr, use_L__stack_usage>: New fields.
21894 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
21895 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
21896 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
21897 (gasisr, *gasisr): New expander and insn.
21898 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21899 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
21900 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
21902 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
21904 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
21907 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21909 Move jump-tables out of .text again.
21912 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
21913 (ASM_OUTPUT_ADDR_VEC): New function.
21914 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
21915 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
21916 INSN_ADDRESSes as asm comment.
21917 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
21918 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
21919 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
21920 * config/avr/avr.md (*tablejump): Adjust comment.
21921 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
21922 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
21924 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
21925 (avr_output_addr_vec): New proto.
21926 (avr_log_t) <insn_addresses>: New field.
21928 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
21931 * config/i386/i386.c (ix86_function_arg_advance): Set
21932 outgoing_args_on_stack to true if there are outgoing arguments
21934 (ix86_function_arg): Likewise.
21935 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
21936 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
21937 * config/i386/i386.h (machine_function): Add
21938 outgoing_args_on_stack.
21940 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
21942 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
21943 supporting pthreds.
21944 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
21946 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
21948 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
21949 (REAL_H): Remove $(MACHMODE_H).
21950 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
21952 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
21953 $(MACHMODE_H) and double-int.h.
21954 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
21956 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
21959 2017-07-07 Andrew Pinski <apinski@cavium.com>
21961 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
21962 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
21964 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21966 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
21967 Add warning if GCC was not configured to link against a GLIBC that
21968 exports the hardware capability bits.
21969 (make_resolver_func): Make resolver function private and not a
21970 COMDAT function. Create the name with clone_function_name instead
21971 of make_unique_name.
21974 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
21975 correct operand in doing the split.
21977 2017-07-07 Carl Love <cel@us.ibm.com>
21979 * config/rs6000/rs6000-c: Add support for built-in function
21980 vector unsigned short vec_pack_to_short_fp32 (vector float,
21982 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
21983 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
21984 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
21985 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
21986 (convert_4f32_8i16): Add define_expand.
21987 * doc/extend.texi: Update the built-in documentation file for the
21988 new built-in function.
21990 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21992 * config/sparc/m8.md: New file.
21993 * config/sparc/sparc.md: Include m8.md.
21995 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21997 * config/sparc/sparc.opt: New option -mvis4b.
21998 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
21999 (sparc_option_override): Handle VIS4B.
22000 (enum sparc_builtins): Define
22001 SPARC_BUILTIN_DICTUNPACK{8,16,32},
22002 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
22003 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
22004 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
22005 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
22006 (check_constant_argument): New function.
22007 (sparc_vis_init_builtins): Define builtins
22008 __builtin_vis_dictunpack{8,16,32},
22009 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
22010 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
22011 __builtin_vis_fpcmpde{8,16,32}shl and
22012 __builtin_vis_fpcmpur{8,16,32}shl.
22013 (sparc_expand_builtin): Check that the constant operands to
22014 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
22015 constant and in range.
22016 * config/sparc/sparc-c.c (sparc_target_macros): Handle
22018 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
22019 (SPARC_IMM5_P): Likewise.
22020 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
22021 (enabled): Handle vis4b.
22022 (UNSPEC_DICTUNPACK): New unspec.
22023 (UNSPEC_FPCMPSHL): Likewise.
22024 (UNSPEC_FPUCMPSHL): Likewise.
22025 (UNSPEC_FPCMPDESHL): Likewise.
22026 (UNSPEC_FPCMPURSHL): Likewise.
22027 (cpu_feature): New CPU feature `vis4b'.
22028 (dictunpack{8,16,32}): New insns.
22029 (FPCSMODE): New mode iterator.
22030 (fpcscond): New code iterator.
22031 (fpcsucond): Likewise.
22032 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
22033 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
22034 (fpcmpde{8,16,32}{si,di}shl): Likewise.
22035 (fpcmpur{8,16,32}{si,di}shl): Likewise.
22036 * config/sparc/constraints.md: Define constraints `q' for unsigned
22037 2-bit integer constants and `t' for unsigned 5-bit integer
22039 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
22041 (imm5_operand_dictunpack16): Likewise.
22042 (imm5_operand_dictunpack32): Likewise.
22043 (imm2_operand): Likewise.
22044 * doc/invoke.texi (SPARC Options): Document -mvis4b.
22045 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
22046 ditunpack* and fpcmp*shl builtins.
22048 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22050 * config.gcc: Handle m8 in --with-{cpu,tune} options.
22051 * config.in: Add HAVE_AS_SPARC6 define.
22052 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
22054 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
22056 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
22057 (CPP_CPU_SPEC): Handle m8.
22058 (ASM_CPU_SPEC): Likewise.
22059 * config/sparc/sparc-opts.h (enum processor_type): Add
22061 * config/sparc/sparc.c (m8_costs): New struct.
22062 (sparc_option_override): Handle TARGET_CPU_m8.
22063 (sparc32_initialize_trampoline): Likewise.
22064 (sparc64_initialize_trampoline): Likewise.
22065 (sparc_issue_rate): Likewise.
22066 (sparc_register_move_cost): Likewise.
22067 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
22068 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
22069 (ASM_CPU64_DEFAULT_SPEC): Likewise.
22070 (CPP_CPU_SPEC): Handle M8.
22071 (ASM_CPU_SPEC): Likewise.
22072 (AS_M8_FLAG): Define.
22073 * config/sparc/sparc.md: Add m8 to the cpu attribute.
22074 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
22075 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
22077 * configure: Regenerate.
22078 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
22081 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22083 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
22085 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
22086 ("*movdi_insn_sp32"): Do not set v3pipe.
22087 ("*movsi_insn"): Likewise.
22088 ("*movdi_insn_sp64"): Likewise.
22089 ("*movsf_insn"): Likewise.
22090 ("*movdf_insn_sp32"): Likewise.
22091 ("*movdf_insn_sp64"): Likewise.
22092 ("*zero_extendsidi2_insn_sp64"): Likewise.
22093 ("*sign_extendsidi2_insn"): Likewise.
22094 ("*mov<VM32:mode>_insn"): Likewise.
22095 ("*mov<VM64:mode>_insn_sp64"): Likewise.
22096 ("*mov<VM64:mode>_insn_sp32"): Likewise.
22097 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
22098 ("<vlop:code><VL:mode>3"): Likewise.
22099 ("*not_<vlop:code><VL:mode>3"): Likewise.
22100 ("*nand<VL:mode>_vis"): Likewise.
22101 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
22102 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
22103 ("one_cmpl<VL:mode>2"): Likewise.
22104 ("faligndata<VM64:mode>_vis"): Likewise.
22105 ("alignaddrsi_vis"): Likewise.
22106 ("alignaddrdi_vis"): Likweise.
22107 ("alignaddrlsi_vis"): Likewise.
22108 ("alignaddrldi_vis"): Likewise.
22109 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
22110 ("bmaskdi_vis"): Likewise.
22111 ("bmasksi_vis"): Likewise.
22112 ("bshuffle<VM64:mode>_vis"): Likewise.
22113 ("cmask8<P:mode>_vis"): Likewise.
22114 ("cmask16<P:mode>_vis"): Likewise.
22115 ("cmask32<P:mode>_vis"): Likewise.
22116 ("pdistn<P:mode>_vis"): Likewise.
22117 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
22119 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22121 * config/sparc/sparc.md ("subtype"): New insn attribute.
22122 ("*wrgsr_sp64"): Set insn subtype.
22123 ("*rdgsr_sp64"): Likewise.
22124 ("alignaddrsi_vis"): Likewise.
22125 ("alignaddrdi_vis"): Likewise.
22126 ("alignaddrlsi_vis"): Likewise.
22127 ("alignaddrldi_vis"): Likewise.
22128 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
22129 ("fexpand_vis"): Likewise.
22130 ("fpmerge_vis"): Likewise.
22131 ("faligndata<VM64:mode>_vis"): Likewise.
22132 ("bshuffle<VM64:mode>_vis"): Likewise.
22133 ("cmask8<P:mode>_vis"): Likewise.
22134 ("cmask16<P:mode>_vis"): Likewise.
22135 ("cmask32<P:mode>_vis"): Likewise.
22136 ("fchksm16_vis"): Likewise.
22137 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
22138 ("fmean16_vis"): Likewise.
22139 ("fp<plusminus_insn>64_vis"): Likewise.
22140 ("<plusminus_insn>v8qi3"): Likewise.
22141 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
22142 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
22143 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
22144 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
22145 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
22146 ("*movqi_insn"): Likewise.
22147 ("*movhi_insn"): Likewise.
22148 ("*movsi_insn"): Likewise.
22149 ("movsi_pic_gotdata_op"): Likewise.
22150 ("*movdi_insn_sp32"): Likewise.
22151 ("*movdi_insn_sp64"): Likewise.
22152 ("movdi_pic_gotdata_op"): Likewise.
22153 ("*movsf_insn"): Likewise.
22154 ("*movdf_insn_sp32"): Likewise.
22155 ("*movdf_insn_sp64"): Likewise.
22156 ("*zero_extendhisi2_insn"): Likewise.
22157 ("*zero_extendqihi2_insn"): Likewise.
22158 ("*zero_extendqisi2_insn"): Likewise.
22159 ("*zero_extendqidi2_insn"): Likewise.
22160 ("*zero_extendhidi2_insn"): Likewise.
22161 ("*zero_extendsidi2_insn_sp64"): Likewise.
22162 ("ldfsr"): Likewise.
22163 ("prefetch_64"): Likewise.
22164 ("prefetch_32"): Likewise.
22165 ("tie_ld32"): Likewise.
22166 ("tie_ld64"): Likewise.
22167 ("*tldo_ldub_sp32"): Likewise.
22168 ("*tldo_ldub1_sp32"): Likewise.
22169 ("*tldo_ldub2_sp32"): Likewise.
22170 ("*tldo_ldub_sp64"): Likewise.
22171 ("*tldo_ldub1_sp64"): Likewise.
22172 ("*tldo_ldub2_sp64"): Likewise.
22173 ("*tldo_ldub3_sp64"): Likewise.
22174 ("*tldo_lduh_sp32"): Likewise.
22175 ("*tldo_lduh1_sp32"): Likewise.
22176 ("*tldo_lduh_sp64"): Likewise.
22177 ("*tldo_lduh1_sp64"): Likewise.
22178 ("*tldo_lduh2_sp64"): Likewise.
22179 ("*tldo_lduw_sp32"): Likewise.
22180 ("*tldo_lduw_sp64"): Likewise.
22181 ("*tldo_lduw1_sp64"): Likewise.
22182 ("*tldo_ldx_sp64"): Likewise.
22183 ("*mov<VM32:mode>_insn"): Likewise.
22184 ("*mov<VM64:mode>_insn_sp64"): Likewise.
22185 ("*mov<VM64:mode>_insn_sp32"): Likewise.
22187 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22189 * config/sparc/sparc.md ("type"): New insn type viscmp.
22190 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
22192 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
22193 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
22194 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
22195 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
22197 ("n7_vis_logical_11cycle"): Likewise.
22198 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
22199 * config/sparc/niagara2.md ("niag3_vis": Likewise.
22200 * config/sparc/niagara.md ("niag_vis"): Likewise.
22201 * config/sparc/ultra3.md ("us3_fga"): Likewise.
22202 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
22204 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22206 * config/sparc/sparc.md: New instruction type `bmask'.
22207 (bmaskdi_vis): Use the `bmask' type.
22208 (bmasksi_vis): Likewise.
22209 * config/sparc/ultra3.md (us3_array): Likewise.
22210 * config/sparc/niagara7.md (n7_array): Likewise.
22211 * config/sparc/niagara4.md (n4_array): Likewise.
22212 * config/sparc/niagara2.md (niag2_vis): Likewise.
22213 (niag3_vis): Likewise.
22214 * config/sparc/niagara.md (niag_vis): Likewise.
22216 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22218 * ipa-comdats.c: Remove optimize check from gate.
22219 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
22220 for functions not optimized.
22221 (ipa_fn_summary_read): Skip optimize check.
22222 (ipa_fn_summary_write): Likewise.
22223 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
22225 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
22227 (can_inline_edge_p): Check flag_pcc_struct_return for match.
22228 (check_callers): Give up on caller which is not optimized.
22229 (inline_small_functions): Likewise.
22230 (ipa_inline): Do not give up when not optimizing.
22231 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
22232 away unoptimizes cdtors.
22233 (whole_program_function_and_variable_visibility): Do
22234 ipa_discover_readonly_nonaddressable_vars in LTO mode.
22235 * ipa.c (process_references): Do not check optimize.
22236 (symbol_table::remove_unreachable_nodes): Update optimize check.
22237 (set_writeonly_bit): Update optimize check.
22238 (pass_ipa_cdtor_merge::gate): Do not check optimize.
22239 (pass_ipa_single_use::gate): Remove.
22241 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22243 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
22244 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
22245 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
22246 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
22247 permute_load, permute_store, adjust_extract, adjust_splat,
22248 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
22249 replace_swap_with_copy, dump_swap_insn_table,
22250 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
22251 recombine_lvx_pattern, recombine_stvx_pattern,
22252 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
22253 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
22254 to file rs6000-p8swap.c.
22255 * config/rs6000/rs6000-p8swap.c: New file.
22256 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
22257 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
22258 and rs6000*-*-* targets.
22260 2017-07-06 David Malcolm <dmalcolm@redhat.com>
22262 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
22264 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22266 * lto-wrapper.c (merge_and_complain): Do not merge
22267 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
22268 fsigned_zeros, ftrapping_math, fwrapv.
22269 (append_compiler_options): Do not track these options.
22270 (append_linker_options): Likewie
22272 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22274 * cgraphunit.c (cgraph_node::finalize_function): When
22275 !flag_toplevel_reorde set no_reorder flag.
22276 (varpool_node::finalize_decl): Likewise.
22277 (symbol_table::compile): Drop no toplevel reorder path.
22279 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22281 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
22282 edges; zero probability is not better than uninitialized.
22284 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
22286 * asan.h (asan_sanitize_allocas_p): Declare.
22287 * asan.c (asan_sanitize_allocas_p): New function.
22288 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
22289 (handle_builtin_alloca): Likewise.
22290 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
22291 if !asan_sanitize_allocas_p.
22292 * params.def (asan-instrument-allocas): Add new option.
22293 * params.h (ASAN_PROTECT_ALLOCAS): Define.
22294 * opts.c (common_handle_option): Disable allocas sanitization for
22297 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
22299 * asan.c: Include gimple-fold.h.
22300 (get_last_alloca_addr): New function.
22301 (handle_builtin_stackrestore): Likewise.
22302 (handle_builtin_alloca): Likewise.
22303 (asan_emit_allocas_unpoison): Likewise.
22304 (get_mem_refs_of_builtin_call): Add new parameter, remove const
22305 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
22306 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
22307 (instrument_builtin_call): Pass gimple iterator to
22308 get_mem_refs_of_builtin_call.
22309 (last_alloca_addr): New global.
22310 * asan.h (asan_emit_allocas_unpoison): Declare.
22311 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
22312 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
22313 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
22314 if function calls alloca.
22315 * gimple-fold.c (replace_call_with_value): Remove static keyword.
22316 * gimple-fold.h (replace_call_with_value): Declare.
22317 * internal-fn.c: Include asan.h.
22318 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
22319 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
22321 2017-07-06 David Malcolm <dmalcolm@redhat.com>
22323 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
22324 (C_SELFTEST_FLAGS): New.
22325 (CPP_SELFTEST_FLAGS): New.
22326 (SELFTEST_DEPS): New, from deps of s-selftest.
22327 (C_SELFTEST_DEPS): New, from deps of s-selftest.
22328 (CPP_SELFTEST_DEPS): New.
22329 (selftest): Add dependency on s-selftest-c++.
22330 (s-selftest): Rename to...
22331 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
22332 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
22333 than SELFTEST_FLAGS.
22334 (selftest-gdb): Rename to...
22335 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
22337 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
22338 (selftest-valgrind): Rename to...
22339 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
22341 (selftest-valgrind): Reintroduce as an alias for
22342 selftest-c-valgrind.
22343 (s-selftest-c++): New.
22344 (selftest-c++-gdb): New.
22345 (selftest-c++-valgrind): New.
22347 2017-07-06 Olivier Hainque <hainque@adacore.com>
22349 * gcc.c (process_command): When deciding if undefined variables
22350 should be ignored when processing specs, accept "gcc -v" as well.
22352 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22354 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
22355 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
22357 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22359 * config/arm/arm-cpus.in (armv8-r): Add new entry.
22360 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
22361 * config/arm/arm-tables.opt: Regenerate.
22362 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
22364 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
22366 2017-07-06 Carl Love <cel@us.ibm.com>
22368 * ChangeLog: Clean up from mid air collision
22370 2017-07-06 Carl Love <cel@us.ibm.com>
22372 * config/rs6000/rs6000-c.c: Add support for built-in functions
22373 vector signed int vec_subc (vector signed int, vector signed int);
22374 vector signed __int128 vec_subc (vector signed __int128,
22375 vector signed __int128);
22376 vector unsigned __int128 vec_subc (vector unsigned __int128,
22377 vector unsigned __int128);
22378 vector signed int vec_sube (vector signed int, vector signed int,
22379 vector signed int);
22380 vector unsigned int vec_sube (vector unsigned int,
22381 vector unsigned int,
22382 vector unsigned int);
22383 vector signed __int128 vec_sube (vector signed __int128,
22384 vector signed __int128,
22385 vector signed__int128);
22386 vector unsigned __int128 vec_sube (vector unsigned __int128,
22387 vector unsigned __int128,
22388 vector unsigned __int128);
22389 vector signed int vec_subec (vector signed int, vector signed int,
22390 vector signed int);
22391 vector unsigned int vec_subec (vector unsigned int,
22392 vector unsigned int,
22393 vector unsigned int);
22394 vector signed __int128 vec_subec (vector signed __int128,
22395 vector signed __int128,
22396 vector signed__int128);
22397 vector unsigned __int128 vec_subec (vector unsigned __int128,
22398 vector unsigned __int128,
22399 vector unsigned __int128);
22400 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
22401 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
22402 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
22403 BU_ALTIVEC_OVERLOAD_X definitions.
22404 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
22405 * doc/extend.texi: Update the built-in documentation file for the new
22406 built-in functions.
22408 2017-07-06 David Malcolm <dmalcolm@redhat.com>
22411 * diagnostic-show-locus.c (layout::layout): Use start and finish
22412 spelling location for the start and finish of each range.
22413 * genmatch.c (linemap_client_expand_location_to_spelling_point):
22414 Add unused aspect param.
22415 * input.c (expand_location_1): Add "aspect" param, and use it
22416 to access the correct part of the location.
22417 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
22419 (expand_location_to_spelling_point): Likewise.
22420 (linemap_client_expand_location_to_spelling_point): Add "aspect"
22421 param, and pass it to expand_location_1.
22423 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
22425 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
22426 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
22427 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
22428 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
22429 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
22430 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
22431 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
22432 _mm_maskz_getmant_ss): New intrinsics.
22433 (__builtin_ia32_getexpss128_mask): Changed to ...
22434 __builtin_ia32_getexpss128_round ... this.
22435 (__builtin_ia32_getexpsd128_mask): Changed to ...
22436 __builtin_ia32_getexpsd128_round ... this.
22437 * config/i386/i386-builtin-types.def
22438 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
22439 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
22440 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
22441 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
22442 __builtin_ia32_getmantss_mask_round): New builtins.
22443 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
22444 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
22445 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
22446 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
22447 * config/i386/sse.md
22448 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
22449 avx512f_sgetexp<mode><mask_scalar_name>
22450 <round_saeonly_scalar_name> ... this.
22451 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
22452 %0, %1, %2<round_saeonly_op3>}): Changed to ...
22453 vgetexp<ssescalarmodesuffix>
22454 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22455 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
22456 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
22457 avx512f_vgetmant<mode><mask_scalar_name>
22458 <round_saeonly_scalar_name> ... this.
22459 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
22460 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
22461 vgetmant<ssescalarmodesuffix>
22462 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
22463 %0<mask_scalar_operand4>, %1, %2
22464 <round_saeonly_scalar_mask_op4>, %3} ... this.
22465 * config/i386/subst.md (mask_scalar_operand4,
22466 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
22467 round_saeonly_scalar_nimm_predicate): New subst attributes.
22469 2017-07-06 Julia Koval <julia.koval@intel.com>
22471 * config/i386/i386.c (ix86_erase_embedded_rounding):
22472 Remove code for old rounding pattern.
22474 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
22476 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
22478 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
22480 * doc/sourcebuild.texi (Test Directives, Variants of
22481 dg-require-support): Add documentation for dg-require-stack-check.
22483 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
22485 * config/i386/subst.md (mask_scalar, round_scalar,
22486 round_saeonly_scalar): New meta-templates.
22487 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
22488 round_scalar_mask_operand3, round_scalar_mask_op3,
22489 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
22490 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
22491 round_saeonly_scalar_constraint,
22492 round_saeonly_scalar_prefix): New subst attribute.
22493 * config/i386/sse.md
22494 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
22495 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
22496 <round_scalar_name> ... this.
22497 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
22498 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
22499 <round_scalar_name> ... this.
22500 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
22501 <sse>_vm<code><mode>3<mask_scalar_name>
22502 <round_saeonly_scalar_name> ... this.
22503 (v<plusminus_mnemonic><ssescalarmodesuffix>
22504 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
22505 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
22506 v<plusminus_mnemonic><ssescalarmodesuffix>
22507 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22508 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
22509 (v<multdiv_mnemonic><ssescalarmodesuffix>
22510 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
22511 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
22512 v<multdiv_mnemonic><ssescalarmodesuffix>
22513 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22514 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
22515 (v<maxmin_float><ssescalarmodesuffix>
22516 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
22517 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
22518 v<maxmin_float><ssescalarmodesuffix>
22519 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22520 %0<mask_scalar_operand3>, %1, %<iptr>2
22521 <round_saeonly_scalar_mask_op3>} ... this.
22523 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
22525 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
22526 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
22528 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
22529 Alan Hayward <alan.hayward@arm.com>
22530 David Sherwood <david.sherwood@arm.com>
22532 * combine.c (simplify_if_then_else): Remove "enum" before
22534 * compare-elim.c (can_eliminate_compare): Likewise.
22535 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
22537 (aarch64_lookup_simd_builtin_type): Likewise.
22538 (aarch64_simd_builtin_type): Likewise.
22539 (aarch64_init_simd_builtin_types): Likewise.
22540 (aarch64_simd_expand_args): Likewise.
22541 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
22543 (aarch64_reverse_mask): Likewise.
22544 (aarch64_simd_emit_reg_reg_move): Likewise.
22545 (aarch64_gen_adjusted_ldpstp): Likewise.
22546 (aarch64_ccmp_mode_to_code): Likewise.
22547 (aarch64_operands_ok_for_ldpstp): Likewise.
22548 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
22549 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
22551 (aarch64_min_divisions_for_recip_mul): Likewise.
22552 (aarch64_reassociation_width): Likewise.
22553 (aarch64_get_condition_code_1): Likewise.
22554 (aarch64_simd_emit_reg_reg_move): Likewise.
22555 (aarch64_simd_attr_length_rglist): Likewise.
22556 (aarch64_reverse_mask): Likewise.
22557 (aarch64_operands_ok_for_ldpstp): Likewise.
22558 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
22559 (aarch64_gen_adjusted_ldpstp): Likewise.
22560 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
22562 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
22563 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
22564 (arm_lookup_simd_builtin_type): Likewise.
22565 (arm_simd_builtin_type): Likewise.
22566 (arm_init_simd_builtin_types): Likewise.
22567 (arm_expand_builtin_args): Likewise.
22568 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
22569 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
22570 (ft32_setup_incoming_varargs): Likewise.
22571 (ft32_function_arg): Likewise.
22572 (ft32_function_arg_advance): Likewise.
22573 (ft32_pass_by_reference): Likewise.
22574 (ft32_arg_partial_bytes): Likewise.
22575 (ft32_valid_pointer_mode): Likewise.
22576 (ft32_addr_space_pointer_mode): Likewise.
22577 (ft32_addr_space_legitimate_address_p): Likewise.
22578 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
22580 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
22581 (ix86_emit_outlined_ms2sysv_restore): Likewise.
22582 (iamcu_alignment): Likewise.
22583 (canonicalize_vector_int_perm): Likewise.
22584 (ix86_noce_conversion_profitable_p): Likewise.
22585 (ix86_mpx_bound_mode): Likewise.
22586 (ix86_operands_ok_for_move_multiple): Likewise.
22587 * config/microblaze/microblaze-protos.h
22588 (microblaze_expand_conditional_branch_reg): Likewise.
22589 * config/microblaze/microblaze.c
22590 (microblaze_expand_conditional_branch_reg): Likewise.
22591 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
22593 (rs6000_reassociation_width): Likewise.
22594 (rs6000_invalid_binary_op): Likewise.
22595 (fusion_p9_p): Likewise.
22596 (emit_fusion_p9_load): Likewise.
22597 (emit_fusion_p9_store): Likewise.
22598 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
22600 (riscv_hard_regno_mode_ok_p): Likewise.
22601 (riscv_address_insns): Likewise.
22602 (riscv_split_symbol): Likewise.
22603 (riscv_legitimize_move): Likewise.
22604 (riscv_function_value): Likewise.
22605 (riscv_hard_regno_nregs): Likewise.
22606 (riscv_expand_builtin): Likewise.
22607 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
22608 (riscv_build_integer): Likewise.
22609 (riscv_split_integer): Likewise.
22610 (riscv_legitimate_constant_p): Likewise.
22611 (riscv_cannot_force_const_mem): Likewise.
22612 (riscv_regno_mode_ok_for_base_p): Likewise.
22613 (riscv_valid_base_register_p): Likewise.
22614 (riscv_valid_offset_p): Likewise.
22615 (riscv_valid_lo_sum_p): Likewise.
22616 (riscv_classify_address): Likewise.
22617 (riscv_legitimate_address_p): Likewise.
22618 (riscv_address_insns): Likewise.
22619 (riscv_load_store_insns): Likewise.
22620 (riscv_force_binary): Likewise.
22621 (riscv_split_symbol): Likewise.
22622 (riscv_force_address): Likewise.
22623 (riscv_legitimize_address): Likewise.
22624 (riscv_move_integer): Likewise.
22625 (riscv_legitimize_const_move): Likewise.
22626 (riscv_legitimize_move): Likewise.
22627 (riscv_address_cost): Likewise.
22628 (riscv_subword): Likewise.
22629 (riscv_output_move): Likewise.
22630 (riscv_canonicalize_int_order_test): Likewise.
22631 (riscv_emit_int_order_test): Likewise.
22632 (riscv_function_arg_boundary): Likewise.
22633 (riscv_pass_mode_in_fpr_p): Likewise.
22634 (riscv_pass_fpr_single): Likewise.
22635 (riscv_pass_fpr_pair): Likewise.
22636 (riscv_get_arg_info): Likewise.
22637 (riscv_function_arg): Likewise.
22638 (riscv_function_arg_advance): Likewise.
22639 (riscv_arg_partial_bytes): Likewise.
22640 (riscv_function_value): Likewise.
22641 (riscv_pass_by_reference): Likewise.
22642 (riscv_setup_incoming_varargs): Likewise.
22643 (riscv_print_operand): Likewise.
22644 (riscv_elf_select_rtx_section): Likewise.
22645 (riscv_save_restore_reg): Likewise.
22646 (riscv_for_each_saved_reg): Likewise.
22647 (riscv_register_move_cost): Likewise.
22648 (riscv_hard_regno_mode_ok_p): Likewise.
22649 (riscv_hard_regno_nregs): Likewise.
22650 (riscv_class_max_nregs): Likewise.
22651 (riscv_memory_move_cost): Likewise.
22652 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
22653 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
22654 (rl78_addr_space_address_mode): Likewise.
22655 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22657 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
22658 (rs6000_reassociation_width): Likewise.
22659 (rs6000_invalid_binary_op): Likewise.
22660 (fusion_p9_p): Likewise.
22661 (emit_fusion_p9_load): Likewise.
22662 (emit_fusion_p9_store): Likewise.
22663 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
22664 (ok_for_simple_move_operands): Likewise.
22665 (ok_for_simple_move_strict_operands): Likewise.
22666 (ok_for_simple_arith_logic_operands): Likewise.
22667 (visium_legitimize_reload_address): Likewise.
22668 (visium_select_cc_mode): Likewise.
22669 (output_cbranch): Likewise.
22670 (visium_split_double_move): Likewise.
22671 (visium_expand_copysign): Likewise.
22672 (visium_expand_int_cstore): Likewise.
22673 (visium_expand_fp_cstore): Likewise.
22674 * config/visium/visium.c (visium_pass_by_reference): Likewise.
22675 (visium_function_arg): Likewise.
22676 (visium_function_arg_advance): Likewise.
22677 (visium_libcall_value): Likewise.
22678 (visium_setup_incoming_varargs): Likewise.
22679 (visium_legitimate_constant_p): Likewise.
22680 (visium_legitimate_address_p): Likewise.
22681 (visium_legitimize_address): Likewise.
22682 (visium_secondary_reload): Likewise.
22683 (visium_register_move_cost): Likewise.
22684 (visium_memory_move_cost): Likewise.
22685 (prepare_move_operands): Likewise.
22686 (ok_for_simple_move_operands): Likewise.
22687 (ok_for_simple_move_strict_operands): Likewise.
22688 (ok_for_simple_arith_logic_operands): Likewise.
22689 (visium_function_value_1): Likewise.
22690 (rtx_ok_for_offset_p): Likewise.
22691 (visium_legitimize_reload_address): Likewise.
22692 (visium_split_double_move): Likewise.
22693 (visium_expand_copysign): Likewise.
22694 (visium_expand_int_cstore): Likewise.
22695 (visium_expand_fp_cstore): Likewise.
22696 (visium_split_cstore): Likewise.
22697 (visium_select_cc_mode): Likewise.
22698 (visium_split_cbranch): Likewise.
22699 (output_cbranch): Likewise.
22700 (visium_print_operand_address): Likewise.
22701 * expmed.c (flip_storage_order): Likewise.
22702 * expmed.h (emit_cstore): Likewise.
22703 (flip_storage_order): Likewise.
22704 * genrecog.c (validate_pattern): Likewise.
22705 * hsa-gen.c (gen_hsa_addr): Likewise.
22706 * internal-fn.c (expand_arith_overflow): Likewise.
22707 * ira-color.c (allocno_copy_cost_saving): Likewise.
22708 * lra-assigns.c (find_hard_regno_for_1): Likewise.
22709 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
22710 (process_invariant_for_inheritance): Likewise.
22711 * lra-eliminations.c (move_plus_up): Likewise.
22712 * omp-low.c (lower_oacc_reductions): Likewise.
22713 * simplify-rtx.c (simplify_subreg): Likewise.
22714 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
22715 (TARGET_CHKP_BOUND_MODE): Likewise..
22716 * targhooks.c (default_chkp_bound_mode): Likewise.
22717 (default_setup_incoming_vararg_bounds): Likewise.
22718 * targhooks.h (default_chkp_bound_mode): Likewise.
22719 (default_setup_incoming_vararg_bounds): Likewise.
22720 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
22721 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
22722 (have_whole_vector_shift): Likewise.
22723 * tree-vect-stmts.c (vectorizable_load): Likewise.
22724 * doc/tm.texi: Regenerate.
22726 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22728 Graceful degrade if Binutils PR21472 is not available.
22731 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
22732 .rodata in flash test fails.
22733 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
22734 * confgure: Regenerate.
22735 * config.in: Regenerate.
22736 * config/avr/avr.c (avr_asm_named_section)
22737 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
22738 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
22739 (avr_asm_init_sections): Same.
22741 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22743 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
22744 (fma<VH:mode>4_intrinsic): Likewise.
22745 (*fmsub<VCVTF:mode>4): Likewise.
22746 (*fmsub<VH:mode>4_intrinsic): Likewise.
22748 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22751 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
22752 Don't depend on "optimize > 0".
22753 (out_movhi_r_mr, out_movqi_mr_r): Same.
22754 (out_movhi_mr_r, out_movqi_r_mr): Same.
22755 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
22756 io_address_operand on "optimize > 0".
22758 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22760 * tree-loop-distribution.c: Add general explanantion on the pass.
22761 (generate_loops_for_partition): Mark distributed loop.
22762 (pg_add_dependence_edges): New parameter. Handle alias data
22763 dependence specially and record it in the parameter if asked.
22764 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
22765 (init_partition_graph_vertices, add_partition_graph_edge): New.
22766 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
22767 (free_partition_graph_vdata, build_partition_graph): New.
22768 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
22769 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
22770 (data_ref_segment_size, latch_dominated_by_data_ref): New.
22771 (compute_alias_check_pairs, version_loop_by_alias_check): New.
22772 (version_for_distribution_p, finalize_partitions): New.
22773 (distribute_loop): Handle alias data dependence specially. Factor
22774 out loop fusion code as functions and call these functions.
22776 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22778 * tree-loop-distribution.c (classify_partition): New parameter and
22779 better handle reduction statement.
22780 (rdg_build_partitions): Revise comment.
22781 (distribute_loop): Compute statements in all partitions and pass it
22782 to classify_partition.
22784 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22786 * tree-loop-distribution.c (enum partition_type): New.
22787 (struct partition): New field type.
22788 (partition_merge_into): Add parameter. Update partition type.
22789 (data_dep_in_cycle_p, update_type_for_merge): New functions.
22790 (build_rdg_partition_for_vertex): Compute partition type.
22791 (rdg_build_partitions): Dump partition type.
22792 (distribute_loop): Update calls to partition_merge_into.
22794 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22796 * tree-loop-distribution.c (struct ddr_hasher): New.
22797 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
22799 (classify_partition): Call get_data_dependence.
22800 (pg_add_dependence_edges): Ditto.
22801 (distribute_loop): Release data dependence hash table.
22803 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22805 * tree-loop-distribution.c (ref_base_address): Delete.
22806 (similar_memory_accesses): Rename ...
22807 (share_memory_accesses): ... to this. Check if partitions access
22808 the same memory reference.
22809 (distribute_loop): Call share_memory_accesses.
22811 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22813 * tree-loop-distribution.c (struct partition): New field recording
22814 its data reference.
22815 (partition_alloc, partition_free): Init and release data refs.
22816 (partition_merge_into): Merge data refs.
22817 (build_rdg_partition_for_vertex): Collect data refs for partition.
22818 (pg_add_dependence_edges): Change parameters from vector to bitmap.
22820 (distribute_loop): Remve data refs from vertice data of partition
22823 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22825 * tree-loop-distribution.c (params.h): Include header file.
22826 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
22827 (datarefs_vec): New global var.
22828 (create_rdg_vertices): Use datarefs_vec directly.
22829 (free_rdg): Don't free data references.
22830 (build_rdg): Update use. Don't free data references.
22831 (distribute_loop): Compute global variable for data references.
22832 Bail out if there are too many data references.
22834 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22836 * tree-loop-distribution.c (loop_nest): New global var.
22837 (build_rdg): Use loop directly, rather than loop nest.
22838 (pg_add_dependence_edges): Remove loop nest parameter. Use global
22840 (distribute_loop): Compute global variable loop nest. Update use.
22842 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22844 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
22845 (partition_merge_into): New parameter. Dump reason for fusion.
22846 (distribute_loop): Update use of partition_merge_into.
22848 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22850 * tree-loop-distribution.c (bb_top_order_index): New.
22851 (bb_top_order_index_size, bb_top_order_cmp): New.
22852 (stmts_from_loop): Use topological order.
22853 (pass_loop_distribution::execute): Compute and release topological
22854 order for basic blocks.
22856 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22858 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
22861 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22863 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
22864 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
22865 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
22866 * internal-fn.def (LOOP_DIST_ALIAS): New.
22867 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
22868 (fold_loop_internal_call): ... this.
22869 (vect_loop_dist_alias_call): New function.
22870 (set_uid_loop_bbs): Call fold_loop_internal_call.
22871 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
22874 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22877 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
22878 Require dead FLAGS_REG at the beginning of a peephole.
22880 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22883 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
22884 arguments in the call to __builtin_ia32_sbb_u32.
22885 (_subborrow_u64): Swap _X and _Y arguments in the call to
22886 __builtin_ia32_sbb_u64.
22888 2017-07-04 Jakub Jelinek <jakub@redhat.com>
22891 * tree-vrp.c (compare_assert_loc): Turn into a function template
22892 with stable template parameter. Only test if a->e is NULL,
22893 !a->e == !b->e has been verified already. Use e == NULL or
22894 e != NULL instead of e or ! e tests. If stable is true, don't use
22895 iterative_hash_expr, on the other side allow a or b or both NULL
22896 and sort the NULLs last.
22897 (process_assert_insertions): Sort using compare_assert_loc<false>
22898 instead of compare_assert_loc, later sort using
22899 compare_assert_loc<true> before calling process_assert_insertions_for
22900 in a loop. Use break instead of continue once seen NULL pointer.
22902 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22904 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
22905 Cortex-R7 and Cortex-R8 processors.
22907 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22909 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
22910 uninitialized while src is not.
22912 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
22914 * common/config/arm/arm-common.c: Adjust include path for
22916 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
22917 (arm-cpu.h): Create in build directory. Adjust dependency rules.
22918 (arm-cpu-data.h): Likewise.
22919 (arm-cpu-cdata.h): Likewise.
22920 * config/arm/arm-cpu.h: Delete.
22921 * config/arm/arm-cpu-cdata.h: Delete.
22922 * config/arm/arm-cpu-data.h: Delete.
22924 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
22926 * config/arm/arm-cpus.in (cortex-a55): New.
22927 (cortex-a75): Likewise.
22928 (cortex-a75.cortex-a55): Likewise.
22929 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
22931 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
22932 * config/arm/arm-cpu-cdata.h: Regenerate.
22933 * config/arm/arm-cpu-data.h: Regenerate.
22934 * config/arm/arm-cpu.h: Regenerate.
22935 * config/arm/arm-tables.opt: Regenerate.
22936 * config/arm/arm-tune.md: Regenerate.
22938 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22940 * haifa-sched.c (sched_create_recovery_edges): Update profile.
22942 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22944 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
22947 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
22949 PR tree-optimization/81292
22950 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
22951 full_string_p, also call adjust_related_strinfos if the adjustment
22952 is simple, otherwise invalidate related strinfos.
22954 2017-07-04 Martin Liska <mliska@suse.cz>
22957 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
22958 newly created variable as DECL_IGNORED_P.
22960 2017-07-04 Martin Liska <mliska@suse.cz>
22963 * ipa-inline.c (inline_small_functions):
22964 Use xstrdup_for_dump.
22966 2017-07-04 Tom de Vries <tom@codesourcery.com>
22968 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
22970 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22973 * config/darwin.c (darwin_function_switched_text_sections):
22976 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
22978 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
22980 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22982 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
22984 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22986 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
22987 min_profitable_iters, and th as inclusive lower bounds.
22988 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
22989 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
22990 for min_profitable_iters and min_profitable_estimate.
22991 (vect_transform_loop): Treat th as an inclusive lower bound.
22992 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22994 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22997 * config/darwin.c (darwin_function_switched_text_sections):
22998 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
22999 in two pieces, and suppress the use of buf.
23001 2017-07-03 Nathan Sidwell <nathan@acm.org>
23003 * hash-table.h (hash_table_mod1): Fix indentation.
23005 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23007 PR middle-end/81290
23008 * predict.c (force_edge_cold): Be more careful about propagation
23010 * profile-count.h (profile_probability::guessed,
23011 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
23013 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
23015 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
23017 * doc/invoke.texi (rcpc architecture extension): Document it.
23019 2017-07-03 Richard Biener <rguenther@suse.de>
23021 PR tree-optimization/60510
23022 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
23023 the scalar reduction PHI and use it.
23024 (vectorizable_reduction): Properly guard the single_defuse_cycle
23025 path for non-SLP reduction chains where we cannot use it.
23026 Rework reduc_def/index and vector type deduction. Rework
23027 vector operand gathering during reduction op code-gen.
23028 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
23029 chains dissolve the chain and leave it to non-SLP reduction
23032 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23034 * tree-data-ref.h (dr_alignment): Declare.
23035 * tree-data-ref.c (dr_alignment): New function.
23036 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
23037 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
23039 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
23041 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23043 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
23044 and base_misalignment fields.
23045 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
23046 * tree-data-ref.c: Include builtins.h.
23047 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
23048 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
23049 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
23050 * tree-vect-data-refs.c: Include tree-cfg.h.
23051 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
23052 fields instead of calculating an alignment here.
23053 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
23054 innermost_loop_behavior fields.
23056 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23058 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
23060 (DR_STEP_ALIGNMENT): New macro.
23061 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
23062 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
23063 (create_data_ref): Print it.
23064 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
23065 to tell whether the step preserves vector (mis)alignment.
23066 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
23067 Move the check for an integer step and generalise to all INTEGER_CST.
23068 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
23069 Print the outer step alignment.
23071 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23073 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
23074 with offset_alignment.
23075 (DR_ALIGNED_TO): Delete.
23076 (DR_OFFSET_ALIGNMENT): New macro.
23077 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
23078 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
23079 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
23080 (create_data_ref): Likewise.
23081 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
23082 (vect_analyze_data_refs): Likewise.
23083 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
23084 creating dummy innermost behavior.
23086 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23088 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
23089 with a "innermost_loop_behavior *" and refeence tree.
23090 * tree-data-ref.c (dr_analyze_innermost): Likewise.
23091 (create_data_ref): Update call accordingly.
23092 * tree-predcom.c (find_looparound_phi): Likewise.
23094 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23096 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
23097 fields with dr_wrt_vec_loop.
23098 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
23099 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
23100 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
23101 (vect_dr_behavior): New function.
23102 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
23103 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
23104 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
23105 track whether the step preserves the misalignment.
23106 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
23107 Use vect_dr_behavior.
23108 (vect_setup_realignment): Update call accordingly.
23109 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
23110 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
23111 call to vect_create_addr_base_for_vector_ref.
23112 (vect_create_cond_for_align_checks): Likewise.
23113 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
23114 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
23115 (vect_recog_mask_conversion_pattern): Likewise.
23116 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
23117 (new_stmt_vec_info): Remove redundant zeroing.
23119 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
23121 * common/config/arm/arm-common.c (arm_be8_option): New function.
23122 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
23123 (ISA_ARMv6): Add isa_bit_be8.
23124 * config/arm/arm.h (arm_be8_option): Add prototype.
23125 (BE8_SPEC_FUNCTION): New define.
23126 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
23127 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
23128 (mlittle-endian): Similarly.
23129 (mbe8, mbe32): New options.
23130 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
23131 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
23133 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23135 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
23137 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23139 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
23140 (cleanup_tree_cfg_bb): Use it.
23141 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
23143 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
23145 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23148 * loop-doloop.c (add_test): Update profile.
23150 2017-07-03 Martin Liska <mliska@suse.cz>
23153 * sanopt.c (rewrite_usage_of_param): New function.
23154 (sanitize_rewrite_addressable_params): Likewise.
23155 (pass_sanopt::execute): Call rewrite_usage_of_param.
23157 2017-07-03 Richard Biener <rguenther@suse.de>
23159 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
23160 back to using VIEW_CONVERT_EXPR.
23162 2017-07-03 Martin Liska <mliska@suse.cz>
23165 * doc/extend.texi: Document when a resolver function is
23166 generated for target_clones.
23168 2017-07-03 Martin Liska <mliska@suse.cz>
23170 * asan.c (asan_emit_stack_protection): Unpoison just red zones
23171 and shadow memory of auto variables which are subject of
23172 use-after-scope sanitization.
23173 (asan_expand_mark_ifn): Add do set only when is_poison.
23175 2016-07-03 Richard Biener <rguenther@suse.de>
23177 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
23179 (vect_force_simple_reduction): Record reduction def -> phi mapping.
23180 (vectorizable_reduction): Perform reduction PHI creation when
23181 visiting a reduction PHI and adjust and simplify code generation
23182 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
23183 (vect_transform_loop): Visit reduction PHIs.
23184 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
23185 defs into the SLP tree.
23186 (vect_build_slp_tree): Reduction defs terminate the recursion.
23187 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
23189 (vect_get_vec_defs_for_stmt_copy): Export.
23190 (vect_get_vec_defs): Likewise.
23191 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
23193 (vect_get_vec_defs_for_stmt_copy): Declare.
23194 (vect_get_vec_defs): Likewise.
23196 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23198 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
23199 parameter with a "loop" parameter and use it instead of the
23200 loop containing DR_STMT. Don't check simple_iv when doing
23201 BB analysis. Describe the two analysis modes in the comment.
23203 2017-07-03 Tom de Vries <tom@codesourcery.com>
23205 PR tree-optimization/69468
23206 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
23207 (find_same_succ_bb): Handle ignore_edge_flags.
23209 2017-07-03 Tom de Vries <tom@codesourcery.com>
23211 PR tree-optimization/81192
23212 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
23214 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
23216 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
23218 2017-07-03 Tom de Vries <tom@codesourcery.com>
23220 PR tree-optimization/81192
23221 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
23222 BB_SAME_SUCC (bb) == NULL.
23224 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23226 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
23229 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23231 * dumpfile.c: Include profile-count.h
23232 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
23234 (insert_cond_bb): Update profile.
23235 * tree-cfg.h (insert_cond_bb): Update prototype.
23236 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
23237 * tree-dump.c: Do not include tree-cfg.
23239 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23241 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
23243 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23245 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
23248 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23250 * tree-complex.c (expand_complex_div_wide): update profile.
23252 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23253 Alan Hayward <alan.hayward@arm.com>
23254 David Sherwood <david.sherwood@arm.com>
23256 * Makefile.in (MACHMODE_H): Remove insn-modes.h
23257 (CORETYPES_H): New define.
23258 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
23259 (insn-modes-inline.h, s-modes-inline-h): New rules.
23260 (generated_files): Add insn-modes-inline.h.
23261 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
23262 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
23263 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
23264 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
23265 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
23266 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
23267 (build/gencodes.o, build/genconditions.o): Likewise.
23268 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
23269 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
23270 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
23271 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
23272 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
23273 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
23274 * coretypes.h: Include everything up to real.h for generators.
23275 Include insn-modes.h first. Include wide-int-print.h after
23276 wide-int.h. Include insn-modes-inline.h and then machmode.h.
23277 * machmode.h: Don't include insn-modes.h here.
23278 * function-tests.c: Remove includes of signop.h, machmode.h,
23279 double-int.h and wide-int.h.
23281 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
23283 * optc-save-gen.awk: Likewise.
23284 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
23285 * godump.c: Remove include of wide-int-print.h.
23286 * pretty-print.h: Likewise.
23287 * wide-int-print.cc: Likewise.
23288 * wide-int.cc: Likewise.
23289 * hash-map-tests.c: Remove include of signop.h.
23290 * hash-set-tests.c: Likewise.
23291 * rtl-tests.c: Likewise.
23292 * mkconfig.sh: Remove include of machmode.h.
23293 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
23295 (emit_insn_modes_inline_h): ...this new function. Emit the code
23296 into an insn-modes-inline.h header file, adding appropriate
23297 include guards and end comments.
23298 (emit_insn_modes_c_header): Remove include of machmode.h.
23299 (emit_min_insn_modes_c_header): Include coretypes.h rather than
23301 (main): Handle -i flag and call emit_insn_modes_inline_h when
23304 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23306 * tree-ssa-strlen.c (strinfo): Rename the length field to
23307 nonzero_chars. Add a full_string_p field.
23308 (compare_nonzero_chars, zero_length_string_p): New functions.
23309 (get_addr_stridx): Add an offset_out parameter.
23310 Use compare_nonzero_chars.
23311 (get_stridx): Update accordingly. Use compare_nonzero_chars.
23312 (new_strinfo): Update after above changes to strinfo.
23313 (set_endptr_and_length): Set full_string_p.
23314 (get_string_length): Update after above changes to strinfo.
23315 (unshare_strinfo): Update call to new_strinfo.
23316 (maybe_invalidate): Likewise.
23317 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
23318 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
23319 as a uhwi instead of an shwi. Update after above changes to
23320 strinfo and new_strinfo.
23321 (zero_length_string): Assert that chainsi contains full strings.
23322 Use zero_length_string_p. Update call to new_strinfo.
23323 (adjust_related_strinfos): Update after above changes to strinfo.
23324 Copy full_string_p from origsi.
23325 (adjust_last_stmt): Use zero_length_string_p.
23326 (handle_builtin_strlen): Update after above changes to strinfo and
23327 new_strinfo. Install the lhs as the string length if the previous
23328 entry didn't describe a full string.
23329 (handle_builtin_strchr): Update after above changes to strinfo
23331 (handle_builtin_strcpy): Likewise.
23332 (handle_builtin_strcat): Likewise.
23333 (handle_builtin_malloc): Likewise.
23334 (handle_pointer_plus): Likewise.
23335 (handle_builtin_memcpy): Likewise. Track nonzero characters
23336 that aren't necessarily followed by a nul terminator.
23337 (handle_char_store): Likewise.
23339 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23341 PR tree-optimization/80769
23342 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
23343 for malloc and calloc. Document the new invariant that all related
23344 strinfos have delayed lengths or none do.
23345 (verify_related_strinfos): Move earlier in file.
23346 (set_endptr_and_length): New function, split out from...
23347 (get_string_length): ...here. Also set the lengths of related
23349 (zero_length_string): Assert that chainsi has known (rather than
23351 (adjust_related_strinfos): Likewise.
23353 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23355 PR tree-optimization/81136
23356 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
23357 assert that two references with the same misalignment have the same
23358 compile-time misalignment if those compile-time misalignments
23361 2017-07-01 Andi Kleen <ak@linux.intel.com>
23363 * print-tree.c (print_node): Print all attributes.
23365 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23367 * cfg.c (scale_bbs_frequencies): New function.
23368 * cfg.h (scale_bbs_frequencies): Declare it.
23369 * cfgloopanal.c (single_likely_exit): Cleanup.
23370 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
23372 (scale_loop_profile): Likewise.
23373 (loop_version): Likewise.
23374 (create_empty_loop_on_edge): Update.
23375 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
23376 scale_loop_frequencies, scale_loop_profile, loopify,
23377 loop_version): Update prototypes.
23378 * modulo-sched.c (sms_schedule): Update.
23379 * predict.c (unlikely_executed_edge_p): Also check probability.
23380 (probably_never_executed_edge_p): Fix typo.
23381 * tree-if-conv.c (version_loop_for_if_conversion): Update.
23382 * tree-parloops.c (gen_parallel_loop): Update.
23383 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
23384 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23385 * tree-ssa-loop-split.c (split_loop): Update.
23386 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
23387 * tree-vect-loop-manip.c (vect_do_peeling): Update.
23388 (vect_loop_versioning): Update.
23389 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23391 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23393 * trans-mem.c (split_bb_make_tm_edge): Update profile.
23395 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23397 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
23398 to keep profile consistent.
23400 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23402 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
23403 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
23404 * profile-count.h (max_safe_multiplier): Make unsigned.
23405 (profile_count::guessed_zero): New.
23407 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23409 * bb-reorder.c (fix_up_crossing_landing_pad,
23410 fix_crossing_conditional_branches): Use make_single_succ_edge
23411 to keep profile consistent.
23413 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23415 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
23418 2017-07-01 Jakub Jelinek <jakub@redhat.com>
23421 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
23422 the right scopes, make sure cond_jump isn't preserved between multiple
23423 iterations. Search for fallthru edge whenever there are 3+ edges and
23424 use find_fallthru_edge for it.
23426 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23428 Patch by Alexander Monakov <amonakov@ispras.ru>
23429 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
23430 probabilities consistently.
23432 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23434 * pa.c (pa_expand_compare_and_swap_loop): Update call of
23435 emit_cmp_and_jump_insns.
23437 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23440 * tree-inline.c (expand_call_inline): Combine profile statuses.
23442 2017-06-30 Andrew Pinski <apinski@cavium.com>
23444 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
23445 fold_stmt returned true.
23447 2017-06-30 Nathan Sidwell <nathan@acm.org>
23449 * ggc.h (empty_string): Delete.
23450 * cfgexpand.c (expand_asm_stmt): Use plain "".
23451 * optabs.c (expand_asm_memory_barrier): Likewise.
23452 * stringpool.c (empty_string): Delete.
23453 (digit_vector, digit_string): Delete.
23454 (ggc_alloc_string): Use plain "", don't optimize single digit
23455 strings. Use ggc_alloc_atomic.
23457 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
23459 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
23460 comparison set and one other set, use the cost of the non-comparison
23463 2017-06-30 Nathan Sidwell <nathan@acm.org>
23465 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
23468 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
23470 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
23471 loops. Remove now unneeded calls to gimple_switch_set_label() that
23472 just set removed labels to NULL_TREE.
23474 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
23476 * tree-ssanames.c (set_range_info_raw): Abstract from ...
23477 (set_range_info): ...here. Only call set_range_info_raw if domain
23479 (set_nonzero_bits): Call set_range_info_raw.
23480 * tree-ssanames.h (set_range_info_raw): New.
23482 2017-06-30 Jakub Jelinek <jakub@redhat.com>
23485 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
23486 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
23487 of nonimmediate_operand and <store_mask_constraint> instead of m
23488 for the input operand. For V8FI iterator, always split if input
23489 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
23490 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
23491 <store_mask_predicate> instead of register_operand and
23492 <store_mask_constraint> instead of v for the input operand. Make
23493 sure both operands aren't MEMs for if not <mask_applied>.
23495 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
23497 * lto-wrapper.c (copy_file) Close both file descriptors before
23500 2017-06-30 Martin Liska <mliska@suse.cz>
23503 * multiple_target.c (create_dispatcher_calls): Make ifunc
23504 also for function that don't have calls or are not referenced.
23506 2017-06-30 Richard Biener <rguenther@suse.de>
23508 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
23509 analyze the first scalar stmt. Move vector type computation
23510 for the BB case here from ...
23511 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
23512 live operation processing in the SLP case properly.
23514 2017-06-30 Richard Biener <rguenther@suse.de>
23516 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
23518 2017-06-30 Martin Liska <mliska@suse.cz>
23521 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
23522 before BUILT_IN_UNWIND_RESUME when ASAN is used.
23524 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
23526 * doc/invoke.texi (AArch64): Add missing options and remove redundant
23529 2017-06-30 Richard Biener <rguenther@suse.de>
23531 PR tree-optimization/81249
23532 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
23533 condition reduction result to original scalar type.
23535 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23537 * profile-count.h (enum profile_quality): Fix typos and whitespace
23540 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23542 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
23543 type for branch probabilities.
23545 2017-06-29 Julian Brown <julian@codesourcery.com>
23546 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23548 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
23549 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
23550 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
23551 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
23553 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23555 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
23556 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
23557 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
23558 CC usage from generic code to here.
23559 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
23560 CC usage into the target macros.
23562 2017-06-29 Maya Rashish <coypu@sdf.org>
23564 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
23567 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23569 * arm/arm-builtins.c: Include profile-count.h
23570 * except.c (sjlj_emit_function_enter): Use
23571 profile_probability::unlikely.
23573 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23575 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
23576 and tocrel_offset be pointer args rather than implicitly using
23578 (legitimate_constant_pool_address_p, rs6000_emit_move,
23579 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
23580 tocrel_offset and use in toc_relative_expr_p call.
23581 (print_operand, print_operand_address): Use static tocrel_base_oac
23582 and tocrel_offset_oac.
23583 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
23586 2017-06-29 Maya Rashish <coypu@sdf.org>
23588 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
23590 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
23592 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
23593 objects, take into account only the alignment of 'op0' and 'mode1' if
23596 2017-06-29 Steve Ellcey <sellcey@cavium.com>
23598 * ccmp.c (ccmp_tree_comparison_p): New function.
23599 (ccmp_candidate_p): Update to use above function.
23600 (get_compare_parts): New function.
23601 (expand_ccmp_next): Update to use new functions.
23602 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
23604 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
23605 take mode as argument.
23606 * ccmp.h (expand_ccmp_expr): Add mode as argument.
23607 * expr.c (expand_expr_real_1): Pass mode as argument.
23609 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
23611 * combine.c (combine_instructions): Print insns to dump_file, together
23614 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23616 * asan.c (asan_emit_stack_protection): Update.
23617 (create_cond_insert_point): Update.
23618 * auto-profile.c (afdo_propagate_circuit): Update.
23619 * basic-block.h (struct edge_def): Turn probability to
23620 profile_probability.
23621 (EDGE_FREQUENCY): Update.
23622 * bb-reorder.c (find_traces_1_round): Update.
23623 (better_edge_p): Update.
23624 (sanitize_hot_paths): Update.
23625 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
23626 (make_single_succ_edge): Update.
23627 (check_bb_profile): Update.
23628 (dump_edge_info): Update.
23629 (update_bb_profile_for_threading): Update.
23630 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
23631 probabilitycount to 0.
23632 * cfgbuild.c (compute_outgoing_frequencies): Update.
23633 * cfgcleanup.c (try_forward_edges): Update.
23634 (outgoing_edges_match): Update.
23635 (try_crossjump_to_edge): Update.
23636 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
23637 (expand_gimple_tailcall): Update.
23638 (construct_init_block): Use make_single_succ_edge.
23639 (construct_exit_block): Use make_single_succ_edge.
23640 * cfghooks.c (verify_flow_info): Update.
23641 (redirect_edge_succ_nodup): Update.
23642 (split_edge): Update.
23643 (account_profile_record): Update.
23644 * cfgloopanal.c (single_likely_exit): Update.
23645 * cfgloopmanip.c (scale_loop_profile): Update.
23646 (set_zero_probability): Remove.
23647 (duplicate_loop_to_header_edge): Update.
23648 * cfgloopmanip.h (loop_version): Update prototype.
23649 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
23650 (force_nonfallthru_and_redirect): Update.
23651 (update_br_prob_note): Update.
23652 (rtl_verify_edges): Update.
23653 (purge_dead_edges): Update.
23654 (rtl_lv_add_condition_to_bb): Update.
23655 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
23656 * cgraphunit.c (init_lowered_empty_function): Update.
23657 (cgraph_node::expand_thunk): Update.
23658 * cilk-common.c: Include profile-count.h
23659 * dojump.c (inv): Remove.
23660 (jumpifnot): Update.
23661 (jumpifnot_1): Update.
23662 (do_jump_1): Update.
23664 (do_jump_by_parts_greater_rtx): Update.
23665 (do_compare_rtx_and_jump): Update.
23666 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
23667 do_jump_1. do_compare_rtx_and_jump): Update prototype.
23668 * dwarf2cfi.c: Include profile-count.h
23669 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
23670 (sjlj_emit_dispatch_table): Likewise.
23671 * explow.c: Include profile-count.h
23672 * expmed.c (emit_store_flag_force): Update.
23673 (do_cmp_and_jump): Update.
23674 * expr.c (compare_by_pieces_d::generate): Update.
23675 (compare_by_pieces_d::finish_mode): Update.
23676 (emit_block_move_via_loop): Update.
23677 (store_expr_with_bounds): Update.
23678 (store_constructor): Update.
23679 (expand_expr_real_2): Update.
23680 (expand_expr_real_1): Update.
23681 * expr.h (try_casesi, try_tablejump): Update prototypes.
23682 * gimple-pretty-print.c (dump_probability): Update.
23683 (dump_profile): New.
23684 (dump_gimple_label): Update.
23685 (dump_gimple_bb_header): Update.
23686 * graph.c (draw_cfg_node_succ_edges): Update.
23687 * hsa-gen.c (convert_switch_statements): Update.
23688 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
23689 (find_if_case_1): Update.
23690 (find_if_case_2): Update.
23691 * internal-fn.c (expand_arith_overflow_result_store): Update.
23692 (expand_addsub_overflow): Update.
23693 (expand_neg_overflow): Update.
23694 (expand_mul_overflow): Update.
23695 (expand_vector_ubsan_overflow): Update.
23696 * ipa-cp.c (good_cloning_opportunity_p): Update.
23697 * ipa-split.c (split_function): Use make_single_succ_edge.
23698 * ipa-utils.c (ipa_merge_profiles): Update.
23699 * loop-doloop.c (add_test): Update.
23700 (doloop_modify): Update.
23701 * loop-unroll.c (compare_and_jump_seq): Update.
23702 (unroll_loop_runtime_iterations): Update.
23703 * lra-constraints.c (lra_inheritance): Update.
23704 * lto-streamer-in.c (input_cfg): Update.
23705 * lto-streamer-out.c (output_cfg): Update.
23706 * mcf.c (adjust_cfg_counts): Update.
23707 * modulo-sched.c (sms_schedule): Update.
23708 * omp-expand.c (expand_omp_for_init_counts): Update.
23709 (extract_omp_for_update_vars): Update.
23710 (expand_omp_ordered_sink): Update.
23711 (expand_omp_for_ordered_loops): Update.
23712 (expand_omp_for_generic): Update.
23713 (expand_omp_for_static_nochunk): Update.
23714 (expand_omp_for_static_chunk): Update.
23715 (expand_cilk_for): Update.
23716 (expand_omp_simd): Update.
23717 (expand_omp_taskloop_for_outer): Update.
23718 (expand_omp_taskloop_for_inner): Update.
23719 * omp-simd-clone.c (simd_clone_adjust): Update.
23720 * optabs.c (expand_doubleword_shift): Update.
23721 (expand_abs): Update.
23722 (emit_cmp_and_jump_insn_1): Update.
23723 (expand_compare_and_swap_loop): Update.
23724 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
23725 * predict.c (predictable_edge_p): Update.
23726 (edge_probability_reliable_p): Update.
23727 (set_even_probabilities): Update.
23728 (combine_predictions_for_insn): Update.
23729 (combine_predictions_for_bb): Update.
23730 (propagate_freq): Update.
23731 (estimate_bb_frequencies): Update.
23732 (force_edge_cold): Update.
23733 * profile-count.c (profile_count::dump): Add missing space into dump.
23734 (profile_count::debug): Add newline.
23735 (profile_count::differs_from_p): Explicitly convert to unsigned.
23736 (profile_count::stream_in): Update.
23737 (profile_probability::dump): New member function.
23738 (profile_probability::debug): New member function.
23739 (profile_probability::differs_from_p): New member function.
23740 (profile_probability::differs_lot_from_p): New member function.
23741 (profile_probability::stream_in): New member function.
23742 (profile_probability::stream_out): New member function.
23743 * profile-count.h (profile_count_quality): Rename to ...
23744 (profile_quality): ... this one.
23745 (profile_probability): New.
23746 (profile_count): Update.
23747 * profile.c (compute_branch_probabilities): Update.
23748 * recog.c (peep2_attempt): Update.
23749 * sched-ebb.c (schedule_ebbs): Update.
23750 * sched-rgn.c (find_single_block_region): Update.
23751 (compute_dom_prob_ps): Update.
23752 (schedule_region): Update.
23753 * sel-sched-ir.c (compute_succs_info): Update.
23754 * stmt.c (struct case_node): Update.
23755 (do_jump_if_equal): Update.
23756 (get_outgoing_edge_probs): Update.
23757 (conditional_probability): Update.
23758 (emit_case_dispatch_table): Update.
23759 (expand_case): Update.
23760 (expand_sjlj_dispatch_table): Update.
23761 (emit_case_nodes): Update.
23762 * targhooks.c: Update.
23763 * tracer.c (better_p): Update.
23764 (find_best_successor): Update.
23765 * trans-mem.c (expand_transaction): Update.
23766 * tree-call-cdce.c: Update.
23767 * tree-cfg.c (gimple_split_edge): Upate.
23768 (move_sese_region_to_fn): Upate.
23769 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
23770 * tree-eh.c (lower_resx): Upate.
23771 (cleanup_empty_eh_move_lp): Upate.
23772 * tree-if-conv.c (version_loop_for_if_conversion): Update.
23773 * tree-inline.c (copy_edges_for_bb): Update.
23774 (copy_cfg_body): Update.
23775 * tree-parloops.c (gen_parallel_loop): Update.
23776 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
23777 (gimple_gen_time_profiler): Update.
23778 * tree-ssa-dce.c (remove_dead_stmt): Update.
23779 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23780 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
23781 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
23782 (unloop_loops): Update.
23783 (try_peel_loop): Update.
23784 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23785 * tree-ssa-loop-split.c (connect_loops): Update.
23786 (split_loop): Update.
23787 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
23788 (hoist_guard): Update.
23789 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
23790 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
23791 (value_replacement): Update.
23792 * tree-ssa-reassoc.c (branch_fixup): Update.
23793 * tree-ssa-tail-merge.c (replace_block_by): Update.
23794 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
23795 (create_edge_and_update_destination_phis): Update.
23796 (compute_path_counts): Update.
23797 (recompute_probabilities): Update.
23798 (update_joiner_offpath_counts): Update.
23799 (freqs_to_counts_path): Update.
23800 (duplicate_thread_path): Update.
23801 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
23802 (struct switch_conv_info): Update.
23803 (gen_inbound_check): Update.
23804 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23805 (vect_do_peeling): Update.
23806 (vect_loop_versioning): Update.
23807 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23808 (optimize_mask_stores): Update.
23809 * ubsan.c (ubsan_expand_null_ifn): Update.
23810 * value-prof.c (gimple_divmod_fixed_value): Update.
23811 (gimple_divmod_fixed_value_transform): Update.
23812 (gimple_mod_pow2): Update.
23813 (gimple_mod_pow2_value_transform): Update.
23814 (gimple_mod_subtract): Update.
23815 (gimple_mod_subtract_transform): Update.
23816 (gimple_ic): Update.
23817 (gimple_stringop_fixed_value): Update.
23818 (gimple_stringops_transform): Update.
23819 * value-prof.h: Update.
23821 2017-06-29 Carl Love <cel@us.ibm.com>
23823 * config/rs6000/rs6000-c.c: Add support for built-in functions
23824 vector signed int vec_signed (vector float);
23825 vector signed long long vec_signed (vector double);
23826 vector signed int vec_signed2 (vector double, vector double);
23827 vector signed int vec_signede (vector double);
23828 vector signed int vec_signedo (vector double);
23829 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
23830 instruction generator.
23831 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23832 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
23833 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
23835 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
23836 vunsignede_v2df): Add define_expands.
23837 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
23838 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
23839 VEC_UNSIGNEDO): Add definitions.
23840 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23841 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
23842 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
23843 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
23844 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
23845 * config/rs6000/altivec.h (vec_signed, vec_signed2,
23846 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
23847 vec_unsignede, vec_unsignedo): Add builtin defines.
23848 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
23850 * doc/extend.texi: Update the built-in documentation file for the
23851 new built-in functions.
23853 2017-06-29 Richard Biener <rguenther@suse.de>
23855 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
23856 reduction chains to LOOP_VINFO_REDUCTIONS.
23857 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
23858 SLP reductions after processing reduction chains.
23860 2017-06-29 Nathan Sidwell <nathan@acm.org>
23862 * builtins.c (fold_builtin_FUNCTION): Use
23863 lang_hooks.decl_printable_name.
23865 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
23867 PR middle-end/81194
23868 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
23869 with only one label.
23870 * stmt.c (expand_case): Assert NCASES is greater than one.
23872 2017-06-29 Richard Biener <rguenther@suse.de>
23874 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
23876 (group_case_labels): Likewise.
23877 (find_taken_edge): Push sanity checking on val to workers...
23878 (find_taken_edge_cond_expr): ... here
23879 (find_taken_edge_switch_expr): ... and here, handle cases
23880 with just a default label.
23881 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
23882 (group_case_labels): Likewise.
23883 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
23884 group_case_labels does anything cleanup the CFG again.
23886 2017-06-29 Bin Cheng <bin.cheng@arm.com>
23888 PR tree-optimization/81196
23889 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
23890 exit condition comparing two IVs.
23892 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
23894 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
23895 profile to the dummy entry at the end of the list of architectures.
23896 * config/arm/arm-cpu-cdata.h: Regenerated.
23898 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23899 Michael Collison <michael.collison@arm.com>
23902 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
23904 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
23905 (*aarch64_reg_<mode>3_minus_mask): New pattern.
23906 (*aarch64_<optab>_reg_di3_mask2): New pattern.
23907 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
23908 of shift when the shift amount is masked with constant equal to
23909 the size of the mode.
23910 * config/aarch64/predicates.md (subreg_lowpart_operator): New
23913 2017-06-29 Martin Liska <mliska@suse.cz>
23915 * config/i386/i386.opt: Change range from [1,5] to [0,5].
23917 2017-06-29 Yury Gribov <tetra2005@gmail.com>
23920 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
23923 (edge_growth_cache_entry::edge_growth_cache_entry): New
23925 (reset_edge_growth_cache): Update to use constructor.
23927 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23929 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
23930 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23931 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23933 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
23935 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
23936 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
23938 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
23940 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
23941 (*-linux-uclibc*): Add t-uclibc tmake_file.
23942 * config/t-musl: New.
23943 * config/t-uclibc: New.
23945 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
23947 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
23949 (gen_comm_data): Emit architectural setting of arch_prof.
23950 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
23952 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
23953 (armv8-m.base, armv8-m.main): Likewise.
23954 * arm-protos.h (arm_build_target): Add profile field.
23955 (arch_option): Likewise.
23956 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
23958 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
23959 arm_active_target.profile.
23961 2017-06-28 Richard Biener <rguenther@suse.de>
23963 PR middle-end/81227
23964 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
23965 TYPE_OVERFLOW_WRAPS.
23966 * match.pd (negate_expr_p): Likewise.
23967 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
23968 fold_build2, not fold_binary.
23970 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23972 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
23973 Convert memory address to Pmode.
23974 (aarch64_print_operand): Assert MEM operands are always Pmode.
23976 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23979 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
23980 Remove redundant if.
23981 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
23982 * config/arm/aarch-common-protos.h
23983 (aarch_forward_to_shift_is_not_shifted_re): Remove.
23984 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
23986 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
23989 * multiple_target.c (create_dispatcher_calls): Set the default
23990 clone to be static, not public.
23992 2017-06-28 Richard Biener <rguenther@suse.de>
23994 * tree-vect-loop.c (vectorizable_reduction): Move special
23995 cond reduction IV var creation ...
23996 (vect_create_epilog_for_reduction): ... here. Remove induction_index
23997 parameter. Use STMT_VINFO_VECTYPE.
23998 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
24001 2017-06-28 Martin Liska <mliska@suse.cz>
24004 * ipa-visibility.c (non_local_p): Handle visibility.
24006 2017-06-28 Martin Liska <mliska@suse.cz>
24009 * common.opt: Add IntegerRange to various options.
24010 * opt-functions.awk (integer_range_info): New function.
24011 * optc-gen.awk: Add integer_range_info to cl_options struct.
24012 * opts-common.c (decode_cmdline_option): Handle
24013 CL_ERR_INT_RANGE_ARG.
24014 (cmdline_handle_error): Likewise.
24015 * opts.c (print_filtered_help): Show valid interval in
24016 when --help is provided.
24017 * opts.h (struct cl_option): Add range_min and range_max fields.
24018 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
24020 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
24022 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
24023 (x * C EQ/NE y * C): New transformation.
24025 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
24027 * genmultilib (combination_space): Accept '+' in option names.
24029 2017-06-28 Martin Liska <mliska@suse.cz>
24032 * asan.c (instrument_derefs): Bail out inner references
24033 that are hard register variables.
24035 2017-06-28 Jakub Jelinek <jakub@redhat.com>
24038 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
24039 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
24041 2017-06-28 Richard Biener <rguenther@suse.de>
24043 * tree-vectorizer.h (vect_get_vec_defs): Remove.
24044 (vect_get_slp_defs): Adjust.
24045 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
24047 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
24049 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
24050 get_initial_defs_for_reduction instead of vect_get_vec_defs.
24051 (vectorizable_reduction): Adjust.
24052 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
24054 (vect_get_slp_defs): Likewise.
24055 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
24056 (vectorizable_bswap): Adjust.
24057 (vectorizable_call): Likewise.
24058 (vectorizable_conversion): Likewise.
24059 (vectorizable_assignment): Likewise.
24060 (vectorizable_shift): Likewise.
24061 (vectorizable_operation): Likewise.
24062 (vectorizable_store): Likewise.
24063 (vectorizable_condition): Likewise.
24064 (vectorizable_comparison): Likewise.
24066 2017-06-28 Michael Collison <michael.collison@arm.com>
24069 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
24071 (arm_gen_movmemqi): Removed unused variable 'i'.
24072 Convert 'for' loop into 'while' loop.
24073 (arm_expand_prologue): Remove last unnecessary set of insn.
24074 (thumb_pop): Remove unused variable 'pushed_words'.
24075 (thumb_exit): Remove last unnecessary set of regs_to_pop.
24077 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24079 * config/s390/predicates.md: Use s390_rel_address_ok_p.
24080 * config/s390/s390-protos.h: Add prototype of
24081 s390_rel_address_ok_p.
24082 * config/s390/s390.c (s390_got_symbol): New function.
24083 (s390_rel_address_ok_p): New function.
24084 (legitimize_pic_address): Use s390_rel_address_ok_p.
24085 (s390_load_got): Use s390_got_symbol.
24086 (s390_option_override): Issue error if
24087 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
24088 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
24090 * config/s390/s390.opt: New option mpic-data-is-text-relative.
24092 2017-06-27 Andrew Pinski <apinski@cavium.com>
24094 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
24095 (X * copysign (1.0, X)): New pattern.
24096 (X * copysign (1.0, -X)): New pattern.
24097 (copysign (-1.0, CST)): New pattern.
24099 2017-06-27 Joseph Myers <joseph@codesourcery.com>
24101 * genmultilib (combination_space): Remove variable.
24102 Validate reuse rules against regular expression for any sequence
24103 of multilib options in any order.
24105 2017-06-27 Michael Collison <michael.collison@arm.com>
24107 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
24108 call aarch64_split_simd_combine.
24109 * (aarch64_combine_internal<mode>): Delete pattern.
24110 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
24111 Allow register and subreg operands.
24113 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24115 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
24116 specific need, just fallback on defaults.
24117 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
24119 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24120 Olivier Hainque <hainque@adacore.com>
24122 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
24124 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
24125 targets. Pick a default if no particular attempt applied.
24126 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
24129 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24131 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
24132 (x86_64-wrs-vxworks7): Likewise.
24134 2017-06-27 Marek Polacek <polacek@redhat.com>
24137 * ubsan.c (instrument_null): Check get_base_address's result for null.
24139 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
24141 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
24143 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
24145 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
24146 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
24147 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
24148 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
24149 New function types.
24150 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
24151 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
24152 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
24153 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
24154 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
24155 BUILT_IN_FEUPDATEENV): New builtins.
24156 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
24157 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
24158 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
24159 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
24161 (builtin_structptr_types): Adjust size.
24162 * tree.c (builtin_structptr_types): Add four entries.
24164 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24165 Olivier Hainque <hainque@adacore.com>
24167 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
24168 (TLS_SYM): New local macro, forcing reference to __tls__ on
24169 link command lines for VxWorks 7 RTPs, triggering initialization
24171 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
24172 OS features TLS support, true for RTPs on VxWorks 7.
24173 * config/vxworks.c (vxworks_override_options): Setup emutls
24176 2017-06-27 Jakub Jelinek <jakub@redhat.com>
24178 * predict.c (test_prediction_value_range): Use -1U instead of -1
24179 to avoid narrowing conversion warning.
24180 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
24181 to avoid narrowing conversion warning.
24182 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
24184 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
24186 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24188 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
24189 64bit configurations.
24190 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
24191 (SIZE_TYPE): Likewise.
24192 * config/vxworks.c (vxworks_emutls_var_fields): Use
24193 long_unsigned_type_node instead of unsigned_type_node as the offset
24194 field type, which is "pointer" mode in emutls.c.
24196 2017-06-27 Jakub Jelinek <jakub@redhat.com>
24199 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
24201 PR middle-end/81207
24202 * gimple-fold.c (replace_call_with_call_and_fold): Handle
24203 gimple_vuse copying separately from gimple_vdef copying.
24205 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24207 * value-prof.c (free_hist): Remove call to memset and the enclosing if
24210 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
24211 Olivier Hainque <hainque@adacore.com>
24213 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
24214 for all vxworks7 targets.
24215 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
24216 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
24217 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
24218 variations for VX6/VX7 and 32/64bits later on in ...
24219 (VXWORKS_LIB_SPEC): Leverage new macros.
24220 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
24221 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
24223 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
24225 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
24226 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
24228 2017-06-26 Carl Love <cel@us.ibm.com>
24230 * config/rs6000/rs6000-c.c: Add support for built-in functions
24231 vector bool char vec_reve (vector bool char);
24232 vector signed char vec_reve (vector signed char);
24233 vector unsigned char vec_reve (vector unsigned char);
24234 vector bool int vec_reve (vector bool int);
24235 vector signed int vec_reve (vector signed int);
24236 vector unsigned int vec_reve (vector unsigned int);
24237 vector bool long long vec_reve (vector bool long long);
24238 vector signed long long vec_reve (vector signed long long);
24239 vector unsigned long long vec_reve (vector unsigned long long);
24240 vector bool short vec_reve (vector bool short);
24241 vector signed short vec_reve (vector signed short);
24242 vector double vec_reve (vector double);
24243 vector float vec_reve (vector float);
24244 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
24245 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
24246 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
24247 (altivec_vreve): New pattern.
24248 * config/rs6000/altivec.h (vec_reve): New define.
24249 * doc/extend.texi (vec_rev): Update the built-in documentation file
24250 for the new built-in functions.
24252 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24254 PR tree-optimization/71815
24255 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
24257 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
24259 (slsr_process_phi): Likewise.
24260 (replace_uncond_cands_and_profitable_phis): Don't replace a
24261 multiply candidate with a stride of 1 (copy or cast).
24262 (phi_incr_cost): Call uses_consumed_by_stmt rather than
24264 (lowest_cost_path): Likewise.
24265 (total_savings): Likewise.
24267 2017-06-26 Richard Biener <rguenther@suse.de>
24270 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
24271 Use def_builtin_pure for all gather builtins.
24273 2017-06-26 Richard Biener <rguenther@suse.de>
24275 PR tree-optimization/81203
24276 * tree-tailcall.c (find_tail_calls): Do not move stmts into
24277 non-dominating BBs.
24279 2017-06-26 Marek Polacek <polacek@redhat.com>
24282 * doc/invoke.texi: Document -Wmultistatement-macros.
24284 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
24286 * doc/sourcebuild.texi (ARM-specific attributes): Document new
24287 arm_neon_ok_no_float_abi effective target.
24289 2017-06-26 Richard Biener <rguenther@suse.de>
24291 PR tree-optimization/80928
24292 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
24293 (copy_bbs): Set BB_DUPLICATED flag early.
24294 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
24296 (execute_on_shrinking_pred): Likewise.
24297 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
24298 BB_DUPLICATED blocks.
24299 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
24300 iterate over all PHIs considering removal of *gsi.
24302 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
24304 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
24307 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24309 * config/rs6000/rs6000-string.c: (expand_block_clear,
24310 do_load_for_compare, select_block_compare_mode,
24311 compute_current_alignment, expand_block_compare,
24312 expand_strncmp_align_check, expand_strn_compare,
24313 expand_block_move, rs6000_output_load_multiple)
24314 Move functions related to string/block move/compare
24315 to a separate file.
24316 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
24317 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
24318 for this function which is now used in two files.
24319 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
24320 * config.gcc: Add rs6000-string.o to extra_objs for
24321 targets powerpc*-*-* and rs6000*-*-*.
24323 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
24326 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
24327 32-bit, since indexed is not valid for DImode.
24328 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
24329 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
24330 (define_peephole2 for Altivec d-form load): Add 32-bit support.
24331 (define_peephole2 for Altivec d-form store): Likewise.
24334 * multiple_target.c (create_dispatcher_calls): Only create the
24335 dispatcher call if the function is the default clone of a
24336 versioned function.
24338 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
24340 PR middle-end/80902
24341 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
24342 a call, force the call to not be a tail call.
24344 2017-06-23 Jeff Law <law@redhat.com>
24346 * doc/contrib.texi: Add entry for Steven Pemberton's work on
24349 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
24351 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
24352 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
24353 handling for early expansion of vector shifts (sl,sr,sra,rl).
24354 (builtin_function_type): Add vector shift right instructions
24355 to the unsigned argument list.
24357 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
24359 rtl-optimizatoin/79286
24360 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
24361 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
24362 trap. PIC register plus a const unspec without offset can never trap.
24364 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
24366 * tree.h (builtin_structptr_type): New type.
24367 (builtin_structptr_types): Declare new array.
24368 * tree.c (builtin_structptr_types): New array.
24369 (free_lang_data, build_common_tree_nodes): Use it.
24371 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
24374 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
24377 2017-06-22 Matt Turner <mattst88@gmail.com>
24379 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
24380 Lake models to skylake case. Assume skylake for unknown
24381 models with clflushopt.
24383 2017-06-22 Jeff Law <law@redhat.com>
24385 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
24386 frame sizes that do not satisfy aarch64_uimm12_shift.
24388 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
24390 * profile-count.h (apply_probability,
24391 apply_scale, probability_in): Fix checks for zero.
24393 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24395 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
24396 * doc/cppdiropts.texi (-I @var{dir}): Document it.
24398 2016-06-22 Richard Biener <rguenther@suse.de>
24400 * tree-vect-loop.c (vect_model_reduction_cost): Handle
24401 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
24402 REDUC_MAX_EXPR support.
24403 (vectorizable_reduction): Likewise.
24404 (vect_create_epilog_for_reduction): Likewise.
24406 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
24408 * match.pd (A / (1 << B) -> A >> B): New.
24409 * generic-match-head.c: Include optabs-tree.h.
24410 * gimple-match-head.c: Likewise.
24411 * optabs-tree.h (target_supports_op_p): New.
24412 * optabs-tree.c (target_supports_op_p): New.
24414 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24416 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
24417 $gcc_cv_ld --help output.
24418 (gcc_cv_ld_demangle): Likewise.
24419 (gcc_cv_ld_eh_frame_hdr): Likewise.
24420 (gcc_cv_ld_pie): Likewise.
24421 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
24422 (gcc_cv_ld_buildid): Likewise.
24423 (gcc_cv_ld_sysroot): Likewise.
24424 (ld_bndplt_support): Likewise.
24425 (ld_pushpopstate_support): Likewise.
24426 * configure: Regenerate.
24427 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
24429 2017-06-21 Jakub Jelinek <jakub@redhat.com>
24432 * config/i386/sse.md (round<mode>2): Renumber match_dup and
24433 operands indexes to avoid gap between operands and match_dups.
24435 2017-06-21 Andrew Pinski <apinski@cavium.com>
24437 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
24438 Increment Arith_shift and Arith_shift_reg by 1.
24439 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
24441 * config/aarch64/aarch64.c (thunderx_tunings): Enable
24442 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
24443 (aarch64_strip_extend): Add new argument and test for it.
24444 (aarch64_cheap_mult_shift_p): New function.
24445 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
24446 add a cost if it is true.
24447 Update calls to aarch64_strip_extend.
24448 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
24450 2017-06-21 Andrew Pinski <apinski@cavium.com>
24452 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
24454 (thunderxt88): Likewise.
24455 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
24456 (thunderx_prefetch_tune): New variable.
24457 (thunderx2t99_prefetch_tune): Update for the correct values.
24458 (thunderxt88_tunings): New variable.
24459 (thunderx_tunings): Use thunderx_prefetch_tune instead of
24460 generic_prefetch_tune.
24461 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
24463 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24465 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
24466 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
24467 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
24468 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
24469 (aarch64_atomic_cas<mode>, GPI): Likewise.
24471 2017-06-21 Martin Liska <mliska@suse.cz>
24473 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
24474 statements on cold and hot labels.
24475 * predict.c (tree_estimate_probability_bb): Remove the
24476 prediction from this place.
24478 2017-06-21 Martin Liska <mliska@suse.cz>
24480 PR tree-optimization/79489
24481 * gimplify.c (maybe_add_early_return_predict_stmt): New
24483 (gimplify_return_expr): Call the function.
24484 * predict.c (tree_estimate_probability_bb): Remove handling
24486 * predict.def: Update comment about early return predictor.
24487 * gimple-predict.h (is_gimple_predict): New function.
24488 * predict.def: Change default value of early return to 66.
24489 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
24491 * passes.def: Put pass_strip_predict_hints to the beginning of
24494 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
24496 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
24497 FUNCTION_DECL declarations.
24498 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
24500 (dwaf2out_decl): Likewise.
24501 * godump.c (go_early_global_decl): Skip call to the real debug hook
24502 for FUNCTION_DECL declarations.
24503 * passes.c (rest_of_decl_compilation): Skip call to the
24504 early_global_decl debug hook for FUNCTION_DECL declarations, unless
24505 -fdump-go-spec is passed.
24507 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
24509 * config/i386/i386.c (struct builtin_isa): New field pure_p.
24510 Reorder for compactness.
24511 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
24512 (def_builtin_pure, def_builtin_pure2): New functions.
24513 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
24515 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
24517 * match.pd (nop_convert): New predicate.
24518 ((A +- CST1) +- CST2): Allow some NOP conversions.
24520 2017-06-21 Jakub Jelinek <jakub@redhat.com>
24523 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
24524 with ctors/dtors if GOVD_SHARED is set.
24526 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24528 * config/aarch64/aarch64.md (movti_aarch64):
24529 Emit mov rather than orr.
24530 (movtf_aarch64): Likewise.
24531 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
24532 Emit mov rather than orr.
24534 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24536 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
24537 Swap alternatives, make integer dup more expensive.
24539 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24541 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
24542 Return true for non-tls symbols.
24544 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
24546 * config/aarch64/aarch64-cores.def (cortex-a55): New.
24547 (cortex-a75): Likewise.
24548 (cortex-a75.cortex-a55): Likewise.
24549 * config/aarch64/aarch64-tune.md: Regenerate.
24550 * doc/invoke.texi (-mtune): Document new values for -mtune.
24552 2017-06-21 Tom de Vries <tom@codesourcery.com>
24554 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
24555 stack_size feature.
24556 (Effective-Target Keywords, Other attributes): Suggest using
24557 dg-add-options stack_size feature to get stack limit in stack_size
24558 effective target documentation.
24560 2017-06-21 Julian Brown <julian@codesourcery.com>
24561 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24563 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
24564 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
24565 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
24567 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
24568 attribute type list for neon_multiply.
24569 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
24570 attribute type list for neon_multiply.
24571 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
24572 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
24573 attribute type list for neon_multiply.
24574 * config/arm/types.md (crypto_pmull): Add.
24575 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
24576 attribute type list.
24578 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
24580 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
24583 2017-06-20 Jakub Jelinek <jakub@redhat.com>
24585 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
24586 to make sure not to dereference a NULL cost_classes_ptr pointer.
24588 2017-06-20 Carl Love <cel@us.ibm.com>
24590 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24591 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
24592 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
24593 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
24594 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
24595 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
24596 VMULOSW): New enum "unspec" values.
24597 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
24598 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
24599 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
24600 altivec_vmulosw): New patterns.
24601 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
24602 VMULOSW): Add definitions.
24604 2017-06-20 Julia Koval <julia.koval@intel.com>
24606 * config/i386/i386.c: Fix rounding expand for new pattern.
24607 * config/i386/subst.md: Fix pattern (parallel -> unspec).
24609 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24611 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
24612 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
24614 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24616 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
24619 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24621 * config/aarch64/aarch64-cores.def: Rearrange to sort by
24622 architecture, then by implementer ID.
24623 * config/aarch64/aarch64-tune.md: Regenerate.
24625 2017-06-20 Richard Biener <rguenther@suse.de>
24627 PR middle-end/81097
24628 * fold-const.c (split_tree): Fold to type before negating.
24630 2017-06-20 David Malcolm <dmalcolm@redhat.com>
24632 * diagnostic-show-locus.c
24633 (selftest::test_fixit_deletion_affecting_newline): New function.
24634 (selftest::diagnostic_show_locus_c_tests): Call it.
24636 2017-06-20 Andreas Schwab <schwab@suse.de>
24639 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
24642 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24644 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
24645 __ARM_FEATURE_COPROC according to support.
24647 2017-06-20 Jakub Jelinek <jakub@redhat.com>
24649 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
24650 Rewritten to avoid overflow for > 32-bit pointers.
24653 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
24654 by removing enum keyword.
24655 (ubsan_type_descriptor): Likewise. Formatting fix.
24658 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
24659 splitter): Require TARGET_SSE2 in the condition.
24661 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
24664 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
24665 for doing vector set of SFmode on ISA 3.0.
24666 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
24667 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
24669 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
24670 SFmode value into a V4SF variable that was extracted from another
24671 V4SF variable without converting the element to double precision
24672 and back to single precision vector format.
24673 (vsx_insert_extract_v4sf_p9_2): Likewise.
24675 2017-06-19 Jakub Jelinek <jakub@redhat.com>
24677 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
24678 in UWHI to avoid undefined overflow.
24681 * ubsan.h (enum ubsan_encode_value_phase): New.
24682 (ubsan_encode_value): Change second argument to
24683 enum ubsan_encode_value_phase with default value of
24684 UBSAN_ENCODE_VALUE_GENERIC.
24685 * ubsan.c (ubsan_encode_value): Change second argument to
24686 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
24687 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
24688 create_tmp_var_raw instead of create_tmp_var and use a
24690 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
24691 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
24692 ubsan_encode_value callers.
24695 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
24696 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
24697 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
24699 2017-06-19 Richard Biener <rguenther@suse.de>
24701 PR middle-end/81118
24702 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
24703 estimates if we changed anything.
24705 2017-06-19 Richard Biener <rguenther@suse.de>
24707 PR tree-optimization/80887
24708 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
24709 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
24710 simplified lookups, then reset mprts_hook.
24711 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
24713 (try_to_simplify): Likewise.
24715 2017-06-19 Martin Liska <mliska@suse.cz>
24718 * gimplify.c (gimplify_switch_expr):
24719 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
24721 2017-06-19 Martin Liska <mliska@suse.cz>
24723 * doc/install.texi: Document that PGO runs in 4 stages.
24725 2017-06-19 Martin Liska <mliska@suse.cz>
24728 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
24729 to dispatcher function name.
24730 * multiple_target.c (replace_function_decl): New function.
24731 (create_dispatcher_calls): Redirect both edges and references.
24733 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
24735 * profile-count.c (profile_count::dump): Dump quality.
24736 (profile_count::differs_from_p): Update for unsigned val.
24737 * profile-count.h (profile_count_quality): New enum.
24738 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
24740 2017-06-19 Richard Biener <rguenther@suse.de>
24742 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
24743 struct function as arg.
24744 (estimate_numbers_of_iterations): Export overload with loop arg.
24745 (free_numbers_of_iterations_estimates_loop): Use an overload of
24746 free_numbers_of_iterations_estimates instead.
24747 * tree-cfg.c (remove_bb): Adjust.
24748 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
24749 * tree-parloops.c (gen_parallel_loop): Likewise.
24750 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
24752 (tree_unroll_loops_completely): Likewise.
24753 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
24754 Use an overload instead and export.
24755 (estimated_loop_iterations): Adjust.
24756 (max_loop_iterations): Likewise.
24757 (likely_max_loop_iterations): Likewise.
24758 (estimate_numbers_of_iterations): Take struct function as arg
24760 (loop_exits_before_overflow): Adjust.
24761 (free_numbers_of_iterations_estimates_loop): Use an overload.
24762 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
24763 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
24765 2017-06-19 Richard Biener <rguenther@suse.de>
24768 * ipa-prop.c (find_constructor_constant_at_offset): Handle
24769 RANGE_EXPR conservatively.
24771 2017-06-16 Carl Love <cel@us.ibm.com>
24773 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24774 definitions for vec_float, vec_float2, vec_floato,
24775 vec_floate built-ins.
24776 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
24777 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
24779 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
24780 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
24781 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
24782 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
24783 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
24784 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
24785 vec_floato): Add builtin defines.
24786 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
24787 Update the built-in documentation file for the new built-in
24790 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24792 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
24793 (mthumb): Mark as the negative of -marm.
24795 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24797 * doc/invoke.texi (ARM Options, -mcpu): Document supported
24799 (ARM Options, -mtune): Document that this accepts the same
24800 extension options as -mcpu.
24801 (ARM Options, -mfpu): Document addition of -mfpu=auto.
24803 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24805 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
24806 permitted extensions.
24808 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24810 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
24811 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
24812 (armv8-m.main): Add option +nodsp.
24813 * config/arm/arm-cpu-cdata.h: Regenerated.
24815 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24817 * config/arm/t-fuchsia: New file.
24818 * config.gcc (arm*-*-fuchsia*): Use it.
24820 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24822 * config/arm/t-symbian: Rewrite for new option infrastructure.
24824 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24826 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
24827 (MULTILIB_REQUIRED): Likewise.
24829 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24831 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
24832 (MULTILIB_RESUE): Likewise.
24833 (MULTILIB_MATCHES): Likewise.
24834 (MULTLIB_REQUIRED): Likewise.
24836 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24838 * config/arm/t-rtems: Rewrite for new option framework.
24840 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24842 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
24843 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
24844 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
24845 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
24846 * config/arm/t-multilib: ... here.
24847 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
24848 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
24849 armv7-a and armv8*-a when A-profile libraries have not been built.
24850 * config/arm/t-rmprofile: Rewrite.
24852 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24854 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
24855 with a backslash. Remove the backslash after substituting unescaped
24857 * doc/fragments.texi (MULTILIB_REUSE): Document it.
24859 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24861 * config.gcc: (arm*-*-*): When building a-profile libraries, force
24862 the driver to pass through the default setting of -mfloat-abi.
24863 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
24865 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
24866 (all_feat_combs): New rule.
24867 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
24869 * config/arm/t-aprofile: Rewrite.
24871 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24873 * config/arm/arm.h (FPUTYPE_AUTO): Define.
24874 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
24875 fpu is not specified by the user/command-line.
24876 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
24877 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
24878 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
24879 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
24880 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
24881 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
24883 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24885 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
24886 * config/arm/t-arm-elf: Rewritten.
24888 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24890 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
24891 have some floating-point instructions.
24892 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
24893 (TARGET_MAYBE_HARD_FLOAT): New macro.
24894 * config/arm/arm-builtins.c (arm_init_builtins): Use
24895 TARGET_MAYBE_HARD_FLOAT.
24896 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
24898 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24900 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
24901 (configargs.h): Include it.
24902 (arm_print_hint_for_fpu_option): New function.
24903 (arm_parse_fpu_option): New function.
24904 (candidate_extension): New class.
24905 (arm_canon_for_multilib): New function.
24906 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
24907 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
24908 (ARCH_CANONICAL_SPECS): New macro.
24909 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
24911 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24913 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
24914 are set after handling multilib fragments. Set target_cpu_default2
24917 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24919 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
24921 (arm*-*-*): Set target_cpu_default2 to a quoted string.
24922 * config/arm/parsecpu.awk (check_cpu): Validate any extension
24924 (check_arch): Likewise.
24925 * config/arm/arm.c (arm_configure_build_target): Handle
24926 TARGET_CPU_DEFAULT being a string constant. Scan any feature
24927 options in the default.
24929 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24931 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
24932 when an option is an alias of another.
24933 * config/arm/parsecpu.awk (optalias): New parser token.
24934 (gen_comm_data): Mark non-alias options as such. Emit entries
24935 for extension aliases.
24936 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
24937 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
24938 (armv6kz, armv6zk, armv6t2): Likewise.
24939 (armv7): Make vfpv3-d16 an alias.
24940 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
24942 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
24943 Sort in canonical order.
24944 (armv8-a): Sort in canonical order.
24945 (armv8.1-a, armv8.2-a): Likewise.
24946 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
24948 (cortex-a9): Sort in canonical order.
24949 * config/arm/arm.c (selftests.h): Include it.
24950 (arm_test_cpu_arch_data): New function.
24951 (arm_run_self_tests): New function.
24952 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
24953 (targetm): Move declaration to the end of the file.
24954 * arm-cpu-cdata.h: Regenerated.
24956 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24958 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
24959 call to target_mode_check describing the type of option passed.
24960 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
24961 (arm_target_thumb_only): Use arm_parse_arch_option_name or
24962 arm_parse_cpu_option_name to match parameters against list of
24964 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
24965 arm_arch_core_flags data structure.
24966 * config/arm/arm-cpu_cdata.h: Regenerated.
24968 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24970 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
24972 (arm_print_hint_for_cpu_option): Likewise.
24973 (arm_print_hint_for_arch_option): Likewise.
24974 (arm_parse_cpu_option_name): Likewise.
24975 (arm_parse_arch_option_name): Likewise.
24976 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
24977 of entries in the all_fpus list.
24978 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
24979 (arm_parse_cpu_option_name): Declare.
24980 (arm_parse_arch_option_name): Declare.
24981 (arm_parse_option_features): Declare.
24982 (arm_intialize_isa): Declare.
24983 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
24985 (gen_comm_data): ... here. Make definitions non-static.
24986 * config/arm/arm-cpu-data.h: Regenerated.
24987 * config/arm/arm-cpu-cdata.h: Regenerated.
24989 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24991 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
24992 (cpu_arch_extension): New structure.
24993 (cpu_arch_option, arch_option, cpu_option): New structures.
24994 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
24995 architecture types.
24996 (gen_data): Generate new format data tables.
24997 * config/arm/arm.c (cpu_tune): New structure.
24998 (cpu_option, processors): Delete.
24999 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
25000 (arm_print_hint_for_cpu_option): ... this and ...
25001 (arm_print_hint_for_arch_option): ... this.
25002 (arm_parse_arch_cpu_name): Delete. Replace with ...
25003 (arm_parse_cpu_option_name): ... this and ...
25004 (arm_parse_arch_option_name): ... this.
25005 (arm_unrecognized_feature): Change type of target parameter to
25007 (arm_parse_arch_cpu_features): Delete. Replace with ...
25008 (arm_parse_option_features): ... this.
25009 (arm_configure_build_target): Rework to use new configuration data
25011 (arm_print_tune_info): Rework for new configuration data tables.
25012 * config/arm/arm-cpu-data.h: Regenerated.
25013 * config/arm/arm-cpu.h: Regenerated.
25015 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25017 * Makefile.in (OBJS): Move sbitmap.o from here ...
25018 (OBJS-libcommon): ... to here.
25020 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25022 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
25023 (ISA_ALL_CRYPTO): New macro.
25024 (ISA_ALL_SIMD): New macro
25025 (ISA_ALL_FP): New macro.
25026 * config/arm/arm.c (fpu_bitlist): Update initializer.
25027 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
25029 (arm9e): Add fpu. Add option for nofp
25030 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
25031 (arm926ej-s, arm1026ej-s): Likewise.
25032 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
25033 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
25034 neon-fp16, neon-vfpv4, nofp and nosimd.
25035 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
25036 (cortex-a8): Add fpu. Add option for nofp.
25037 (cortex-a9): Add fpu. Add options for nosimd and nofp.
25038 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
25039 (cortex-r4f): Add fpu.
25040 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
25041 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
25043 (cortex-r8): Likewise.
25044 (cortex-m4): Add fpu. Add option for nofp.
25045 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
25046 (cortex-a17.cortex-a7): Likewise.
25047 (cortex-a32): Add fpu. Add options for crypto and nofp.
25048 (cortex-a35, cortex-a53): Likewise.
25049 (cortex-a57): Add fpu. Add option for crypto.
25050 (cortex-a72, cortex-a73): Likewise.
25051 (exynos-m1): Likewise.
25052 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
25053 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
25054 (cortex-m33): Add fpu. Add option for nofp.
25055 * config/arm/arm-cpu-cdata.h: Regenerated
25056 * config/arm/arm-cpu-data.h: Regenerated.
25058 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25060 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
25061 (armv5te, armv5tej): Likewise.
25062 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
25063 (armv7): Add options fp and vfpv3-d16.
25064 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
25065 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
25067 (armv7ve): Likewise.
25068 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
25069 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
25070 (armv8-a): Add nocrypto option.
25071 (armv8.1-a, armv8.2-a): Likewise.
25072 (armv8-m.main): add options fp, fp.dp and nofp.
25074 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25076 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
25078 (armv8-a+crc): Delete.
25079 (armv8.1-a): Add options simd, crypto and nofp.
25080 (armv8.2-a): Add options fp16, simd, crypto and nofp.
25081 (armv8.2-a+fp16): Delete.
25082 (armv8-m.main): Add option dsp.
25083 (armv8-m.main+dsp): Delete.
25084 (cortex-a8): Add fpu. Add nofp option.
25085 (cortex-a9): Add fpu. Add nofp and nosimd options.
25086 * config/arm/parsecpu.awk (gen_data): Generate option tables and
25087 link to main cpu and architecture data structures.
25088 (gen_comm_data): Only put isa attributes from the main architecture
25090 (option): New statement for architecture and CPU entries.
25091 * arm.c (struct cpu_option): New structure.
25092 (struct processors): Add entry for options.
25093 (arm_unrecognized_feature): New function.
25094 (arm_parse_arch_cpu_name): Ignore any characters after the first
25096 (arm_parse_arch_cpu_feature): New function.
25097 (arm_configure_build_target): Separate out any CPU and architecture
25098 features and parse separately. Don't error out if -mfpu=auto is
25099 used with only an architecture string.
25100 (arm_print_asm_arch_directives): New function.
25101 (arm_file_start): Call it.
25102 * config/arm/arm-cpu-cdata.h: Regenerated.
25103 * config/arm/arm-cpu-data.h: Likewise.
25104 * config/arm/arm-tables.opt: Likewise.
25106 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25108 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
25109 assembler when it is not -mfpu=auto.
25111 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25113 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
25114 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
25115 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
25116 (ASM_CPU_SPEC): Rewrite.
25117 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
25118 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
25119 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
25120 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
25121 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
25122 copied string is NUL-terminated. Also strip any characters prefixed
25124 (arm_rewrite_selected_arch): New function.
25125 (arm_rewrite_march): New function.
25127 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25129 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
25130 (x_arm_cpu_string, x_arm_tune_string): Likewise.
25131 (march, mcpu, mtune): Convert to string-based options.
25132 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
25133 (arm_parse_arch_cpu_name): New function.
25134 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
25135 identify selected architecture or CPU.
25136 (arm_option_save): New function.
25137 (TARGET_OPTION_SAVE): Redefine.
25138 (arm_option_restore): Restore string options.
25139 (arm_option_print): Print string options.
25141 2017-06-16 Martin Sebor <msebor@redhat.com>
25143 PR tree-optimization/80933
25144 PR tree-optimization/80934
25145 * builtins.c (fold_builtin_3): Do not handle bcmp here.
25146 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
25147 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
25148 (gimple_fold_builtin): Call them.
25150 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25152 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
25153 as unlikely; update profile.
25155 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25157 * predict.c (force_edge_cold): Handle declaring edges impossible
25160 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25162 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
25164 (try_unroll_loop_completely): Fix reporting.
25166 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25168 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
25170 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
25173 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
25174 if given a non-constant argument for an intrinsic which requires a
25177 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25179 * profile.c (compare_freqs): New function.
25180 (branch_prob): Sort edge list.
25181 (find_spanning_tree): Assume that the list is priority sorted.
25183 2017-06-16 Richard Biener <rguenther@suse.de>
25185 PR tree-optimization/81090
25186 * passes.def (pass_record_bounds): Remove.
25187 * tree-pass.h (make_pass_record_bounds): Likewise.
25188 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
25189 make_pass_record_bounds): Likewise.
25190 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
25191 not free niter estimates at the beginning but at the end.
25192 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
25194 2017-06-16 Richard Biener <rguenther@suse.de>
25196 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
25197 initializer to workaround ICE in host GCC 4.8.
25199 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25201 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
25203 (clone_inlined_nodes): Update.
25205 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25207 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
25208 prefetch settings, and enable prefetching by default at -O3.
25210 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25212 * config/aarch64/aarch64.c (aarch64_override_options_internal):
25213 Set flag_prefetch_loop_arrays according to tuning data.
25215 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25217 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
25218 New tune structure.
25219 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
25220 [Unrelated to main purpose of the patch] Place the pointer field last
25221 to enable type checking errors when tune structure are wrongly merged.
25222 * config/aarch64/aarch64.c (generic_prefetch_tune,)
25223 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
25224 (thunderx2t99_prefetch_tune): New tune constants.
25225 (tune_params *_tunings): Update all tunings (no functional change).
25226 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
25227 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
25228 from tunings structures.
25230 2017-06-16 Jakub Jelinek <jakub@redhat.com>
25233 * ubsan.c (instrument_null): Add T argument, use it instead
25234 of computing it based on IS_LHS.
25235 (instrument_object_size): Likewise.
25236 (pass_ubsan::execute): Adjust instrument_null and
25237 instrument_object_size callers to pass gimple_get_lhs or
25238 gimple_assign_rhs1 result to it. Use instrument_null instead of
25239 calling get_base_address and instrument_mem_ref. Handle
25240 aggregate call arguments for object-size sanitization.
25242 2017-06-16 Yury Gribov <tetra2005@gmail.com>
25244 PR tree-optimization/81089
25245 * tree-vrp.c (is_masked_range_test): Validate operands of
25248 2017-06-15 Martin Sebor <msebor@redhat.com>
25251 * dumpfile.c (dump_register): Avoid calling memset to initialize
25252 a class with a default ctor.
25253 * gcc.c (struct compiler): Remove const qualification.
25254 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
25255 * hash-table.h: Ditto.
25256 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
25258 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
25259 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
25261 * params.h (struct param_info): Make struct members non-const.
25262 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
25263 with default initialization.
25264 * vec.h (vec_copy_construct, vec_default_construct): New helper
25266 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
25267 with vec_copy_construct.
25268 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
25269 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
25270 * doc/invoke.texi (-Wclass-memaccess): Document.
25272 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25274 * emit-rtl.h (is_leaf): Update comment about local
25275 register allocator.
25277 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
25280 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
25281 for a variable to have a section before checking if the section has a
25283 Set section to.persistent if persistent attribute is set.
25284 Warn if .persistent attribute is used on an automatic variable.
25286 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
25288 PR rtl-optimization/80474
25289 * reorg.c (update_block): Do not ignore instructions in a delay slot.
25291 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
25293 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
25296 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
25298 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
25299 (casesi): Emit bounds checking as RTL.
25300 (casesi_internal_mips16_<mode>): Remove bounds checking.
25302 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
25304 * config/xtensa/xtensa.c (xtensa_option_override): Append
25305 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
25306 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
25307 xtensa_doloop_hooks): Define unconditionally.
25308 (xtensa_reorg_loops): Only call reorg_loops in the presence of
25310 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
25311 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
25312 for it in xtensa_option_override.
25313 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
25314 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
25316 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
25318 * doc/cppopts.texi: Document '-' special value to -MF.
25320 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
25322 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
25323 (cortex_a53_fconst): Likewise.
25324 (cortex_a53_fpmul): Likewise.
25325 (cortex_a53_f_load_64): Likewise.
25326 (cortex_a53_f_load_many): Likewise.
25327 (cortex_a53_advsimd_alu): Likewise.
25328 (cortex_a53_advsimd_alu_q): Likewise.
25329 (cortex_a53_advsimd_mul): Likewise.
25330 (cortex_a53_advsimd_mul_q): Likewise.
25331 (fpmac bypass): Add new bypass for fpmac-fpmac case.
25332 Add missing fmul, r2f_cvt and fconst cases.
25334 2017-06-14 Richard Biener <rguenther@suse.de>
25336 PR middle-end/81088
25337 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
25339 (fold_binary_loc): When associating do not treat pre-existing
25340 TREE_OVERFLOW on literal constants as a reason to allow
25341 TREE_OVERFLOW on associated literal constants.
25343 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
25345 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
25346 (MASK_FEATURES): New macro.
25347 * config/sparc/sparc.c (sparc_option_override): Remove the special
25348 handling of -mfpu and generalize it to all MASK_FEATURES switches.
25350 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
25352 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
25353 a division of 0 if non-call exceptions are enabled.
25355 2017-06-14 Andrew Pinski <apinski@cavium.com>
25356 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25359 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
25360 Improve vector initialization code gen for only variable case.
25362 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
25364 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
25366 2017-06-14 Richard Biener <rguenther@suse.de>
25368 PR tree-optimization/81083
25369 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
25372 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25374 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
25375 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
25376 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
25377 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
25378 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
25379 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
25381 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25383 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
25384 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
25386 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25388 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
25390 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25392 * config/rs6000/t-rtems: Don't handle SPE.
25394 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25396 * config/rs6000/t-linux: Don't handle SPE.
25398 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25400 * config/rs6000/eabispe.h: Delete file.
25402 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25404 * config/rs6000/t-spe: Delete file.
25406 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25408 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
25409 (rs6000_legitimate_offset_address_p): Return false for anything in
25410 V2SImode or V2SFmode.
25412 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25414 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
25415 except V2SF and V2SI. Rearrange the vector modes, and add comments.
25416 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
25418 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
25419 (rs6000_legitimate_offset_address_p): Ditto.
25420 (rs6000_emit_move): Ditto.
25421 (rs6000_init_builtins): Remove V4HI_type_node.
25423 2017-06-13 Martin Liska <mliska@suse.cz>
25426 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
25427 (gate_asan): Likewise.
25428 * asan.h (asan_no_sanitize_address_p): Remove the function.
25429 (sanitize_flags_p): New function.
25430 * builtins.def: Fix coding style.
25431 * common.opt: Use renamed enum value.
25432 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
25433 * doc/extend.texi: Document no_sanitize attribute.
25434 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
25435 to SANITIZE_UNDEFINED_NONDEFAULT.
25436 * gcc.c (sanitize_spec_function): Use the renamed enum value.
25437 * gimple-fold.c (optimize_atomic_compare_exchange_p):
25438 Use sanitize_flags_p.
25439 * gimplify.c (gimplify_function_tree): Likewise.
25440 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
25441 * opts.c (parse_no_sanitize_attribute): New function.
25442 (common_handle_option): Use renamed enum value.
25443 * opts.h (parse_no_sanitize_attribute): Declare.
25444 * tree.c (sanitize_flags_p): New function.
25445 * tree.h: Declared here.
25446 * tsan.c: Use sanitize_flags_p.
25447 * ubsan.c (ubsan_expand_null_ifn): Likewise.
25448 (instrument_mem_ref): Likewise.
25449 (instrument_bool_enum_load): Likewise.
25450 (do_ubsan_in_current_function): Remove the function.
25451 (pass_ubsan::execute): Use sanitize_flags_p.
25452 * ubsan.h: Remove do_ubsan_in_current_function
25453 * tree-cfg.c (print_no_sanitize_attr_value): New function.
25454 (dump_function_to_file): Use it here.
25456 2017-06-13 Martin Jambor <mjambor@suse.cz>
25458 PR tree-optimization/80803
25459 PR tree-optimization/81063
25460 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
25461 (propagate_subaccesses_across_link): Enqueue subtree whenever
25462 necessary instead of relying on the caller.
25464 2017-06-13 Martin Jambor <mjambor@suse.cz>
25466 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
25467 that have a first_link.
25468 (sort_and_splice_var_accesses): Do not check first_link before
25470 (subtree_mark_written_and_enqueue): Likewise.
25471 (propagate_all_subaccesses): Likewise and do not stop at first
25472 parent with a first_link.
25474 2017-06-13 Martin Jambor <mjambor@suse.cz>
25476 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
25479 2017-06-13 Yury Gribov <tetra2005@gmail.com>
25481 * match.pd: New pattern.
25483 2017-06-13 Yury Gribov <tetra2005@gmail.com>
25485 * tree-vrp.c (is_masked_range_test): New function.
25486 (register_edge_assert_for): Determine ranges for
25489 2017-06-13 Yury Gribov <tetra2005@gmail.com>
25491 PR tree-optimization/67328
25492 * fold-const.c (maskable_range_p): New function.
25493 (build_range_check): Generate bittests if possible.
25495 2017-06-13 Martin Liska <mliska@suse.cz>
25497 * gimple-pretty-print.c (dump_probability): Add new argument.
25498 (dump_edge_probability): Dump both probability and count.
25499 (dump_gimple_label): Likewise.
25500 (dump_gimple_bb_header): Likewise.
25502 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
25505 * config/avr/avr-devices.c: Fix indentation.
25506 * config/avr/gen-avr-mmcu-specs.c: Dito.
25508 2017-06-13 Richard Biener <rguenther@suse.de>
25510 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
25511 instead get vector type from stmt_info.
25512 (vectorizable_reduction): Adjust. Remove dead code.
25514 2017-06-13 Richard Biener <rguenther@suse.de>
25516 PR middle-end/81065
25517 * fold-const.c (extract_muldiv_1): Remove bogus distribution
25518 case of C * (x * C2 + C3).
25519 (fold_addr_of_array_ref_difference): Properly fold index difference.
25521 2017-06-12 David S. Miller <davem@davemloft.net>
25524 * config/sparc/sparc.md (return expander): Emit frame blockage if
25525 function uses alloca.
25527 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
25529 * combine.c (make_field_assignment): Check len rather than the mode
25530 precision when calling force_to_mode.
25532 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
25534 Support multilibs and devices that see flash in RAM address range.
25537 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
25538 (avr_mcu_t) <flash_pm_offset>: New field.
25539 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
25540 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
25541 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
25542 (AVR_TINY_PM_OFFSET): Remove macro.
25543 * config/avr/avr.opt (-mshort-calls): New option.
25544 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
25545 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
25546 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
25547 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
25548 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
25549 instead of avr_arch->have_jmp_call.
25550 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
25551 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
25552 avr_arch->flash_pm_offset to define.
25553 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
25554 new field flash_pm_offset. Add entry for avrxmega3.
25555 (avr_texinfo): Add entry for avrxmega3.
25556 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
25557 attiny212, attiny214,
25558 attiny412, attiny414, attiny416, attiny417,
25559 attiny814, attiny816, attiny817,
25560 attiny1614, attiny1616, attiny1617,
25561 attiny3214, attiny3216, attiny3217.
25562 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
25563 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
25564 (avr_print_operand_address) [AVR_TINY]: Same.
25565 (avr_asm_init_sections) <readonly_data_section>: Only patch
25566 callback if avr_arch->flash_pm_offset = 0.
25567 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
25568 for rodata if avr_arch->flash_pm_offset != 0.
25569 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
25570 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
25571 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
25572 (m_options): Append opt_rcall.
25573 (m_dirnames): Append dir_rcall.
25574 * config/avr/t-multilib: Regenerate.
25576 * configure.ac [target=avr]: Check whether avrxmega3 default
25577 linker description file works as needed.
25578 * configure: Regenerate.
25579 * doc/avr-mmcu.texi: Regenerate.
25580 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
25581 <__AVR_ARCH__>: Document avrxmega3 and 103.
25582 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
25583 <__AVR_SHORT_CALLS__>: Document it.
25584 <__AVR_PM_BASE_ADDRESS__>: Document it.
25585 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
25586 (AVR Variable Attributes) <progmem>: Document this is
25587 not needed for avrxmega3.
25588 (AVR Named Address Spaces) <__flash>: Dito.
25590 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
25592 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
25594 2017-06-12 Doug Rupp <rupp@adacore.com>
25596 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
25597 Append vxworks-stdint.h to the tm_file list.
25598 * config/vxworks-stdint.h: New file.
25600 2017-06-12 Martin Liska <mliska@suse.cz>
25602 PR tree-optimization/81041
25603 * tree-profile.c (gimple_gen_ic_func_profiler):
25604 Create an extra BB in profile-generate
25605 (gimple_gen_time_profiler): Likewise.
25607 2017-06-12 Jakub Jelinek <jakub@redhat.com>
25609 PR tree-optimization/81003
25610 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
25611 (update_range_test): Use it instead of force_gimple_operand_gsi.
25613 2017-06-12 Richard Biener <rguenther@suse.de>
25615 PR tree-optimization/81053
25616 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
25617 with backedge value not defined in loop. Simplify def stmt
25620 2017-06-11 Tom de Vries <tom@codesourcery.com>
25623 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
25625 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
25626 nvptx_cannot_force_const_mem.
25628 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25630 * opts.c (finish_options): Move test for flag_split_stack after
25631 it has been initialized.
25633 2017-06-11 Jason Merrill <jason@redhat.com>
25635 * tree.h (id_equal): New.
25636 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
25637 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
25638 instead of strcmp of IDENTIFIER_POINTER.
25640 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25642 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
25643 (mark_all_inlined_calls_cdtor): Fix formating.
25644 (inline_transform): Rescale profile before inlining.
25646 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25648 * cgraph.h (cgraph_edge::clone): Update prototype.
25649 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
25650 (cgraph_node::create_clone): Update.
25651 (cgraph_node::create_version_clone): Update.
25652 * tree-inline.c (copy_bb): Update.
25653 (expand_call_inline): Update.
25655 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25657 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
25658 factored out from ...
25659 (rs6000_emit_prologue): ... here.
25661 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25663 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
25664 factored out from ...
25665 (rs6000_emit_prologue): ... here.
25667 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25669 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
25671 (handle_missing_profiles): Fix computation of tp_first_run.
25672 (counts_to_freqs): Do not touch freqs when count is 0.
25674 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25676 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
25679 2017-06-10 Tom de Vries <tom@codesourcery.com>
25681 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
25682 attributes): Document signal effective target.
25684 2017-06-10 Tom de Vries <tom@codesourcery.com>
25686 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
25687 Document effective target stack_size.
25689 2017-06-09 David Malcolm <dmalcolm@redhat.com>
25691 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
25692 to the edit_context if they can be auto-applied.
25694 2017-06-9 Ian Lance Taylor <iant@golang.org>
25696 * opts.c (finish_options): If -fsplit-stack, disable implicit
25697 -forder-blocks-and-partition.
25698 * doc/invoke.texi (Optimize Options): Document that when using
25699 -fsplit-stack -forder-blocks-and-partition is not implicitly
25702 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25704 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
25705 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
25706 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
25707 * builtins.def (abort, trap, unreachable): Declare cold.
25708 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
25709 * tree-core.h (ECF_COLD): New.
25710 * tree.c (set_call_expr_flags): Handle ECF_COLD.
25711 (build_common_builtin_nodes): Mark unreachable and abort as cold.
25713 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25715 * predict.c (unlikely_executed_stmt_p): Cleanup.
25717 2017-06-09 Richard Biener <rguenther@suse.de>
25719 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
25720 model if the ref is always written to.
25722 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25724 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
25726 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25728 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
25731 2017-06-09 Tom de Vries <tom@codesourcery.com>
25734 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
25735 "target cannot support label values" when encountering LABEL_REF.
25737 2017-06-09 Martin Liska <mliska@suse.cz>
25739 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
25740 (gimple_gen_ic_func_profiler): Emit direct comparison
25741 of __gcov_indirect_call_callee with NULL.
25742 (gimple_gen_time_profiler): Change probability from
25743 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
25745 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25747 * profile.c (edge_gcov_counts): Turn to pointer.
25748 (compute_branch_probabilities, compute_branch_probabilities): Update.
25749 (branch_prob): Do not clear edge_gcov_count.
25750 * profile.h (edge_gcov_counts): Turn to pointer.
25751 (edge_gcov_count): Update.
25753 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25755 * gimple.h (gimple_check_failed): Mark cold.
25757 2017-06-09 Richard Biener <rguenther@suse.de>
25759 PR tree-optimization/66623
25760 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
25761 refactor check_reduction into two parts, properly computing
25762 whether we have to check reduction validity for outer loop
25765 2017-06-09 Richard Biener <rguenther@suse.de>
25767 PR tree-optimization/79483
25768 * graphite-scop-detection.c (order): New global.
25769 (get_order): Compute bb to order mapping that satisfies code
25770 generation constraints.
25771 (cmp_pbbs): New helper.
25772 (build_scops): Start domwalk at entry block, sort generated
25775 2017-06-09 Richard Biener <rguenther@suse.de>
25777 PR middle-end/81007
25778 * ipa-polymorphic-call.c
25779 (ipa_polymorphic_call_context::restrict_to_inner_class):
25780 Skip FIELD_DECLs with error_mark_node type.
25781 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
25784 2017-06-09 Martin Liska <mliska@suse.cz>
25786 * predict.c (struct branch_predictor): New struct.
25787 (test_prediction_value_range): New test.
25788 (predict_c_tests): New function.
25789 * selftest-run-tests.c (selftest::run_tests): Run the function.
25790 * selftest.h: Declare new tests.
25792 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
25795 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
25796 gen_add3_insn did not fail.
25797 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
25798 r0, construct that number in a temporary reg and add that reg to r0.
25799 If asked to put the result in r0 as well, fail.
25801 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
25803 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25804 for early expansion of vec_eqv.
25806 2017-06-08 Jakub Jelinek <jakub@redhat.com>
25808 PR middle-end/81005
25809 * ubsan.c (instrument_null): Avoid pointless code temporary.
25810 (pass_ubsan::execute): Instrument aggregate arguments of calls.
25812 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
25816 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
25819 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
25822 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25824 * predict.c (unlikely_executed_edge_p): Move ahead.
25825 (probably_never_executed_edge_p): Use it.
25827 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25829 PR middle-end/79988
25830 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
25831 gimple_call_builtin_p call.
25833 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25835 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
25836 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
25837 rtl_check_failed_type2, rtl_check_failed_code1,
25838 rtl_check_failed_code2, rtl_check_failed_code_mode,
25839 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
25840 rtvec_check_failed_bounds, rtl_check_failed_flag,
25841 _fatal_insn_not_found, _fatal_insn): Likewise.
25842 * tree.h (tree_contains_struct_check_failed,
25843 tree_check_failed, tree_not_check_failed,
25844 tree_class_check_failed, tree_range_check_failed,
25845 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
25846 tree_vec_elt_check_failed, phi_node_elt_check_failed,
25847 tree_operand_check_failed, omp_clause_check_failed,
25848 omp_clause_operand_check_failed, omp_clause_range_check_failed):
25851 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25853 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
25854 flag_branch_probabilities.
25855 * ipa-inline.c (edge_badness): Likewise.
25856 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
25857 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
25858 * predict.c (maybe_hot_frequency_p): Likewise.
25859 (probably_never_executed): Likewise.
25860 * sched-ebb.c (schedule_ebbs): Likewise.
25861 * sched-rgn.c (find_single_block_region): Likewise.
25862 * tracer.c (tail_duplicate): Likewise.
25864 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25866 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
25867 longer requires x_flag_profile_use.
25869 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25871 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
25872 instead of flag_reorder_blocks_and_partition.
25873 * dbxout.c (dbxout_function_end): Likewise.
25874 * dwarf2out.c (gen_subprogram_die): Likewise.
25875 * haifa-sched.c (sched_create_recovery_edges): Likewise.
25876 * hw-doloop.c (reorg_loops): Likewise.
25877 * varasm.c (assemble_start_function,
25878 assemble_end_function): Likewise.
25879 (decide_function_section): Do not check for
25880 flag_reorder_blocks_and_partition.
25882 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25884 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
25886 (chkp_get_hard_register_fake_addr_expr): Ditto.
25887 (chkp_build_addr_expr): Add check for hard reg case.
25888 (chkp_parse_array_and_component_ref): Ditto.
25889 (chkp_find_bounds_1): Ditto.
25890 (chkp_process_stmt): Don't generate bounds store for
25893 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25895 * predict.c (maybe_hot_bb_p): Do not check profile status.
25896 (maybe_hot_edge_p): Likewise.
25897 (probably_never_executed): Check for zero counts even if profile
25899 (unlikely_executed_edge_p): New function.
25900 (unlikely_executed_stmt_p): New function.
25901 (unlikely_executed_bb_p): New function.
25902 (set_even_probabilities): Use unlikely predicates.
25903 (combine_predictions_for_bb): Likewise.
25904 (predict_paths_for_bb): Likewise.
25905 (predict_paths_leading_to_edge): Likewise.
25906 (determine_unlikely_bbs): New function.
25907 (estimate_bb_frequencies): Use it.
25908 (compute_function_frequency): Use zero counts even if profile is
25910 * profile-count.h: Fix typo.
25912 2017-08-08 Julia Koval <julia.koval@intel.com>
25914 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
25915 _mm512_mask_cvtsepi16_storeu_epi8,
25916 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
25917 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
25918 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
25919 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
25920 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
25921 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
25922 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
25923 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
25924 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
25925 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
25926 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
25927 __builtin_ia32_pmovuswb256mem_mask,
25928 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
25929 __builtin_ia32_pmovwb512mem_mask): New builtins.
25931 2017-08-08 Julia Koval <julia.koval@intel.com>
25933 PR target/73350,80862
25934 * config/i386/subst.md (round): Fix round pattern.
25935 * config/i386/i386.c (ix86_erase_embedded_rounding):
25936 Fix erasing rounding for the fixed pattern.
25938 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25940 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
25942 2017-06-08 Martin Liska <mliska@suse.cz>
25944 PR gcov-profile/80911
25945 * gcov.c (block_info::block_info): New constructor.
25947 2017-06-07 Carl Love <cel@us.ibm.com>
25949 * config/rs6000/rs6000-c: The return type of the following
25950 built-in functions was implemented as int not long long. Fix sign
25951 of return value for the unsigned version of vec_mulo and vec_mule.
25952 vector unsigned long long vec_bperm (vector unsigned long long,
25953 vector unsigned char)
25954 vector signed long long vec_mule (vector signed int,
25956 vector unsigned long long vec_mule (vector unsigned int,
25957 vector unsigned int)
25958 vector signed long long vec_mulo (vector signed int,
25960 vector unsigned long long vec_mulo (vector unsigned int,
25961 vector unsigned int)
25962 * doc/extend.texi: Fix the documentation for the built-in
25965 2017-06-07 Carl Love <cel@us.ibm.com>
25968 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
25971 2017-06-07 Carl Love <cel@us.ibm.com>
25973 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
25974 support, Generate doublehv for signed int/float for BE case only.
25976 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
25978 * doc/invoke.texi (mcx16): Rewrite.
25980 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25982 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
25983 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
25984 *mov<mode>_softfloat, and an anonymous splitter): Use
25985 nonimmediate_operand instead of rs6000_nonimmediate_operand.
25987 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25989 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
25991 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25992 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
25993 (rs6000_debug_reg_global): Adjust.
25994 (rs6000_init_hard_regno_mode_ok): Adjust.
25995 (rs6000_dbx_register_number): Adjust.
25996 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
25997 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25998 Remove SPE_ACC and SPEFSCR.
25999 (REG_ALLOC_ORDER): Ditto.
26000 (FRAME_POINTER_REGNUM): Change to 111.
26001 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
26002 (REG_CLASS_NAMES): Ditto.
26003 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
26004 (REGISTER_NAMES): Ditto.
26005 (ADDITIONAL_REG_NAMES): Ditto.
26006 (rs6000_reg_names): Ditto.
26007 * config/rs6000/rs6000.md: Renumber some register number
26010 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26012 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
26014 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
26015 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
26017 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
26018 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
26019 Delete the SPE high registers.
26020 (REG_ALLOC_ORDER): Ditto.
26021 (enum reg_class): Remove SPE_HIGH_REGS.
26022 (REG_CLASS_NAMES): Ditto.
26023 (REG_CLASS_CONTENTS): Delete the SPE high registers.
26024 (REGISTER_NAMES): Ditto.
26025 (rs6000_reg_names): Ditto.
26026 * doc/tm.texi.in: Remove SPE as example.
26027 * doc/tm.texi: Regenerate.
26029 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26031 * config/rs6000/8540.md (ppc8540_brinc): Delete.
26032 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
26033 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
26034 * config/rs6000/rs6000.md (type): Remove "brinc".
26036 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26038 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
26039 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
26040 * config/rs6000/linuxspe.h: Delete file.
26041 * config/rs6000/rs6000.md: Don't include spe.md.
26042 * config/rs6000/spe.h: Delete file.
26043 * config/rs6000/spe.md: Delete file.
26044 * config/rs6000/t-rs6000: Remove spe.md.
26046 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26048 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
26049 (reg_or_none500mem_operand): Delete.
26050 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
26051 instead of reg_or_none500mem_operand.
26053 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26055 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26056 handling of SPE flags.
26057 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
26059 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26061 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
26063 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
26064 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
26065 paired_divv2sf3): Similar.
26066 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
26067 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
26068 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
26070 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
26071 Rename the paired_* instruction patterns.
26072 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
26074 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
26075 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
26076 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
26077 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
26078 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
26079 PAIRED_VECTOR_MODE.
26080 (struct machine_function): Delete field spe_insn_chain_scanned_p.
26081 (spe_func_has_64bit_regs_p): Delete.
26082 (spe_expand_predicate_builtin): Delete.
26083 (spe_expand_evsel_builtin): Delete.
26084 (TARGET_DWARF_REGISTER_SPAN): Do not define.
26085 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
26086 (invalid_e500_subreg): Delete.
26087 (rs6000_legitimize_address): Always force_reg op2 as well, for
26088 paired single memory accesses.
26089 (rs6000_member_type_forces_blk): Delete.
26090 (rs6000_spe_function_arg): Delete.
26091 (rs6000_expand_unop_builtin): Delete SPE handling.
26092 (rs6000_expand_binop_builtin): Ditto.
26093 (spe_expand_stv_builtin): Delete.
26094 (bdesc_2arg_spe): Delete.
26095 (spe_expand_builtin): Delete.
26096 (spe_expand_predicate_builtin): Delete.
26097 (spe_expand_evsel_builtin): Delete.
26098 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
26099 (spe_init_builtins): Delete.
26100 (spe_func_has_64bit_regs_p): Delete.
26101 (savres_routine_name): Delete "info" parameter. Adjust callers.
26102 (rs6000_emit_stack_reset): Ditto.
26103 (rs6000_dwarf_register_span): Delete.
26104 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
26105 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
26106 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
26108 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
26110 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
26111 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
26112 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
26113 mulv2sf3, divv2sf3): Delete expanders.
26115 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26117 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
26119 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26121 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
26122 * config/rs6000/rs6000.c: Ditto.
26124 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26126 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
26127 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
26128 comparison_operator.
26130 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26132 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
26133 * config/rs6000/rs6000.opt: Ditto.
26134 * config/rs6000/t-rtems: Ditto.
26136 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26138 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
26139 TARGET_E500_SINGLE by 0, simplify.
26140 * config/rs6000/rs6000.c: Ditto.
26141 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
26142 (spe_build_register_parallel): Delete.
26143 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
26144 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
26145 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
26146 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
26147 (E500_CONVERT): Delete.
26148 * config/rs6000/spe.md: Remove many patterns and all define_constants.
26150 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26152 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
26153 * config/rs6000/dfp.md: Ditto.
26154 (negdd2, *negdd2_fpr): Merge.
26155 (absdd2, *absdd2_fpr): Merge.
26156 (negtd2, *negtd2_fpr): Merge.
26157 (abstd2, *abstd2_fpr): Merge.
26158 * config/rs6000/e500.h: Delete file.
26159 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
26160 TARGET_FPRS by 1 and simplify.
26161 * config/rs6000/rs6000-c.c: Ditto.
26162 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
26163 TARGET_DF_SPE by 0.
26164 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
26166 * config/rs6000/rs6000.md: Ditto.
26167 (floatdidf2, *floatdidf2_fpr): Merge.
26168 (move_from_CR_gt_bit): Delete.
26169 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
26170 (E500_CR_IOR_COMPARE): Delete.
26171 (All patterns that require !TARGET_FPRS): Delete.
26172 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
26174 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26176 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
26178 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26180 * graphds.c (add_edge): Intitialize edge's attached data.
26181 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
26182 pointer parameter. Call pointed function on each edge during
26183 graph traversing. Skip traversing the edge when the function
26185 (graphds_dfs, graphds_scc): Ditto.
26186 (for_each_edge): New parameter. Pass the new parameter to callback
26188 * graphds.h (skip_edge_callback): New function pointer type.
26189 (graphds_dfs, graphds_scc): New function pointer parameter.
26190 (graphds_edge_callback, for_each_edge): New parameter.
26192 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26194 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
26195 out code checking if runtime alias check is possible to below ...
26196 Call the new function.
26197 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
26198 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
26200 2017-06-07 Marek Polacek <polacek@redhat.com>
26203 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
26204 TYPE_OVERFLOW_WRAPS check.
26206 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26208 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
26209 if versioning is required.
26210 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
26211 peeling with the check for versioning.
26213 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26215 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
26216 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
26217 Set true to new parameter if new ssa variable is defined.
26218 (vect_gen_vector_loop_niters): Refactor. Set range information
26219 for the new vector loop bound variable.
26220 (vect_do_peeling): Ditto.
26222 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26224 * tree-affine.c (ssa.h): Include header file.
26225 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
26226 has wrapping overflow behavior.
26228 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26230 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
26232 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26234 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
26235 (tree_to_aff_combination): ... here.
26237 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26239 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
26240 reg_pressure model function.
26241 (ivopts_global_cost_for_size): Delete.
26242 (determine_set_costs, iv_ca_recount_cost): Call new model function
26243 ivopts_estimate_reg_pressure.
26245 2017-06-07 Tamar Christina <tamar.christina@arm.com>
26247 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
26248 expensive than udiv. Remove floating point cases from mod.
26250 2017-06-07 Tamar Christina <tamar.christina@arm.com>
26252 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
26253 Increase idiv cost.
26255 2017-06-07 Tamar Christina <tamar.christina@arm.com>
26257 * config/aarch64/aarch64.md
26258 (copysignsf3): Fix mask generation.
26260 2017-06-07 Jakub Jelinek <jakub@redhat.com>
26262 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
26264 (class dump_manager): Add register_dumps method.
26265 * dumpfile.c: Include langhooks.h.
26266 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
26267 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
26268 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
26269 (dump_manager::dump_register): Start with 512 entries instead of 32.
26270 (dump_manager::register_dumps): New method.
26271 * toplev.c (general_init): Instead of invoking register_dumps
26272 langhook, invoke register_dumps method on the dump manager.
26273 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
26276 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
26278 * doc/md.texi: Clarify the restrictions on a define_insn condition.
26279 Say that # requires an associated define_split to exist, and that
26280 the define_split must be suitable for use after register allocation.
26282 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26284 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
26285 (compute_outgoing_frequencies): Also initialize zero counts.
26286 (find_many_sub_basic_blocks): Do not produce uninitialized profile
26287 around loops; preserve more of profile when nothing changes.
26289 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
26291 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
26293 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
26294 * config/arm/arm-cpu-cdata.h: Regenerate.
26295 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
26296 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
26297 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
26298 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
26299 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
26301 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
26303 * config/arm/t-rmprofile: Likewise.
26304 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
26306 2017-06-06 David S. Miller <davem@davemloft.net>
26309 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
26310 blockage if function uses alloca.
26312 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26314 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
26315 New "uid" fields to hold pretty-print IDs of group and ref.
26316 Memory references are now identified as <group_id>:<ref_id>
26317 instead of using [random] addresses.
26318 (dump_mem_details): Simplify, no functional change.
26319 (dump_mem_ref): Simplify and make output more concise.
26320 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
26321 (find_or_create_group): Initialize group uid.
26322 (record_ref): Initialize ref uid. Improve debug output.
26323 (prune_group_by_reuse, should_issue_prefetch_p,)
26324 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
26325 (mark_nontemporal_store, determine_loop_nest_reuse):
26326 Improve debug output.
26328 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26330 * dbgcnt.def (prefetch): New debug counter.
26331 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
26332 (schedule_prefetches): Stop issueing prefetches if debug counter
26335 2017-06-06 Tom de Vries <tom@codesourcery.com>
26337 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
26338 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
26340 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26342 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
26343 Use aarch64_reg_or_zero predicate for operand 4.
26344 (aarch64_compare_and_swap<mode> define_insn_and_split):
26345 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
26346 (aarch64_store_exclusive<mode>): Likewise for operand 2.
26348 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
26350 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
26351 (arm_compute_save_core_reg_mask): This.
26352 (thumb1_compute_save_reg_mask): Rename into ...
26353 (thumb1_compute_save_core_reg_mask): This.
26354 (arm_compute_save_reg0_reg12_mask): Adapt comment.
26355 (arm_compute_frame_layout): Likewise.
26357 2017-06-06 Richard Biener <rguenther@suse.de>
26359 PR tree-optimization/80974
26360 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
26361 keep or clear leaders SSA info.
26363 2017-06-06 Tom de Vries <tom@codesourcery.com>
26365 * config/nvptx/nvptx.c (split_mode_p): New function.
26366 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
26368 2017-06-06 Tom de Vries <tom@codesourcery.com>
26370 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
26372 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26375 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
26378 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26380 * shrink-wrap.c (handle_simple_exit): Update profile.
26381 (try_shrink_wrapping): Upate profile.
26383 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26385 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
26386 (tree_guess_outgoing_edge_probabilities): New.
26387 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
26388 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
26390 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26392 * ipa-split.c (split_function): Initialize return bb profile.
26394 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26396 * profile.c (compute_branch_probabilities): Also initialize
26397 EXIT_BLOCK profile.
26399 2017-06-06 Richard Biener <rguenther@suse.de>
26401 PR tree-optimization/80928
26402 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
26403 (vect_analyze_loop_operations): Properly guard analysis for
26405 (vect_transform_loop): Likewise.
26406 (vect_analyze_loop_2): Also reset SLP type on PHIs.
26407 (vect_model_induction_cost): Do not cost for pure SLP.
26408 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
26409 of induction in inner loop vectorization.
26410 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
26411 (vect_get_and_check_slp_defs): Handle vect_induction_def.
26412 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
26414 (vect_analyze_slp_cost_1): Cost induction.
26415 (vect_detect_hybrid_slp_stmts): Handle PHIs.
26416 (vect_get_slp_vect_defs): Likewise.
26417 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
26418 (vect_transform_stmt): Handle SLP reductions.
26419 * tree-vectorizer.h (vectorizable_induction): Adjust.
26421 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
26423 * config/rs6000/rs6000.c (make_resolver_func): Update
26424 init_lowered_empty_function call.
26426 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
26428 * doc/invoke.texi: Document the -fprofile-abs-path option.
26429 * common.opt (fprofile-abs-path): New option.
26430 * gcov-io.h (gcov_write_filename): Declare.
26431 * gcov-io.c (gcov_write_filename): New function.
26432 * coverage.c (coverage_begin_function): Use gcov_write_filename.
26433 * profile.c (output_location): Likewise.
26435 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26437 * shring-wrap.c: Revert accidental commit.
26439 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
26441 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
26443 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26445 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
26447 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
26448 profile in callgraph edge.
26449 * profile-count.h (apply_probability): If THIS is 0, then result is 0
26450 (apply_scale): Likewise.
26451 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
26452 Also scale profile when inlining function with zero profile.
26453 (initialize_cfun): Update exit block profile even when it is zero.
26454 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
26455 when profile is read.
26457 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
26459 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
26460 (CLONE_*): New constants to define the processors we can generate
26461 code for with the target_clone attribute.
26462 (rs6000_clone_map): New array to identify which clone processors
26463 the current program is running on.
26464 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
26465 target_clone attribute.
26466 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
26467 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
26468 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
26469 (cpu_expand_builtin): Add support for target_clone attribute.
26470 (rs6000_valid_attribute_p): Allow "default" attribute.
26471 (get_decl_name): New debug function to simplify printing the
26472 current function name in debugging statements.
26473 (rs6000_clone_priority): New functions to support the target_clone
26474 attribute, and be able to generate code to switch between ISA 2.05
26475 through ISA 3.0 (power6 through power9).
26476 (rs6000_compare_version_priority): Likewise.
26477 (rs6000_get_function_versions_dispatcher): Likewise.
26478 (make_resolver_func): Likewise.
26479 (add_condition_to_bb): Likewise.
26480 (dispatch_function_versions): Likewise.
26481 (rs6000_generate_version_dispatcher_body): Likewise.
26482 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
26483 (fusion_gpr_load_p): Fix a spacing issue.
26484 * doc/extend.texi (Common Function Attributes): Document that the
26485 PowerPC supports the target_clone attribute.
26487 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
26489 * config/arm/arm.h: explain F symbol found in description of ARM
26490 register allocation in its legend.
26492 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26494 * config/mips/frame-header-opt.c: Include profile-count.h.
26495 * config/riscv/riscv.c: Include profile-count.h
26497 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26499 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
26501 (sm_set_flag_if_changed): Add bbs field.
26502 (execute_sm_if_changed_flag_set): Pass BBS.
26503 (execute_sm): Update.
26505 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26507 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
26510 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26512 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
26513 (peephole2): New peephole2 to emit the above.
26514 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
26516 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26518 * config/aarch64/aarch64.c (define_peephole2 above
26519 *sub_<shift>_<mode>): New peephole.
26521 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26523 * config/i386/i386.c (make_resolver_func): Update.
26524 * Makefile.in: Add profile-count.h and profile-count.o
26525 * auto-profile.c (afdo_indirect_call): Update to new API.
26526 (afdo_set_bb_count): Update.
26527 (afdo_propagate_edge): Update.
26528 (afdo_propagate_circuit): Update.
26529 (afdo_calculate_branch_prob): Update.
26530 (afdo_annotate_cfg): Update.
26531 * basic-block.h: Include profile-count.h
26532 (struct edge_def): Turn count to profile_count.
26533 (struct basic_block_def): Likewie.
26534 (REG_BR_PROB_BASE): Move to profile-count.h
26535 (RDIV): Move to profile-count.h
26536 * bb-reorder.c (max_entry_count): Turn to profile_count.
26537 (find_traces): Update.
26538 (rotate_loop):Update.
26539 (connect_traces):Update.
26540 (sanitize_hot_paths):Update.
26541 * bt-load.c (migrate_btr_defs): Update.
26542 * cfg.c (RDIV): Remove.
26543 (init_flow): Use alloc_block.
26544 (alloc_block): Uninitialize count.
26545 (unchecked_make_edge): Uninitialize count.
26546 (check_bb_profile): Update.
26547 (dump_edge_info): Update.
26548 (dump_bb_info): Update.
26549 (update_bb_profile_for_threading): Update.
26550 (scale_bbs_frequencies_int): Update.
26551 (scale_bbs_frequencies_gcov_type): Update.
26552 (scale_bbs_frequencies_profile_count): New.
26553 * cfg.h (update_bb_profile_for_threading): Update.
26554 (scale_bbs_frequencies_profile_count): Declare.
26555 * cfgbuild.c (compute_outgoing_frequencies): Update.
26556 (find_many_sub_basic_blocks): Update.
26557 * cfgcleanup.c (try_forward_edges): Update.
26558 (try_crossjump_to_edge): Update.
26559 * cfgexpand.c (expand_gimple_tailcall): Update.
26560 (construct_exit_block): Update.
26561 * cfghooks.c (verify_flow_info): Update.
26562 (dump_bb_for_graph): Update.
26563 (split_edge): Update.
26564 (make_forwarder_block): Update.
26565 (duplicate_block): Update.
26566 (account_profile_record): Update.
26567 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
26568 (get_estimated_loop_iterations): Update.
26569 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
26570 (single_likely_exit): Update.
26571 * cfgloopmanip.c (scale_loop_profile): Update.
26573 (set_zero_probability): Update.
26574 (lv_adjust_loop_entry_edge): Update.
26575 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
26576 (purge_dead_edges): Update.
26577 (rtl_account_profile_record): Update.
26578 * cgraph.c (cgraph_node::create): Uninitialize count.
26579 (symbol_table::create_edge): Uninitialize count.
26580 (cgraph_update_edges_for_call_stmt_node): Update.
26581 (cgraph_edge::dump_edge_flags): Update.
26582 (cgraph_node::dump): Update.
26583 (cgraph_edge::maybe_hot_p): Update.
26584 * cgraph.h: Include profile-count.h
26585 (create_clone), create_edge, create_indirect_edge): Update.
26586 (cgraph_node): Turn count to profile_count.
26587 (cgraph_edge0: Likewise.
26588 (make_speculative, clone): Update.
26589 (create_edge): Update.
26590 (init_lowered_empty_function): Update.
26591 * cgraphclones.c (cgraph_edge::clone): Update.
26592 (duplicate_thunk_for_node): Update.
26593 (cgraph_node::create_clone): Update.
26594 * cgraphunit.c (cgraph_node::analyze): Update.
26595 (cgraph_node::expand_thunk): Update.
26596 * final.c (dump_basic_block_info): Update.
26597 * gimple-streamer-in.c (input_bb): Update.
26598 * gimple-streamer-out.c (output_bb): Update.
26599 * graphite.c (print_global_statistics): Update.
26600 (print_graphite_scop_statistics): Update.
26601 * hsa-brig.c: Include basic-block.h.
26602 * hsa-dump.c: Include basic-block.h.
26603 * hsa-gen.c (T sum_slice): Update.
26604 (convert_switch_statements):Update.
26605 * hsa-regalloc.c: Include basic-block.h.
26606 * ipa-chkp.c (chkp_produce_thunks): Update.
26607 * ipa-cp.c (struct caller_statistics): Update.
26608 (init_caller_stats): Update.
26609 (gather_caller_stats): Update.
26610 (ipcp_cloning_candidate_p): Update.
26611 (good_cloning_opportunity_p): Update.
26612 (get_info_about_necessary_edges): Update.
26613 (dump_profile_updates): Update.
26614 (update_profiling_info): Update.
26615 (update_specialized_profile): Update.
26616 (perhaps_add_new_callers): Update.
26617 (decide_about_value): Update.
26618 (ipa_cp_c_finalize): Update.
26619 * ipa-devirt.c (struct odr_type_warn_count): Update.
26620 (struct decl_warn_count): Update.
26621 (struct final_warning_record): Update.
26622 (possible_polymorphic_call_targets): Update.
26623 (ipa_devirt): Update.
26624 * ipa-fnsummary.c (redirect_to_unreachable): Update.
26625 * ipa-icf.c (sem_function::merge): Update.
26626 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
26627 * ipa-inline.c (compute_uninlined_call_time): Update.
26628 (compute_inlined_call_time): Update.
26629 (want_inline_small_function_p): Update.
26630 (want_inline_self_recursive_call_p): Update.
26631 (edge_badness): Update.
26632 (lookup_recursive_calls): Update.
26633 (recursive_inlining): Update.
26634 (inline_small_functions): Update.
26635 (dump_overall_stats): Update.
26636 (dump_inline_stats): Update.
26637 * ipa-profile.c (ipa_profile_generate_summary): Update.
26638 (ipa_propagate_frequency): Update.
26639 (ipa_profile): Update.
26640 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26641 * ipa-utils.c (ipa_merge_profiles): Update.
26642 * loop-doloop.c (doloop_modify): Update.
26643 * loop-unroll.c (report_unroll): Update.
26644 (unroll_loop_runtime_iterations): Update.
26645 * lto-cgraph.c (lto_output_edge): Update.
26646 (lto_output_node): Update.
26647 (input_node): Update.
26648 (input_edge): Update.
26649 (merge_profile_summaries): Update.
26650 * lto-streamer-in.c (input_cfg): Update.
26651 * lto-streamer-out.c (output_cfg): Update.
26652 * mcf.c (create_fixup_graph): Update.
26653 (adjust_cfg_counts): Update.
26654 (sum_edge_counts): Update.
26655 * modulo-sched.c (sms_schedule): Update.
26656 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
26657 * predict.c (maybe_hot_count_p): Update.
26658 (probably_never_executed): Update.
26659 (dump_prediction): Update.
26660 (combine_predictions_for_bb): Update.
26661 (propagate_freq): Update.
26662 (handle_missing_profiles): Update.
26663 (counts_to_freqs): Update.
26664 (rebuild_frequencies): Update.
26665 (force_edge_cold): Update.
26666 * predict.h: Include profile-count.h
26667 (maybe_hot_count_p, counts_to_freqs): UPdate.
26668 * print-rtl-function.c: Do not include cfg.h
26669 * print-rtl.c: Include basic-block.h
26670 * profile-count.c: New file.
26671 * profile-count.h: New file.
26672 * profile.c (is_edge_inconsistent): Update.
26673 (correct_negative_edge_counts): Update.
26674 (is_inconsistent): Update.
26675 (set_bb_counts): Update.
26676 (read_profile_edge_counts): Update.
26677 (compute_frequency_overlap): Update.
26678 (compute_branch_probabilities): Update; Initialize and deinitialize
26680 (branch_prob): Update.
26681 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
26682 (edge_gcov_count): New.
26683 (bb_gcov_count): New.
26684 * shrink-wrap.c (try_shrink_wrapping): Update.
26685 * tracer.c (better_p): Update.
26686 * trans-mem.c (expand_transaction): Update.
26687 (ipa_tm_insert_irr_call): Update.
26688 (ipa_tm_insert_gettmclone_call): Update.
26689 * tree-call-cdce.c: Update.
26690 * tree-cfg.c (gimple_duplicate_sese_region): Update.
26691 (gimple_duplicate_sese_tail): Update.
26692 (gimple_account_profile_record): Update.
26693 (execute_fixup_cfg): Update.
26694 * tree-inline.c (copy_bb): Update.
26695 (copy_edges_for_bb): Update.
26696 (initialize_cfun): Update.
26697 (freqs_to_counts): Update.
26698 (copy_cfg_body): Update.
26699 (expand_call_inline): Update.
26700 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26701 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
26702 (try_unroll_loop_completely): Update.
26703 (try_peel_loop): Update.
26704 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26705 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
26706 * tree-ssa-loop-split.c (connect_loops): Update.
26707 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
26708 * tree-ssa-reassoc.c (branch_fixup): Update.
26709 * tree-ssa-tail-merge.c (replace_block_by): Update.
26710 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
26711 (compute_path_counts): Update.
26712 (update_profile): Update.
26713 (recompute_probabilities): Update.
26714 (update_joiner_offpath_counts): Update.
26715 (estimated_freqs_path): Update.
26716 (freqs_to_counts_path): Update.
26717 (clear_counts_path): Update.
26718 (ssa_fix_duplicate_block_edges): Update.
26719 (duplicate_thread_path): Update.
26720 * tree-switch-conversion.c (case_bit_test_cmp): Update.
26721 (struct switch_conv_info): Update.
26722 * tree-tailcall.c (decrease_profile): Update.
26723 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26724 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26725 * value-prof.c (check_counter): Update.
26726 (gimple_divmod_fixed_value): Update.
26727 (gimple_mod_pow2): Update.
26728 (gimple_mod_subtract): Update.
26729 (gimple_ic_transform): Update.
26730 (gimple_stringop_fixed_value): Update.
26731 * value-prof.h (gimple_ic): Update.
26733 2017-06-02 Carl Love <cel@us.ibm.com>
26735 * config/rs6000/rs6000-c: Add support for built-in functions
26736 vector double vec_doublee (vector signed int);
26737 vector double vec_doublee (vector unsigned int);
26738 vector double vec_doublee (vector float);
26739 vector double vec_doubleh (vector signed int);
26740 vector double vec_doubleh (vector unsigned int);
26741 vector double vec_doubleh (vector float);
26742 vector double vec_doublel (vector signed int);
26743 vector double vec_doublel (vector unsigned int);
26744 vector double vec_doublel (vector float);
26745 vector double vec_doubleo (vector signed int);
26746 vector double vec_doubleo (vector unsigned int);
26747 vector double vec_doubleo (vector float);.
26748 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
26749 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
26751 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
26752 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
26753 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
26755 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
26756 vec_doublel, vec_doubleh.
26757 * doc/extend.texi: Update the built-in documentation file for the
26758 new built-in functions.
26760 2017-06-02 David Malcolm <dmalcolm@redhat.com>
26763 * ipa-inline-analysis.c (free_growth_caches): Set
26764 edge_removal_hook_holder to NULL after removing it.
26766 2017-06-02 Sudakshina Das <sudi.das@arm.com>
26768 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
26769 comparision with zero.
26771 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26772 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26773 for early expansion of vec_min and vec_max builtins.
26774 (builtin_function_type): Add min/max unsigned variants to those
26775 identified as having unsigned arguments.
26777 2017-06-02 Olivier Hainque <hainque@adacore.com>
26779 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
26781 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26783 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
26784 Use VALL_F16 iterator rather than VALL.
26786 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26788 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
26789 Emit CBNZ inside loop when doing a strong exchange and comparing
26790 against zero. Generate the CC flags after the loop.
26792 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
26794 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
26795 (dl_section_ref): New.
26796 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
26797 On AIX, append an expression to subtract the size of the
26798 section length to dl_section_ref.
26800 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26802 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26803 for early expansion of vector absolute builtins.
26805 2017-06-02 Richard Biener <rguenther@suse.de>
26807 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
26808 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
26810 2017-06-02 Richard Biener <rguenther@suse.de>
26812 PR tree-optimization/80948
26813 * tree-tailcall.c (find_tail_calls): Track stmts to move in
26814 stmt order as well.
26816 2017-06-02 Richard Biener <rguenther@suse.de>
26818 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
26820 * tree-vect-stmts.c (process_use): Do not mark backedge defs
26821 for inductions as relevant.
26823 2017-06-02 Richard Biener <rguenther@suse.de>
26825 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
26826 (vectorizable_induction): ... this. Remove dead code.
26828 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
26830 * builtins. (expand_builtin_alloca): Remove second parameter and
26831 infer its value from the first parameter instead.
26832 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
26834 2017-06-02 Jakub Jelinek <jakub@redhat.com>
26836 PR rtl-optimization/80903
26837 * loop-doloop.c (add_test): Unshare sequence.
26839 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
26841 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
26843 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
26845 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
26847 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
26848 xlogue_layout::get_instance, logue_layout::xlogue_layout,
26849 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
26850 (xlogue_layout::get_stub_rtx): Make static.
26851 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
26852 (xlogue_layout::compute_stub_managed_regs): Rename to...
26853 (xlogue_layout::count_stub_managed_regs): ...this.
26854 (xlogue_layout::is_stub_managed_reg): New function.
26855 (xlogue_layout::m_stub_names): Rename to...
26856 (xlogue_layout::s_stub_names): ...this, make static.
26857 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
26858 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
26859 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
26860 xlogue_layout::s_stub_names): Instantiate statics.
26861 (stub_managed_regs): Remove.
26862 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
26863 (disable_call_ms2sysv_xlogues): Rename to...
26864 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
26865 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
26867 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
26868 change after reload_completed.
26869 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
26871 (ix86_expand_prologue): Likewise.
26872 (ix86_expand_epilogue): Likewise.
26873 (ix86_expand_split_stack_prologue): Likewise.
26874 (ix86_compute_frame_layout): Remove frame parameter ...
26875 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
26876 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
26878 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
26879 (ix86_frame): Move from here ...
26880 * config/i386/i386.h (ix86_frame): ... to here.
26881 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
26882 complete ix86_frame data structure instead. Remove some_ld_name.
26884 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26886 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
26887 symbols that hold a DECL_VALUE_EXPR.
26889 2017-06-01 Martin Jambor <mjambor@suse.cz>
26891 PR tree-optimization/80898
26892 * tree-sra.c (process_subtree_disqualification): Removed.
26893 (disqualify_candidate): Do not acll
26894 process_subtree_disqualification.
26895 (subtree_mark_written_and_enqueue): New function.
26896 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
26897 RHS has been disqualified and re-queue LHS if necessary. Apart
26898 from that, ignore disqualified RHS.
26900 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26902 * config/s390/s390.c (s390_emit_epilogue): Disable early return
26903 address fetch for z10 or later.
26905 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26907 * config/arc/arc.md (tst_movb): Add guard when splitting.
26909 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26911 * config/arc/arc.c (arc_can_eliminate): Test against
26912 arc_frame_pointer_needed.
26914 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26916 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
26917 to prevent store reordering.
26918 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
26919 (type): Add block type.
26920 (stack_tie): Define special instruction to be used in
26923 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26925 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
26926 constraint. It is not valid for the pattern.
26927 (noncommutative_binary_comparison): Likewise.
26929 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26931 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
26934 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26936 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
26937 be used by the reg-alloc.
26939 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26941 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
26942 reg-alloc when having mul64 or mul32x16 instructions.
26943 (mulsidi3): Likewise.
26944 (umulsidi3): Likewise.
26945 (mulsi32x16): New pattern.
26946 (mulsi64): Likewise.
26947 (mulsidi64): Likewise.
26948 (umulsidi64): Likewise.
26949 (MUL32x16_REG): Define.
26950 (mul64_600): Use MUL32x16_REG.
26951 (mac64_600): Likewise.
26952 (umul64_600): Likewise.
26953 (umac64_600): Likewise.
26955 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26957 * config/arc/arc.md (mulsi3_700): Make it commutative.
26959 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
26961 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
26963 (*sign_extendsidi2_insn): Likewise for movstosw.
26965 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26967 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
26968 the type of the input discriminant value. Convert the
26969 discriminant value of signedness vary.
26971 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
26973 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
26974 Add to -Wall section.
26976 2017-06-01 Richard Biener <rguenther@suse.de>
26978 PR middle-end/66313
26979 * fold-const.c (fold_plusminus_mult_expr): If the factored
26980 factor may be zero use a wrapping type for the inner operation.
26981 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
26982 and handle moved defs.
26983 (process_assignment): Properly guard the unary op case. Return a
26984 tri-state indicating that moving the stmt before the call may allow
26985 to continue. Pass through to_move.
26986 (find_tail_calls): Handle moving unrelated defs before
26989 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
26992 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
26993 splitter result in the canonical way.
26995 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
26997 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
26998 also for 32bit target. Update insn attributes.
26999 (zero-extendsidi2 splitter): Allow all registers for operand 1.
27001 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
27003 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
27004 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
27005 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
27006 (_mm_maskz_min_ss): New intrinsics.
27008 2017-05-31 Martin Liska <mliska@suse.cz>
27010 * tree-vect-loop.c (vect_create_epilog_for_reduction):
27011 Change comment style to one we normally use.
27012 (vectorizable_reduction): Likewise.
27013 (vectorizable_induction): Likewise.
27014 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
27015 (vectorizable_call): Likewise.
27016 (vectorizable_simd_clone_call): Likewise.
27017 (vectorizable_conversion): Likewise.
27018 (vectorizable_assignment): Likewise.
27019 (vectorizable_shift): Likewise.
27020 (vectorizable_operation): Likewise.
27021 (vectorizable_store): Likewise.
27022 (vectorizable_load): Likewise.
27023 * tree-vectorizer.h: Likewise.
27025 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
27027 * passes.c (emergency_dump_function): New.
27028 * tree-pass.h (emergency_dump_function): Declare.
27029 * plugin.c (plugins_internal_error_function): Remove.
27030 * plugin.h (plugins_internal_error_function): Remove declaration.
27031 * toplev.c (internal_error_function): New static function. Use it...
27032 (general_init): ...here.
27034 2017-05-31 Graham Markall <graham.markall@embecosm.com>
27036 * config/arc/arc.c (arc_print_operand): Handle constant operands.
27037 (arc_rtx_costs): Add costs for new patterns.
27038 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
27039 * config/arc/predicates.md: Add _1_2_3_operand predicate.
27041 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
27043 * tree-ssa-strlen.c (get_next_strinfo): New function.
27044 (get_stridx_plus_constant): Use it.
27045 (zero_length_string): Likewise.
27046 (adjust_related_strinfos): Likewise.
27047 (adjust_last_stmt): Likewise.
27049 2017-05-31 Richard Biener <rguenther@suse.de>
27052 * config/i386/i386.c (ix86_expand_builtin): Remove assert
27053 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
27055 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
27057 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
27058 loop_vinfo argument and use of dependence distance vectors.
27059 Check instead whether the two references differ only in their
27060 initial value and assume that they have the same alignment if the
27061 difference is a multiple of the vector alignment.
27062 (vect_analyze_data_refs_alignment): Update call accordingly.
27064 2017-05-31 Martin Liska <mliska@suse.cz>
27067 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
27069 2017-05-31 Bin Cheng <bin.cheng@arm.com>
27071 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
27072 (create_intersect_range_checks): Move from ...
27073 * tree-data-ref.c (create_intersect_range_checks_index)
27074 (create_intersect_range_checks): ... to here.
27075 (create_runtime_alias_checks): New function factored from ...
27076 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
27077 here. Call above function.
27078 * tree-data-ref.h (create_runtime_alias_checks): New function.
27080 2017-05-31 Bin Cheng <bin.cheng@arm.com>
27082 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
27083 segment length for dr_b and compute it in wide_int.
27085 2017-05-31 Richard Biener <rguenther@suse.de>
27087 PR tree-optimization/80906
27088 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
27089 and pass through iv_map.
27090 (copy_bb_and_scalar_dependences): Adjust.
27091 (translate_pending_phi_nodes): Likewise.
27092 (copy_loop_close_phi_args): Handle code-generating IVs instead
27095 2017-05-30 David Malcolm <dmalcolm@redhat.com>
27097 * diagnostic-color.c (color_dict): Add "type-diff".
27098 (parse_gcc_colors): Update comment.
27099 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
27100 -fdiagnostics-show-template-tree and -fno-elide-type.
27101 (GCC_COLORS): Add type-diff to example.
27103 (-fdiagnostics-show-template-tree): New.
27104 (-fno-elide-type): New.
27105 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
27106 the pp_format_decoder callback. Call any m_format_postprocessor's
27108 (pretty_printer::pretty_printer): Initialize
27109 m_format_postprocessor.
27110 (pretty_printer::~pretty_printer): Delete any
27111 m_format_postprocessor.
27112 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
27113 (class format_postprocessor): New class.
27114 (struct pretty_printer::format_decoder): Document the new parameters.
27115 (struct pretty_printer::m_format_postprocessor): New field.
27116 * tree-diagnostic.c (default_tree_printer): Update for new
27117 bool and const char ** params.
27118 * tree-diagnostic.h (default_tree_printer): Likewise.
27120 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
27122 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
27123 (lwa_operand): Delete rs6000_gen_cell_microcode test.
27124 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27125 rs6000_gen_cell_microcode code.
27126 (rs6000_final_prescan_insn): Delete.
27127 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
27128 "warn-cell-microcode" entries.
27129 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
27130 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
27131 throughout. Change cc_reg_not_micro_cr0_operand to
27132 cc_reg_not_cr0_operand throughout.
27133 (*extendhi<mode>2_noload): Delete.
27134 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
27135 (mwarn-cell-microcode): Delete.
27136 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
27137 -mgen-cell-microcode and -mwarn-cell-microcode.
27139 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
27142 * config/i386/constraints.md (Yd): New constraint.
27144 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
27145 and (?Yd, r) alternatives. Update insn attributes.
27146 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
27147 and (?*Yd, r) alternatives. Update insn attributes.
27148 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
27150 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
27152 * gimplify.c (gimplify_modify_expr): Don't create a
27153 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
27156 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
27158 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
27160 2017-05-30 Richard Biener <rguenther@suse.de>
27162 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
27163 and reduc_def fields.
27164 (STMT_VINFO_REDUC_TYPE): New define.
27165 (STMT_VINFO_REDUC_DEF): Likewise.
27166 (vect_force_simple_reduction): Adjust prototype.
27167 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
27168 (vect_is_simple_reduction): Remove check_reduction argument.
27169 (vect_force_simple_reduction): Adjust and set
27170 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
27171 (vectorizable_reduction): Do not re-do reduction analysis
27172 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
27173 * tree-parloops.c (gather_scalar_reductions): Adjust.
27175 2017-05-30 Richard Biener <rguenther@suse.de>
27177 PR middle-end/80901
27178 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
27181 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
27183 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
27184 Introduce unknown_misalignment parameter and remove vf.
27185 (vect_peeling_hash_get_lowest_cost):
27186 Pass unknown_misalignment parameter.
27187 (vect_enhance_data_refs_alignment):
27188 Fix unsupportable data ref treatment.
27190 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27192 * tree-vect-data-refs.c (vect_get_data_access_cost):
27193 Workaround for SLP handling.
27194 (vect_enhance_data_refs_alignment):
27195 Compute costs for doing no peeling at all, compare to the best
27196 peeling costs so far and avoid peeling if cheaper.
27198 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27200 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
27201 Return peeling info and set costs to zero for unlimited cost
27203 (vect_enhance_data_refs_alignment): Also inspect all datarefs
27204 with unknown misalignment. Compute and costs for unknown
27205 misalignment, compare them to the costs for known misalignment
27206 and choose the cheapest for peeling.
27208 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27210 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
27211 (vect_get_peeling_costs_all_drs): Create function.
27212 (vect_peeling_hash_get_lowest_cost):
27213 Use vect_get_peeling_costs_all_drs.
27214 (vect_peeling_supportable): Create function.
27215 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
27217 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27219 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
27220 DR_HAS_NEGATIVE_STEP.
27221 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
27222 (vect_enhance_data_refs_alignment): Use.
27223 (vect_duplicate_ssa_name_ptr_info): Use.
27224 * tree-vectorizer.h (dr_misalignment): Use.
27225 (known_alignment_for_access_p): Use.
27227 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27230 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
27232 (has_section_name): New function.
27234 2017-05-30 Martin Liska <mliska@suse.cz>
27237 * auto-profile.c (get_function_decl_from_block): Fix
27240 2017-05-30 Richard Biener <rguenther@suse.de>
27242 PR middle-end/80876
27243 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
27245 2017-05-30 Martin Liska <mliska@suse.cz>
27247 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
27248 * dumpfile.h (struct dump_file_info): Remove ctors.
27250 2017-05-30 Martin Liska <mliska@suse.cz>
27252 * predict.def: Fix GNU coding style.
27254 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
27256 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
27257 Mark 'to' argument with ATTRIBUTE_UNUSED.
27259 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
27261 * config/xtensa/xtensa.c (xtensa_emit_call): Use
27262 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
27263 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
27266 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
27268 * doc/install.texi (Options specification): Restore entry of
27269 --enable-sjlj-exceptions.
27271 2017-05-27 Michael Eager <eager@eagercon.com>
27274 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
27276 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
27278 * config/microblaze/microblaze.h
27279 (FIXED_REGISTERS): Update in macro.
27280 (CALL_USED_REGISTERS): Update in macro.
27282 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
27284 * doc/install.texi: Add links to macOS binary distributions.
27286 2017-05-27 Jakub Jelinek <jakub@redhat.com>
27290 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27292 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
27294 2017-05-26 Martin Liska <mliska@suse.cz>
27296 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
27298 2017-05-26 Martin Liska <mliska@suse.cz>
27300 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
27301 always leading ';; '.
27302 (dump_bb_info): Likewise.
27303 (brief_dump_cfg): Likewise.
27304 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
27305 * dumpfile.c: Remove usage of TDF_VERBOSE.
27306 * dumpfile.h (enum dump_kind): Likewise.
27307 (dump_gimple_bb_header): Do not use TDF_COMMENT.
27308 * print-tree.c (debug_verbose): Remove.
27309 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
27310 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
27311 * tree-diagnostic.c (default_tree_printer): Replace
27312 TDF_DIAGNOSTIC with TDF_SLIM.
27314 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27316 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
27317 in parameter loop, rather than loop_vinfo.
27318 (create_intersect_range_checks): Ditto.
27319 (vect_create_cond_for_alias_checks): Update call to above functions.
27321 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27323 PR tree-optimization/80815
27324 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
27325 for merging runtime alias checks. Handle negative DR_STEPs.
27327 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27329 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
27331 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
27332 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
27333 out code pruning runtime alias checks.
27334 * tree-data-ref.c (prune_runtime_alias_test_list): New function
27335 factored out from above.
27336 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
27338 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
27340 (prune_runtime_alias_test_list): New decalaration.
27342 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27344 * tree-vect-data-refs.c (compare_tree): Rename and move ...
27345 * tree-data-ref.c (data_ref_compare_tree): ... to here.
27346 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
27347 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
27348 (operator==, comp_dr_with_seg_len_pair): Ditto.
27349 (vect_prune_runtime_alias_test_list): Ditto.
27351 2017-05-26 Martin Liska <mliska@suse.cz>
27354 * params.def: Bound partial-inlining-entry-probability param.
27356 2017-05-26 Marek Polacek <polacek@redhat.com>
27359 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
27362 2017-05-26 Richard Biener <rguenther@suse.de>
27364 PR tree-optimization/80842
27365 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
27368 2017-05-26 Richard Biener <rguenther@suse.de>
27370 PR tree-optimization/80844
27371 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
27373 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
27375 * doc/md.texi (Machine Constraints): Update x86 family
27376 machine constraints section to match 'config/i386/constraints.md'.
27378 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
27380 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
27382 2017-05-25 Nathan Sidwell <nathan@acm.org>
27384 * doc/invoke.texi (--enable-languages): Update documentation.
27386 2017-05-25 Martin Liska <mliska@suse.cz>
27388 * dumpfile.c: Add TDF_FOLDING.
27389 * dumpfile.h (enum dump_kind): Likewise.
27390 * genmatch.c (dt_simplify::gen_1): Use it.
27392 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27394 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
27396 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27398 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
27399 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
27401 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27403 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
27404 * match.pd (X == C): Rewrite it here.
27405 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
27406 with_certain_nonzero_bits2): New predicates.
27407 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
27409 2017-05-24 Nathan Sidwell <nathan@acm.org>
27411 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
27414 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
27417 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
27419 * config/powerpcspe: New port. Files are copied from the rs6000
27420 port, with "rs6000" in filenames replaced by "powerpcspe".
27422 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
27424 PR rtl-optimization/80754
27425 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
27427 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
27429 * config/sparc/sparc.md (length): Return the correct value for -mflat
27430 sibcalls to match output_sibcall.
27432 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
27436 * config/rs6000/rs6000.c (struct machine_function): Add new field
27438 (rs6000_get_separate_components): Init that field, use it.
27439 (rs6000_components_for_bb): Use the field.
27441 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
27443 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
27445 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
27447 PR middle-end/80823
27448 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
27450 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27453 * config/s390/s390.c (s390_check_qrst_address): Check incoming
27454 address against address_operand predicate.
27455 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
27457 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
27459 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
27460 parameters passed indirectly.
27462 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
27464 * config/i386/i386.md (*movdi_internal): Remove SSE4
27465 alternative 18 (?r, *v). Update insn attributes.
27466 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
27467 Update insn attributes.
27468 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
27469 Update insn attributes.
27470 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
27471 alternative 1 (r, v). Remove isa attribute.
27472 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
27473 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
27474 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
27476 2017-05-23 Tom de Vries <tom@codesourcery.com>
27478 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
27481 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27483 * cgraphunit.c (symbol_table::process_new_functions): Update.
27484 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
27485 (inline_generate_summary): Rename to ...
27486 (ipa_fn_summary_generate): ... this one.
27487 (inline_read_summary): Rename to ...
27488 (ipa_fn_summary_read): ... this one.
27489 (inline_write_summary): Rename to ...
27490 (ipa_fn_summary_write): ... this one.
27491 (inline_free_summary): Rename to ...
27492 (ipa_free_fn_summary): ... this one.
27493 (pass_data_local_fn_summary, pass_local_fn_summary,
27494 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
27495 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
27496 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
27497 make_pass_ipa_fn_summary): New.
27498 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
27499 inline_write_summary, inline_free_summary): Remove.
27500 (ipa_free_fn_summary) : New.
27501 * ipa-inline.c (ipa_inline): Update.
27502 (pass_ipa_inline): Do not generate summaries.
27503 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
27505 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
27506 and add pass_ipa_fn_summary.
27507 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
27509 (make_pass_inline_parameters): Remove.
27511 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
27513 * omp-low.c (struct omp_context): Remove "default_kind" member.
27516 * omp-offload.c (execute_oacc_device_lower): Remove the
27517 parallelism dimensions function attributes for unparallelized
27518 OpenACC kernels constructs.
27520 2017-05-23 Martin Liska <mliska@suse.cz>
27522 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
27524 (cgraph_edge::make_speculative): Likewise.
27525 (cgraph_edge::resolve_speculation): Likewise.
27526 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
27527 (cgraph_node::dump): Likewise.
27528 * cgraph.h: Likewise.
27529 * cgraphunit.c (analyze_functions): Likewise.
27530 (symbol_table::compile): Likewise.
27531 * ipa-cp.c (print_all_lattices): Likewise.
27532 (determine_versionability): Likewise.
27533 (initialize_node_lattices): Likewise.
27534 (ipcp_verify_propagated_values): Likewise.
27535 (estimate_local_effects): Likewise.
27536 (update_profiling_info): Likewise.
27537 (create_specialized_node): Likewise.
27538 (perhaps_add_new_callers): Likewise.
27539 (decide_about_value): Likewise.
27540 (decide_whether_version_node): Likewise.
27541 (identify_dead_nodes): Likewise.
27542 (ipcp_store_bits_results): Likewise.
27543 * ipa-devirt.c (dump_targets): Likewise.
27544 (ipa_devirt): Likewise.
27545 * ipa-icf.c (sem_item::dump): Likewise.
27546 (sem_function::equals): Likewise.
27547 (sem_variable::equals): Likewise.
27548 (sem_item_optimizer::read_section): Likewise.
27549 (sem_item_optimizer::execute): Likewise.
27550 (congruence_class::dump): Likewise.
27551 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
27552 (dump_inline_summary): Likewise.
27553 (estimate_node_size_and_time): Likewise.
27554 (inline_analyze_function): Likewise.
27555 * ipa-inline-transform.c (inline_call): Likewise.
27556 * ipa-inline.c (report_inline_failed_reason): Likewise.
27557 (want_early_inline_function_p): Likewise.
27558 (edge_badness): Likewise.
27559 (update_edge_key): Likewise.
27560 (inline_small_functions): Likewise.
27561 * ipa-profile.c (ipa_profile): Likewise.
27562 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
27563 (ipa_make_edge_direct_to_target): Likewise.
27564 (remove_described_reference): Likewise.
27565 (ipa_impossible_devirt_target): Likewise.
27566 (propagate_controlled_uses): Likewise.
27567 (ipa_print_node_params): Likewise.
27568 (ipcp_transform_function): Likewise.
27569 * ipa-pure-const.c (pure_const_read_summary): Likewise.
27570 (propagate_pure_const): Likewise.
27571 * ipa-reference.c (generate_summary): Likewise.
27572 (read_write_all_from_decl): Likewise.
27573 (propagate): Likewise.
27574 (ipa_reference_read_optimization_summary): Likewise.
27575 * ipa-utils.c (ipa_merge_profiles): Likewise.
27576 * ipa.c (walk_polymorphic_call_targets): Likewise.
27577 (symbol_table::remove_unreachable_nodes): Likewise.
27578 (ipa_single_use): Likewise.
27579 * passes.c (execute_todo): Likewise.
27580 * predict.c (drop_profile): Likewise.
27581 * symtab.c (symtab_node::get_dump_name): New function.
27582 (symtab_node::dump_name): Likewise.
27583 (symtab_node::dump_asm_name): Likewise.
27584 (symtab_node::dump_references): Likewise.
27585 (symtab_node::dump_referring): Likewise.
27586 (symtab_node::dump_base): Likewise.
27587 (symtab_node::debug_symtab): Likewise.
27588 * tree-sra.c (convert_callers_for_node): Likewise.
27589 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
27590 * value-prof.c (init_node_map): Likewise.
27592 2017-05-23 Martin Liska <mliska@suse.cz>
27594 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
27595 and symtab_node::debug_symtab to symbol_table::debug.
27596 * cgraphunit.c (analyze_functions): Use the renamed function.
27597 (symbol_table::compile): Likewise.
27598 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
27599 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
27600 * passes.c (execute_todo): Likewise.
27601 * symtab.c (symbol_table::dump): New function.
27602 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
27604 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27606 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
27607 that nonconst implies exec.
27609 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27611 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27612 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27613 (inline_edge_summary_vec): Turn into ...
27614 (ipa_call_summaries): ... this one.
27615 (redirect_to_unreachable, edge_set_predicate,
27616 evaluate_properties_for_edge, inline_summary_alloc,
27617 reset_ipa_call_summary, reset_inline_summary,
27618 inline_summary_t::duplicate): Update.
27619 (inline_edge_duplication_hook): Turn to ...
27620 (ipa_call_summary_t::duplicate): ... this one.
27621 (inline_edge_removal_hook): Turn to ...
27622 (ipa_call_summary_t::remove): ... this one.
27623 (dump_inline_edge_summary): Turn to ...
27624 (dump_ipa_call_summary): ... this one.
27625 (estimate_function_body_sizes): Update.
27626 (inline_update_callee_summaries): Update.
27627 (remap_edge_change_prob): Update.
27628 (remap_edge_summaries): Update.
27629 (inline_merge_summary): Update.
27630 (do_estimate_edge_time): Update.
27631 (inline_generate_summary): Update.
27632 (inline_read_section): Update.
27633 (inline_read_summary): Update.
27634 (inline_free_summary): Update.
27635 * ipa-inline.c (can_inline_edge_p): Update.
27636 (compute_inlined_call_time): Update.
27637 (want_inline_small_function_p): Update.
27638 (edge_badness): Update.
27639 (early_inliner): Update.
27640 * ipa-inline.h (inline_edge_summary): Turn to ...
27641 (ipa_call_summary): ... this one.
27642 (ipa_call_summary_t): New class.
27643 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27644 (ipa_call_summaries): New.
27645 (inline_edge_summary): Remove.
27646 (estimate_edge_growth): Update.
27647 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27648 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27649 * ipa-split.c (execute_split_functions): Update.
27650 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27652 2017-05-23 Tom de Vries <tom@codesourcery.com>
27654 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27655 attributes): Document rdrand effective target.
27657 2017-05-23 Tom de Vries <tom@codesourcery.com>
27659 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27660 attributes): Sort alphabetically.
27662 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
27664 * config/avr/genmultilib.awk: Use gsub instead of gensub.
27666 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
27669 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
27670 V2DF/V2DI splat into two separate patterns, one that handles
27671 registers, and the other that only handles memory. Drop support
27672 for splatting from a GPR on ISA 2.07 and then splitting the
27673 splat into direct move and splat.
27674 (vsx_splat_<mode>_reg): Likewise.
27675 (vsx_splat_<mode>_mem): Likewise.
27677 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
27679 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
27681 2017-05-22 Jakub Jelinek <jakub@redhat.com>
27683 PR middle-end/80809
27684 * omp-low.c (finish_taskreg_remap): New function.
27685 (finish_taskreg_scan): If unit size of ctx->record_type
27686 is non-constant, unshare the size expression and replace
27687 decls in it with possible outer var refs.
27689 PR middle-end/80809
27690 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
27691 GOVD_SHARED rather than GOVD_PRIVATE with it.
27692 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
27693 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
27695 PR middle-end/80853
27696 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
27697 as last argument to build_outer_var_ref for pointer bases of array
27698 section reductions.
27700 2017-05-19 Martin Sebor <msebor@redhat.com>
27702 * print-tree.c (print_node): Print DECL_READ_P flag.
27704 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27706 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
27707 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
27708 * cgraph.c: Likewise.
27709 * cgraphunit.c: Likewise.
27710 * gengtype.c: Likewise.
27711 * ipa-cp.c: Likewise.
27712 * ipa-devirt.c: Likewise.
27713 * ipa-icf.c: Likewise.
27714 * ipa-predicate.c: Likewise.
27715 * ipa-profile.c: Likewise.
27716 * ipa-prop.c: Likewise.
27717 * ipa-split.c: Likewise.
27719 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
27720 edge_predicate_pool, dump_inline_hints,
27721 inline_summary::account_size_time, redirect_to_unreachable,
27722 edge_set_predicate, set_hint_predicate,
27723 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
27724 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
27725 inline_summary_t::remove, remap_hint_predicate_after_duplication,
27726 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
27727 ipa_call_summary_t::remove, initialize_growth_caches,
27728 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
27729 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
27730 mark_modified, unmodified_parm_1, unmodified_parm,
27731 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
27732 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
27733 compute_bb_predicates, will_be_nonconstant_expr_predicate,
27734 will_be_nonconstant_predicate, record_modified_bb_info,
27735 get_minimal_bb, record_modified, param_change_prob,
27736 phi_result_unknown_predicate, predicate_for_phi_result,
27737 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
27738 estimate_function_body_sizes, compute_inline_parameters,
27739 compute_inline_parameters_for_curren, pass_data_inline_parameters,
27740 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
27741 inline_update_callee_summaries, remap_edge_change_prob,
27742 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
27743 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
27744 inline_analyze_function, inline_summary_t::insert,
27745 inline_generate_summary, read_ipa_call_summary, inline_read_section,
27746 inline_read_summary, write_ipa_call_summary, inline_write_summary,
27747 inline_free_summary): Move to ipa-fnsummary.h
27748 (predicate_t): Remove.
27749 * ipa-fnsummary.c: New file.
27750 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
27751 (enum inline_hints_vals, inline_hints, agg_position_info,
27752 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
27753 inline_summaries, ipa_call_summary, ipa_call_summary_t,
27754 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
27755 dump_inline_summary, dump_inline_hints, inline_generate_summary,
27756 inline_read_summary, inline_write_summary, inline_free_summary,
27757 inline_analyze_function, initialize_inline_failed,
27758 inline_merge_summary, inline_update_overall_summary,
27759 compute_inline_parameters): Move to ipa-fnsummary.h
27760 * ipa-fnsummary.h: New file.
27761 * ipa-inline-transform.h: Include ipa-inline.h.
27762 * ipa-inline.c: LIkewise.
27764 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27766 * ipa-inline.c (edge_badness): Use inlined_time instead of
27767 inline_summaries->get.
27769 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27771 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
27773 2017-05-22 Nathan Sidwell <nathan@acm.org>
27775 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
27776 (fdump-lang): Document 'raw' option.
27777 * dumpfile.h (TDI_tu): Delete.
27778 * dumpfile.c (dump_files): Remove translation-unit.
27779 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
27781 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
27783 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
27784 command option from $(AWK) call.
27785 * config/avr/genmultilib.awk: Simplify and rewrite so that it
27786 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
27787 [FORMAT]: Remove handling of variable.
27788 * config/avr/t-multilib: Regenerate.
27790 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27792 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
27794 (dump_inline_summary): Do not print self_time.
27795 (estimate_function_body_sizes): Do not set self_time.
27796 (compute_inline_parameters): Likewise.
27797 (inline_read_section, inline_write_summary): Do not stream self_time.
27798 * ipa-inline.h (inline_summary): Drop self_time.
27800 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27802 * ipa-inline-analysis.c (account_size_time): Rename to ...
27803 (inline_summary::account_size_time): ... this one.
27804 (reset_ipa_call_summary): Turn to ...
27805 (ipa_call_summary::reset): ... this one.
27806 (reset_inline_summary): Turn to ...
27807 (inline_summary::reset): ... this one.
27808 (inline_summary_t::remove): Update.
27809 (inline_summary_t::duplicate): Update.
27810 (ipa_call_summary_t::remove): Update.
27811 (dump_inline_summary): Update.
27812 (estimate_function_body_sizes): Update.
27813 (compute_inline_parameters): Update.
27814 (estimate_node_size_and_time): Update.
27815 (inline_merge_summary): Update.
27816 (inline_update_overall_summary): Update.
27817 (inline_read_section): Update.
27818 (inline_write_summary): Update.
27819 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
27820 add account_size_time and reset member functions.
27821 (ipa_call_summary): Add reset function.
27822 * ipa-predicate.h (predicate::operator &): Constify.
27824 2017-05-22 Richard Biener <rguenther@suse.de>
27826 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
27828 2017-05-19 Jason Merrill <jason@redhat.com>
27830 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
27832 2017-05-19 Marek Polacek <polacek@redhat.com>
27835 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
27836 TYPE_OVERFLOW_WRAPS checks.
27838 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
27840 * tree-core.h (enum omp_clause_default_kind): Add
27841 "OMP_CLAUSE_DEFAULT_PRESENT".
27842 * tree-pretty-print.c (dump_omp_clause): Handle it.
27843 * gimplify.c (enum gimplify_omp_var_data): Add
27844 "GOVD_MAP_FORCE_PRESENT".
27845 (gimplify_adjust_omp_clauses_1): Map it to
27846 "GOMP_MAP_FORCE_PRESENT".
27847 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
27849 * gimplify.c (oacc_default_clause): Clarify.
27851 2017-05-19 Nathan Sidwell <nathan@acm.org>
27853 LANG_HOOK_REGISTER_DUMPS
27854 * toplev.c (general_init): Call register dump lang hook.
27855 * doc/invoke.texi: Document -fdump-lang option family.
27856 * dumpfile.c (dump_files): Remove class dump here.
27857 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
27858 * dumpfile.h (tree_dump_index): Remove TDI_class.
27859 * langhooks-def.h (lhd_register_dumps): Declare.
27860 (LANG_HOOKS_REGISTER_DUMPS): Define.
27861 (LANG_HOOKS_INITIALIZER): Add it.
27862 * langhooks.c (lhd_register_dumps): Define.
27863 * langhooks.h (struct lang_hooks): Add register_dumps.
27865 2017-05-19 Nathan Sidwell <nathan@acm.org>
27867 * context.h (context::set_passes): New.
27868 * context.c (context::context): Do not create pass manager.
27869 * toplev.c (general_init): Create pass manager here.
27871 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
27873 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
27874 use this splitter if two add or or instructions would also work for
27875 the constant we want to generate.
27877 2017-05-19 Richard Biener <rguenther@suse.de>
27880 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
27881 predicate evaluation.
27883 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27885 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
27887 * ipa-inline.c (want_inline_small_function_p): Do not cast to
27890 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27892 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27893 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27894 (inline_edge_summary_vec): Turn into ...
27895 (ipa_call_summaries): ... this one.
27896 (redirect_to_unreachable, edge_set_predicate,
27897 evaluate_properties_for_edge, inline_summary_alloc,
27898 reset_ipa_call_summary, reset_inline_summary,
27899 inline_summary_t::duplicate): Update.
27900 (inline_edge_duplication_hook): Turn to ...
27901 (ipa_call_summary_t::duplicate): ... this one.
27902 (inline_edge_removal_hook): Turn to ...
27903 (ipa_call_summary_t::remove): ... this one.
27904 (dump_inline_edge_summary): Turn to ...
27905 (dump_ipa_call_summary): ... this one.
27906 (estimate_function_body_sizes): Update.
27907 (inline_update_callee_summaries): Update.
27908 (remap_edge_change_prob): Update.
27909 (remap_edge_summaries): Update.
27910 (inline_merge_summary): Update.
27911 (do_estimate_edge_time): Update.
27912 (inline_generate_summary): Update.
27913 (inline_read_section): Update.
27914 (inline_read_summary): Update.
27915 (inline_free_summary): Update.
27916 * ipa-inline.c (can_inline_edge_p): Update.
27917 (compute_inlined_call_time): Update.
27918 (want_inline_small_function_p): Update.
27919 (edge_badness): Update.
27920 (early_inliner): Update.
27921 * ipa-inline.h (inline_edge_summary): Turn to ...
27922 (ipa_call_summary): ... this one.
27923 (ipa_call_summary_t): New class.
27924 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27925 (ipa_call_summaries): New.
27926 (inline_edge_summary): Remove.
27927 (estimate_edge_growth): Update.
27928 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27929 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27930 * ipa-split.c (execute_split_functions): Update.
27931 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27933 2017-05-19 Richard Biener <rguenther@suse.de>
27935 PR middle-end/80764
27936 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
27938 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
27940 * config/rs6000/rs6000.c (struct machine_function): Add field
27941 fpr_is_wrapped_separately.
27942 (rs6000_get_separate_components): Use 64 components. Handle the
27943 new FPR components.
27944 (rs6000_components_for_bb): Handle the FPR components.
27945 (rs6000_emit_prologue_components): Handle the FPR components.
27946 (rs6000_emit_epilogue_components): Handle the FPR components.
27947 (rs6000_set_handled_components): Handle the FPR components.
27948 (rs6000_emit_prologue): Don't output prologue code for those FPRs
27949 that are already separately shrink-wrapped.
27950 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
27951 that are already separately shrink-wrapped.
27953 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27956 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
27959 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
27960 (define_peephole2 for Altivec d-form load): Add peepholes to catch
27961 cases where the register allocator uses a move and an offsettable
27962 memory operation to/from a FPR register on ISA 2.06/2.07.
27963 (define_peephole2 for Altivec d-form store): Likewise.
27965 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
27968 * config/i386/mmx.md (*mov<mode>_internal): Enable
27969 alternatives 11, 12, 13 and 14 also for 32bit targets.
27970 Remove alternatives 15, 16, 17 and 18.
27971 * config/i386/sse.md (vec_concatv2di): Change
27972 alternative (!x, *y) to (x, ?!*Yn).
27974 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
27976 * dumpfile.h (enum dump_kind): Remove stray comma.
27978 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27980 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
27981 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
27982 predicate::num_conditions
27983 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
27984 (CHANGED): turn into predicate::changed.
27985 (agg_position_info): Move to ipa-predicate.h
27986 (add_condition, predicate::add_clause, predicate::operator &=,
27987 predicate::or_with, predicate::evaluate, predicate::probability,
27988 dump_condition, dump_clause, predicate::dump,
27989 predicate::remap_after_duplication, predicate::remap_after_inlining,
27990 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
27991 (evaluate_conditions_for_known_args): Update.
27992 (set_cond_stmt_execution_predicate): Update.
27993 * ipa-inline.h: Include ipa-predicate.h
27994 (condition, inline_param_summary, conditions, agg_position_info,
27995 predicate): Move to ipa-predicate.h
27996 * ipa-predicate.c: New file.
27997 * ipa-predicate.h: New file.
27999 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
28001 * final.c (leaf_function_p): Check we are not in a sequence.
28003 2017-05-18 Martin Liska <mliska@suse.cz>
28005 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
28006 * dumpfile.c (dump_register): Use new enum dump_kind.
28007 (get_dump_file_name): Likewise.
28008 (dump_enable_all): Likewise.
28009 (dump_switch_p_1): Likewise.
28010 (enable_rtl_dump_file): Remove usage of TDF_RTL.
28011 * dumpfile.h (enum dump_kind): New enum type.
28012 (struct dump_file_info): Create constructor and
28013 format fields and comments.
28014 * passes.c (pass_manager::register_one_dump_file):
28016 * statistics.c (statistics_early_init): Likewise.
28017 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
28018 TDF_TREE with TDF_SLIM.
28019 (gather_memory_references_ref): Likewise.
28021 2017-05-18 Martin Liska <mliska@suse.cz>
28023 * vec.h (struct vnull): Use it.
28025 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
28027 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
28028 (true_predicate, false_predicate, true_predicate_p,
28029 false_predicate_p): Remove.
28030 (single_cond_predicate, not_inlined_predicate): Turn to member function
28032 (add_condition): Update.
28033 (add_clause): Turn to...
28034 (predicate::add_clause): ... this one; update; allow passing NULL
28036 (and_predicates): Turn to ...
28037 (predicate::operator &=): ... this one.
28038 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
28039 (or_predicates): Turn to ...
28040 (predicate::or_with): ... this one.
28041 (evaluate_predicate): Turn to ...
28042 (predicate::evaluate): ... this one.
28043 (predicate_probability): Turn to ...
28044 (predicate::probability): ... this one.
28045 (dump_condition): Update.
28046 (dump_predicate): Turn to ...
28047 (predicate::dump): ... this one.
28048 (account_size_time): Update.
28049 (edge_set_predicate): Update.
28050 (set_hint_predicate): UPdate.
28051 (evaluate_conditions_for_known_args): Update.
28052 (evaluate_properties_for_edge): Update.
28053 (remap_predicate_after_duplication): Turn to...
28054 (predicate::remap_after_duplication): ... this one.
28055 (remap_hint_predicate_after_duplication): Update.
28056 (inline_summary_t::duplicate): UPdate.
28057 (dump_inline_edge_summary): Update.
28058 (dump_inline_summary): Update.
28059 (set_cond_stmt_execution_predicate): Update.
28060 (set_switch_stmt_execution_predicate): Update.
28061 (compute_bb_predicates): Update.
28062 (will_be_nonconstant_expr_predicate): Update.
28063 (will_be_nonconstant_predicate): Update.
28064 (phi_result_unknown_predicate): Update.
28065 (predicate_for_phi_result): Update.
28066 (array_index_predicate): Update.
28067 (estimate_function_body_sizes): Update.
28068 (estimate_node_size_and_time): Update.
28069 (estimate_ipcp_clone_size_and_time): Update.
28070 (remap_predicate): Rename to ...
28071 (predicate::remap_after_inlining): ... this one.
28072 (remap_hint_predicate): Update.
28073 (inline_merge_summary): Update.
28074 (inline_update_overall_summary): Update.
28075 (estimate_size_after_inlining): Update.
28076 (read_predicate): Rename to ...
28077 (predicate::stream_in): ... this one.
28078 (read_inline_edge_summary): Update.
28079 (write_predicate): Rename to ...
28080 (predicate::stream_out): ... this one.
28081 (write_inline_edge_summary): Update.
28082 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
28083 (clause_t): Turn to uint32_t
28084 (predicate): Turn to class; implement constructor and operators
28086 (size_time_entry): Update.
28087 (inline_summary): Update.
28088 (inline_edge_summary): Update.
28090 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
28092 * fold-const.c (fold_binary_loc): Move transformation...
28093 * match.pd (C - X CMP X): ... here.
28095 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
28097 * config/sparc/sparc.c (sparc_option_override): Set function
28098 alignment for -mcpu=niagara7 to 64 to match the I$ line.
28099 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
28101 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
28103 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
28105 2017-05-18 Marek Polacek <polacek@redhat.com>
28108 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
28109 (pass_ubsan::execute): Call gimple_assign_single_p instead of
28110 gimple_assign_load_p.
28112 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
28114 PR middle-end/80692
28115 * real.c (do_compare): Give decimal_do_compare preference over
28116 comparing just the signs.
28118 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
28120 * doc/md.texi (Canonicalization of Instructions): Describe the
28121 canonical form of instructions that inherently set a condition
28124 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
28126 PR middle-end/80775
28127 * tree-cfg.c: Move deletion of unreachable case statements to after
28128 the merging of consecutive case labels.
28130 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
28132 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
28133 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
28134 restoring of callee-saved registers.
28136 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
28138 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
28139 * config/visium/visium.c (single_set_and_flags): Likewise.
28140 * config/visium/visium.md (Substitutions): Likewise.
28142 2017-05-17 Martin Liska <mliska@suse.cz>
28144 * cfg.c: Introduce dump_flags_t type and
28145 use it instead of int type.
28147 * cfghooks.c: Likewise.
28148 * cfghooks.h (struct cfg_hooks): Likewise.
28149 * cfgrtl.c: Likewise.
28150 * cfgrtl.h: Likewise.
28151 * cgraph.c (cgraph_node::get_body): Likewise.
28152 * coretypes.h: Likewise.
28153 * domwalk.c: Likewise.
28154 * domwalk.h: Likewise.
28155 * dumpfile.c (struct dump_option_value_info): Likewise.
28156 (dump_enable_all): Likewise.
28157 (dump_switch_p_1): Likewise.
28158 (opt_info_switch_p): Likewise.
28159 * dumpfile.h (enum tree_dump_index): Likewise.
28160 (struct dump_file_info): Likewise.
28161 * genemit.c: Likewise.
28162 * generic-match-head.c: Likewise.
28163 * gengtype.c (open_base_files): Likewise.
28164 * gimple-pretty-print.c: Likewise.
28165 * gimple-pretty-print.h: Likewise.
28166 * graph.c (print_graph_cfg): Likewise.
28167 * graphite-scop-detection.c (dot_all_sese): Likewise.
28168 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
28169 * loop-unroll.c (report_unroll): Likewise.
28170 * passes.c (pass_manager::register_one_dump_file): Likewise.
28171 * print-tree.c: Likewise.
28172 * statistics.c: Likewise.
28173 * tree-cfg.c: Likewise.
28174 * tree-cfg.h: Likewise.
28175 * tree-dfa.c: Likewise.
28176 * tree-dfa.h: Likewise.
28177 * tree-dump.c (dump_function): Likewise.
28178 * tree-dump.h (struct dump_info): Likewise.
28179 * tree-pretty-print.c: Likewise.
28180 * tree-pretty-print.h: Likewise.
28181 * tree-ssa-live.c: Likewise.
28182 * tree-ssa-live.h: Likewise.
28183 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
28184 * tree-vect-loop.c: Likewise.
28185 * tree-vect-slp.c: Likewise.
28187 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
28188 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28190 PR tree-optimization/80457
28191 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
28192 of all arguments to a statement as scalar_to_vec operations.
28193 (vectorizable_call): Adjust call to vect_model_simple_cost for
28195 (vectorizable_conversion): Likewise.
28196 (vectorizable_assignment): Likewise.
28197 (vectorizable_shift): Likewise.
28198 (vectorizable_operation): Likewise.
28199 (vectorizable_comparison): Likewise.
28200 (vect_is_simple_cond): Record the def types for operands.
28201 (vectorizable_condition): Likewise, call vect_model_simple_cost.
28202 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
28203 for statement argument count.
28205 2017-05-16 Carl Love <cel@us.ibm.com>
28207 * config/rs6000/rs6000-c: Add support for built-in functions
28208 vector unsigned long long vec_bperm (vector unsigned long long,
28209 vector unsigned char)
28210 vector signed long long vec_mule (vector signed int,
28212 vector unsigned long long vec_mule (vector unsigned int,
28213 vector unsigned int)
28214 vector signed long long vec_mulo (vector signed int,
28216 vector unsigned long long vec_mulo (vector unsigned int,
28217 vector unsigned int)
28218 vector signed char vec_sldw (vector signed char,
28219 vector signed char,
28221 vector unsigned char vec_sldw (vector unsigned char,
28222 vector unsigned char,
28224 vector signed short vec_sldw (vector signed short,
28225 vector signed short,
28227 vector unsigned short vec_sldw (vector unsigned short,
28228 vector unsigned short,
28230 vector signed int vec_sldw (vector signed int,
28233 vector unsigned int vec_sldw (vector unsigned int,
28234 vector unsigned int,
28236 vector signed long long vec_sldw (vector signed long long,
28237 vector signed long long,
28239 vector unsigned long long vec_sldw (vector unsigned long long,
28240 vector unsigned long long,
28242 * config/rs6000/rs6000-c: Add support for built-in functions
28243 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
28244 * config/rs6000/altivec.h: Add defintion for vec_sldw.
28245 * doc/extend.texi: Update the built-in documentation for the
28246 new built-in functions.
28248 2017-05-16 Marek Polacek <polacek@redhat.com>
28252 * tree.c (save_expr): Don't fold the expression.
28254 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
28256 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
28257 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
28258 and (?*y,m). Update insn attributes.
28260 2017-05-16 Martin Liska <mliska@suse.cz>
28262 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
28263 flags argument of print_gimple_stmt, print_gimple_expr,
28264 print_generic_stmt and print_generic_expr.
28265 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
28266 * coretypes.h: Likewise.
28267 * except.c (dump_eh_tree): Likewise.
28268 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
28269 * gimple-pretty-print.h: Likewise.
28270 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
28271 (backprop::push_to_worklist): Likewise.
28272 (backprop::pop_from_worklist): Likewise.
28273 (backprop::process_use): Likewise.
28274 (backprop::intersect_uses): Likewise.
28275 (note_replacement): Likewise.
28276 * gimple-ssa-store-merging.c
28277 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
28278 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
28279 (pass_store_merging::execute): Likewise.
28280 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
28281 (ssa_base_cand_dump_callback): Likewise.
28282 (dump_incr_vec): Likewise.
28283 (replace_refs): Likewise.
28284 (replace_mult_candidate): Likewise.
28285 (create_add_on_incoming_edge): Likewise.
28286 (create_phi_basis): Likewise.
28287 (insert_initializers): Likewise.
28288 (all_phi_incrs_profitable): Likewise.
28289 (introduce_cast_before_cand): Likewise.
28290 (replace_one_candidate): Likewise.
28291 * gimplify.c (gimplify_expr): Likewise.
28292 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
28293 (set_rename): Likewise.
28294 (rename_uses): Likewise.
28295 (copy_loop_phi_nodes): Likewise.
28296 (add_close_phis_to_merge_points): Likewise.
28297 (copy_loop_close_phi_args): Likewise.
28298 (copy_cond_phi_args): Likewise.
28299 (graphite_copy_stmts_from_block): Likewise.
28300 (translate_pending_phi_nodes): Likewise.
28301 * graphite-poly.c (print_pdr): Likewise.
28302 (dump_gbb_cases): Likewise.
28303 (dump_gbb_conditions): Likewise.
28304 (print_scop_params): Likewise.
28305 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
28306 (build_cross_bb_scalars_use): Likewise.
28307 (gather_bbs::before_dom_children): Likewise.
28308 * hsa-dump.c (dump_hsa_immed): Likewise.
28309 * ipa-cp.c (print_ipcp_constant_value): Likewise.
28310 (get_replacement_map): Likewise.
28311 * ipa-inline-analysis.c (dump_condition): Likewise.
28312 (estimate_function_body_sizes): Likewise.
28313 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
28314 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
28315 * ipa-prop.c (ipa_dump_param): Likewise.
28316 (ipa_print_node_jump_functions_for_edge): Likewise.
28317 (ipa_modify_call_arguments): Likewise.
28318 (ipa_modify_expr): Likewise.
28319 (ipa_dump_param_adjustments): Likewise.
28320 (ipa_dump_agg_replacement_values): Likewise.
28321 (ipcp_modif_dom_walker::before_dom_children): Likewise.
28322 * ipa-pure-const.c (check_stmt): Likewise.
28323 (pass_nothrow::execute): Likewise.
28324 * ipa-split.c (execute_split_functions): Likewise.
28325 * omp-offload.c (dump_oacc_loop_part): Likewise.
28326 (dump_oacc_loop): Likewise.
28327 * trans-mem.c (tm_log_emit): Likewise.
28328 (tm_memopt_accumulate_memops): Likewise.
28329 (dump_tm_memopt_set): Likewise.
28330 (dump_tm_memopt_transform): Likewise.
28331 * tree-cfg.c (gimple_verify_flow_info): Likewise.
28332 (print_loop): Likewise.
28333 * tree-chkp-opt.c (chkp_print_addr): Likewise.
28334 (chkp_gather_checks_info): Likewise.
28335 (chkp_get_check_result): Likewise.
28336 (chkp_remove_check_if_pass): Likewise.
28337 (chkp_use_outer_bounds_if_possible): Likewise.
28338 (chkp_reduce_bounds_lifetime): Likewise.
28339 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
28340 (chkp_mark_completed_bounds): Likewise.
28341 (chkp_register_incomplete_bounds): Likewise.
28342 (chkp_mark_invalid_bounds): Likewise.
28343 (chkp_maybe_copy_and_register_bounds): Likewise.
28344 (chkp_build_returned_bound): Likewise.
28345 (chkp_get_bound_for_parm): Likewise.
28346 (chkp_build_bndldx): Likewise.
28347 (chkp_get_bounds_by_definition): Likewise.
28348 (chkp_generate_extern_var_bounds): Likewise.
28349 (chkp_get_bounds_for_decl_addr): Likewise.
28350 * tree-chrec.c (chrec_apply): Likewise.
28351 * tree-data-ref.c (dump_data_reference): Likewise.
28352 (dump_subscript): Likewise.
28353 (dump_data_dependence_relation): Likewise.
28354 (analyze_overlapping_iterations): Likewise.
28355 * tree-inline.c (expand_call_inline): Likewise.
28356 (tree_function_versioning): Likewise.
28357 * tree-into-ssa.c (dump_defs_stack): Likewise.
28358 (dump_currdefs): Likewise.
28359 (dump_names_replaced_by): Likewise.
28360 (dump_update_ssa): Likewise.
28361 (update_ssa): Likewise.
28362 * tree-object-size.c (pass_object_sizes::execute): Likewise.
28363 * tree-parloops.c (build_new_reduction): Likewise.
28364 (try_create_reduction_list): Likewise.
28365 (ref_conflicts_with_region): Likewise.
28366 (oacc_entry_exit_ok_1): Likewise.
28367 (oacc_entry_exit_single_gang): Likewise.
28368 * tree-pretty-print.h: Likewise.
28369 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
28370 (get_scalar_evolution): Likewise.
28371 (add_to_evolution): Likewise.
28372 (get_loop_exit_condition): Likewise.
28373 (analyze_evolution_in_loop): Likewise.
28374 (analyze_initial_condition): Likewise.
28375 (analyze_scalar_evolution): Likewise.
28376 (instantiate_scev): Likewise.
28377 (number_of_latch_executions): Likewise.
28378 (gather_chrec_stats): Likewise.
28379 (final_value_replacement_loop): Likewise.
28380 (scev_const_prop): Likewise.
28381 * tree-sra.c (dump_access): Likewise.
28382 (disqualify_candidate): Likewise.
28383 (create_access): Likewise.
28384 (reject): Likewise.
28385 (maybe_add_sra_candidate): Likewise.
28386 (create_access_replacement): Likewise.
28387 (analyze_access_subtree): Likewise.
28388 (analyze_all_variable_accesses): Likewise.
28389 (sra_modify_assign): Likewise.
28390 (initialize_constant_pool_replacements): Likewise.
28391 (find_param_candidates): Likewise.
28392 (decide_one_param_reduction): Likewise.
28393 (replace_removed_params_ssa_names): Likewise.
28394 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
28395 * tree-ssa-copy.c (dump_copy_of): Likewise.
28396 (copy_prop_visit_cond_stmt): Likewise.
28397 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
28398 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
28399 (record_equivalences_from_stmt): Likewise.
28400 * tree-ssa-dse.c (compute_trims): Likewise.
28401 (delete_dead_call): Likewise.
28402 (delete_dead_assignment): Likewise.
28403 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
28404 (forward_propagate_into_cond): Likewise.
28405 (pass_forwprop::execute): Likewise.
28406 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
28407 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
28409 (move_computations_worker): Likewise.
28410 (execute_sm): Likewise.
28411 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
28412 (remove_exits_and_undefined_stmts): Likewise.
28413 (remove_redundant_iv_tests): Likewise.
28414 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
28415 (adjust_iv_update_pos): Likewise.
28416 * tree-ssa-math-opts.c (bswap_replace): Likewise.
28417 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
28418 (value_replacement): Likewise.
28419 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
28420 * tree-ssa-pre.c (print_pre_expr): Likewise.
28421 (get_representative_for): Likewise.
28422 (create_expression_by_pieces): Likewise.
28423 (insert_into_preds_of_block): Likewise.
28424 (eliminate_insert): Likewise.
28425 (eliminate_dom_walker::before_dom_children): Likewise.
28426 (eliminate): Likewise.
28427 (remove_dead_inserted_code): Likewise.
28428 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
28429 * tree-ssa-reassoc.c (get_rank): Likewise.
28430 (eliminate_duplicate_pair): Likewise.
28431 (eliminate_plus_minus_pair): Likewise.
28432 (eliminate_not_pairs): Likewise.
28433 (undistribute_ops_list): Likewise.
28434 (eliminate_redundant_comparison): Likewise.
28435 (update_range_test): Likewise.
28436 (optimize_range_tests_var_bound): Likewise.
28437 (optimize_vec_cond_expr): Likewise.
28438 (rewrite_expr_tree): Likewise.
28439 (rewrite_expr_tree_parallel): Likewise.
28440 (linearize_expr): Likewise.
28441 (break_up_subtract): Likewise.
28442 (linearize_expr_tree): Likewise.
28443 (attempt_builtin_powi): Likewise.
28444 (attempt_builtin_copysign): Likewise.
28445 (transform_stmt_to_copy): Likewise.
28446 (transform_stmt_to_multiply): Likewise.
28447 (dump_ops_vector): Likewise.
28448 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
28449 (print_scc): Likewise.
28450 (set_ssa_val_to): Likewise.
28451 (visit_reference_op_store): Likewise.
28452 (visit_use): Likewise.
28453 (sccvn_dom_walker::before_dom_children): Likewise.
28454 (run_scc_vn): Likewise.
28455 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
28457 (expr_hash_elt::print): Likewise.
28458 (const_and_copies::pop_to_marker): Likewise.
28459 (const_and_copies::record_const_or_copy_raw): Likewise.
28460 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
28461 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
28462 (dump_predicates): Likewise.
28463 (find_uninit_use): Likewise.
28464 (warn_uninitialized_phi): Likewise.
28465 (pass_late_warn_uninitialized::execute): Likewise.
28466 * tree-ssa.c (verify_vssa): Likewise.
28467 (verify_ssa): Likewise.
28468 (maybe_optimize_var): Likewise.
28469 * tree-vrp.c (dump_value_range): Likewise.
28470 (dump_all_value_ranges): Likewise.
28471 (dump_asserts_for): Likewise.
28472 (register_edge_assert_for_2): Likewise.
28473 (vrp_visit_cond_stmt): Likewise.
28474 (vrp_visit_switch_stmt): Likewise.
28475 (vrp_visit_stmt): Likewise.
28476 (vrp_visit_phi_node): Likewise.
28477 (simplify_cond_using_ranges_1): Likewise.
28478 (fold_predicate_in): Likewise.
28479 (evrp_dom_walker::before_dom_children): Likewise.
28480 (evrp_dom_walker::push_value_range): Likewise.
28481 (evrp_dom_walker::pop_value_range): Likewise.
28482 (execute_early_vrp): Likewise.
28484 2017-05-16 Richard Biener <rguenther@suse.de>
28486 * dwarf2out.c (loc_list_from_tree_1): Do not create
28487 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
28489 2017-05-16 Richard Biener <rguenther@suse.de>
28491 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
28493 (note_variable_value_in_expr): If we resolved the decl ref
28494 do not push to the stack.
28496 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
28498 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
28499 operations in fast-math mode.
28500 (vaddq_f16): Likewise.
28501 (vmul_f16): Likewise.
28502 (vmulq_f16): Likewise.
28503 (vsub_f16): Likewise.
28504 (vsubq_f16): Likewise.
28505 * config/arm/neon.md (add<mode>3): New.
28507 (fma:<VH:mode>3): New. Also remove outdated comment.
28510 2017-05-16 Martin Liska <mliska@suse.cz>
28514 * ipa-devirt.c (warn_types_mismatch): Fix typo.
28515 (odr_types_equivalent_p): Likewise.
28517 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
28519 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
28521 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
28524 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
28525 non-interunit SSE move alternatives with '?'.
28526 (zero-extendsidi peephole2): New peephole to skip intermediate
28527 general register in SSE zero-extend sequence.
28529 2017-05-15 Jeff Law <law@redhat.com>
28531 * reorg.c (relax_delay_slots): Create a new variable to hold
28532 the temporary target rather than clobbering TARGET_LABEL.
28534 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
28535 missing argument to extract_bit_field call.
28536 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
28538 2017-05-15 Martin Liska <mliska@suse.cz>
28541 * gcc.c (process_command): Do not allow empty argument of -o option.
28543 2017-05-15 Renlin Li <renlin.li@arm.com>
28545 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
28546 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
28547 * config/aarch64/constraints.md (Usf): Add long call check.
28548 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
28549 (call_value): Likewise.
28550 (sibcall): Likewise.
28551 (sibcall_value): Likewise.
28553 (call_value_insn): New.
28554 (sibcall_insn): Update rtx pattern.
28555 (sibcall_value_insn): Likewise.
28556 (call_internal): Remove.
28557 (call_value_internal): Likewise.
28558 (sibcall_internal): Likewise.
28559 (sibcall_value_internal): Likewise.
28560 (call_reg): Likewise.
28561 (call_symbol): Likewise.
28562 (call_value_reg): Likewise.
28563 (call_value_symbol): Likewise.
28565 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
28568 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
28570 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
28572 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
28573 compatible with CCGOCmode and with CCZmode.
28575 2017-05-14 Martin Sebor <msebor@redhat.com>
28577 PR middle-end/77671
28578 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
28579 (gimple_fold_builtin_snprintf): Same.
28580 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
28581 (gimple_fold_builtin_snprintf): Same.
28582 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
28583 of character types.
28584 (is_call_safe): New function.
28585 (try_substitute_return_value): Call it.
28586 (try_simplify_call): New function.
28587 (pass_sprintf_length::handle_gimple_call): Call it.
28589 2017-05-14 Martin Sebor <msebor@redhat.com>
28591 PR middle-end/80669
28592 * builtins.c (expand_builtin_stpncpy): Simplify.
28594 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
28596 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
28597 * config/i386/i386.h
28598 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
28599 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
28600 (struct machine_function): Add new members call_ms2sysv,
28601 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
28602 (struct machine_frame_state): New fields sp_realigned and
28603 sp_realigned_offset.
28604 * config/i386/i386.c
28605 (enum xlogue_stub): New enum.
28606 (enum xlogue_stub_sets): New enum.
28607 (class xlogue_layout): New class.
28608 (struct ix86_frame): New fields stack_realign_allocate_offset,
28609 stack_realign_offset and outlined_save_offset. Modify comments to
28610 detail stack layout when using out-of-line stubs.
28611 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
28612 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
28613 -mcall-ms2sysv-xlogues.
28614 (stub_managed_regs): New static variable.
28615 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
28616 registers managed by out-of-line stub.
28617 (disable_call_ms2sysv_xlogues): New function.
28618 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
28619 m->call_ms2sysv when appropriate and compute frame layout for
28621 (sp_valid_at, fp_valid_at): New inline functions.
28622 (choose_basereg): New function.
28623 (choose_baseaddr): Add align parameter, use choose_basereg and modify
28625 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
28626 Use align parameter of choose_baseaddr to generated aligned SSE movs
28628 (pro_epilogue_adjust_stack): Modify to track
28629 machine_frame_state::sp_realigned.
28630 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
28631 (ix86_nsaved_sseregs): Likewise.
28632 (ix86_emit_save_regs): Likewise.
28633 (ix86_emit_save_regs_using_mov): Likewise.
28634 (ix86_emit_save_sse_regs_using_mov): Likewise.
28635 (get_scratch_register_on_entry): Likewise.
28636 (gen_frame_set): New function.
28637 (gen_frame_load): Likewise.
28638 (gen_frame_store): Likewise.
28639 (emit_outlined_ms2sysv_save): Likewise.
28640 (emit_outlined_ms2sysv_restore): Likewise.
28641 (ix86_expand_prologue): Modify stack re-alignment code and call
28642 emit_outlined_ms2sysv_save when appropriate.
28643 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
28644 parameter rtx_insn *insn, which allows the function to be used to only
28645 generate the notes.
28646 (ix86_expand_epilogue): Modify validity checks of frame and stack
28647 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
28648 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
28649 * config/i386/predicates.md
28650 (save_multiple): New predicate.
28651 (restore_multiple): Likewise.
28652 * config/i386/sse.md
28653 (save_multiple<mode>): New pattern.
28654 (save_multiple_realign<mode>): Likewise.
28655 (restore_multiple<mode>): Likewise.
28656 (restore_multiple_and_return<mode>): Likewise.
28657 (restore_multiple_leave_return<mode>): Likewise.
28658 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
28660 2017-05-14 Julia Koval <julia.koval@intel.com>
28662 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
28663 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
28664 (__builtin_ia32_xsetbv): New builtins.
28665 * config/i386/i386.c (ix86_expand_special_args_builtin):
28667 (ix86_expand_builtin): Special expand for new intrinsics.
28668 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
28669 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
28670 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
28672 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28674 * cfganal.c (inverted_post_order_compute): Change argument type
28676 * cfganal.h (inverted_post_order_compute): Adjust prototype.
28677 * df-core.c (rest_of_handle_df_initialize): Adjust.
28678 (rest_of_handle_df_finish): Likewise.
28679 (df_analyze_1): Likewise.
28680 (df_analyze): Likewise.
28681 (loop_inverted_post_order_compute): Change argument to be a vec *.
28682 (df_analyze_loop): Adjust.
28683 (df_get_n_blocks): Likewise.
28684 (df_get_postorder): Likewise.
28685 * df.h (struct df_d): Change field to be a vec.
28686 * lcm.c (compute_laterin): Adjust.
28687 (compute_available): Likewise.
28688 * lra-lives.c (lra_create_live_ranges_1): Likewise.
28689 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
28690 * tree-ssa-pre.c (compute_antic): Likewise.
28692 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28694 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
28695 (depth_first_search::depth_first_search): Change structure init
28696 function to this constructor.
28697 (depth_first_search::add_bb): Rename function to this member.
28698 (depth_first_search::execute): Likewise.
28699 (flow_dfs_compute_reverse_finish): Adjust.
28701 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28703 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
28704 (longest_simple_path): Likewise.
28705 * shrink-wrap.c (spread_components): Likewise.
28706 (disqualify_problematic_components): Likewise.
28707 (emit_common_heads_for_components): Likewise.
28708 (emit_common_tails_for_components): Likewise.
28709 (insert_prologue_epilogue_for_components): Likewise.
28711 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28713 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
28716 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28718 * df-core.c (df_set_blocks): Start using auto_bitmap.
28719 (df_compact_blocks): Likewise.
28720 * df-problems.c (df_rd_confluence_n): Likewise.
28721 * df-scan.c (df_insn_rescan_all): Likewise.
28722 (df_process_deferred_rescans): Likewise.
28723 (df_update_entry_block_defs): Likewise.
28724 (df_update_exit_block_uses): Likewise.
28725 (df_entry_block_bitmap_verify): Likewise.
28726 (df_exit_block_bitmap_verify): Likewise.
28727 (df_scan_verify): Likewise.
28728 * lra-constraints.c (lra_constraints): Likewise.
28729 (undo_optional_reloads): Likewise.
28730 (lra_undo_inheritance): Likewise.
28731 * lra-remat.c (calculate_gen_cands): Likewise.
28732 (do_remat): Likewise.
28733 * lra-spills.c (assign_spill_hard_regs): Likewise.
28734 (spill_pseudos): Likewise.
28735 * tree-ssa-pre.c (bitmap_set_and): Likewise.
28736 (bitmap_set_subtract_values): Likewise.
28738 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28740 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
28741 management with auto_bitmap.
28742 (fix_inter_tick): Likewise.
28743 (fix_recovery_deps): Likewise.
28744 * ira.c (add_store_equivs): Likewise.
28745 (find_moveable_pseudos): Likewise.
28746 (split_live_ranges_for_shrink_wrap): Likewise.
28747 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
28748 (rtx_reuse_manager::seen_def_p): Likewise.
28749 (rtx_reuse_manager::set_seen_def): Likewise.
28750 * print-rtl.h (class rtx_reuse_manager): Likewise.
28752 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28754 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
28756 (migrate_btr_def): Likewise.
28757 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
28758 * df-core.c (loop_post_order_compute): Likewise.
28759 (loop_inverted_post_order_compute): Likewise.
28760 * hsa-common.h: Likewise.
28761 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
28762 * init-regs.c (initialize_uninitialized_regs): Likewise.
28763 * ipa-inline.c (resolve_noninline_speculation): Likewise.
28764 (inline_small_functions): Likewise.
28765 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
28766 * ira.c (combine_and_move_insns): Likewise.
28767 (build_insn_chain): Likewise.
28768 * loop-invariant.c (find_invariants): Likewise.
28769 * lower-subreg.c (propagate_pseudo_copies): Likewise.
28770 * predict.c (tree_predict_by_opcode): Likewise.
28771 (predict_paths_leading_to): Likewise.
28772 (predict_paths_leading_to_edge): Likewise.
28773 (estimate_loops_at_level): Likewise.
28774 (estimate_loops): Likewise.
28775 * shrink-wrap.c (try_shrink_wrapping): Likewise.
28776 (spread_components): Likewise.
28777 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
28778 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
28779 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
28780 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
28781 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
28782 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
28783 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
28784 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
28785 (mark_threaded_blocks): Likewise.
28786 (thread_through_all_blocks): Likewise.
28787 * tree-ssa.c (verify_ssa): Likewise.
28788 (execute_update_addresses_taken): Likewise.
28789 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
28791 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28793 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
28795 (post_order_compute): Likewise.
28796 (inverted_post_order_compute): Likewise.
28797 (pre_and_rev_post_order_compute_fn): Likewise.
28799 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28801 * genrecog.c (int_set::int_set): Explicitly construct our
28802 auto_vec base class.
28803 * vec.h (auto_vec::auto_vec): New constructor.
28805 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28807 * bitmap.h (class auto_bitmap): New constructor taking
28808 bitmap_obstack * argument.
28810 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28812 * bitmap.h (class auto_bitmap): Change type of m_bits to
28813 bitmap_head, and adjust ctor / dtor and member operators.
28815 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
28817 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
28818 when returned register mode doesn't match original mode.
28820 2017-05-12 Jeff Law <law@redhat.com>
28821 Jakub Jelinek <jakub@redhat.com>
28823 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
28824 we look for cc setter after the compare-elim changes.
28825 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
28826 within the vector to match what compare-elim now expects.
28827 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
28828 (xorsi3_flags, one_cmplsi2_flags): Likewise.
28830 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
28831 after the compare-elim changes.
28832 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
28833 the vector to match what compare-elim now expects.
28834 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
28835 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
28836 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
28837 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
28838 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
28840 * config/visium/visium.c (single_set_and_flags): Fix where
28841 we look for cc setter after the compare-elim changes.
28842 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
28843 with the vector to match what compare-elim now expects.
28844 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
28845 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
28846 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
28847 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
28848 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
28849 (neg<mode>2_insn_set_overflow): Likewise.
28851 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
28853 PR middle-end/79794
28854 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
28855 maybe_expand_insn call, set ops[0].target. If still set after call,
28856 set alt_rtl. Add extra arg to recursive calls.
28857 (extract_bit_field): Add alt_rtl argument. Pass to
28859 * expmed.h (extract_bit_field): Fix prototype.
28860 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
28861 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
28862 to extract_bit_field_calls.
28863 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
28864 Pass alt_rtl to extract_bit_field calls.
28865 * calls.c (store_unaligned_arguments_into_psuedos)
28866 load_register_parameters): Pass extra NULL to extract_bit_field calls.
28867 * optabs.c (maybe_legitimize_operand): Clear op->target when call
28869 * optabs.h (struct expand_operand): Add target bitfield.
28871 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28873 * compare-elim.c (try_eliminate_compare): Canonicalize
28874 operation with embedded compare to
28875 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
28876 (set (reg) (operation)].
28878 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
28880 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28883 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
28884 cost of adding a carry flag for ADC instruction.
28885 [case MINUS]: Ignore the cost of subtracting a carry flag
28886 for SBB instruction.
28888 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
28890 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
28892 * config/rs6000/bmiintrin.h: New file.
28893 * config/rs6000/bmi2intrin.h: New file.
28894 * config/rs6000/x86intrin.h: New file.
28896 2017-05-12 Jeff Law <law@redhat.com>
28898 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
28901 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
28903 PR middle-end/80707
28904 * tree-cfg.c: Remove cfg edges of unreachable case statements.
28906 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28908 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28909 early expansion of vector divide builtins.
28910 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
28911 builtins identified as having unsigned arguments.
28913 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28915 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
28916 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
28917 expansion of vector logical operations (and, andc, or, xor,
28920 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28922 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
28923 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
28925 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28927 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28928 early GIMPLE expansion of vector multiplies.
28930 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28932 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
28933 TARGET_HAVE_MOVT conditional.
28934 (movt splitter): Likewise.
28936 2017-05-12 Richard Biener <rguenther@suse.de>
28938 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
28939 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28940 Fold all stmts not inplace.
28942 2017-05-12 Richard Biener <rguenther@suse.de>
28944 PR tree-optimization/80713
28945 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
28946 inserted_exprs bit for not removed stmts.
28948 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
28950 PR middle-end/69921
28951 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
28952 parallelized" attribute for parallelized OpenACC kernels.
28953 * omp-offload.c (execute_oacc_device_lower): Use it.
28955 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
28956 Set "oacc kernels" attribute.
28957 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
28958 parameter. Adjust all users.
28959 (oacc_fn_attrib_kernels_p): Remove function.
28960 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
28961 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
28962 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
28963 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
28964 assert "oacc kernels" attribute is set.
28966 2017-05-11 Carl Love <cel@us.ibm.com>
28968 * config/rs6000/rs6000-c: Add support for built-in functions
28969 vector unsigned char vec_popcnt (vector signed char)
28970 vector unsigned char vec_popcnt (vector unsigned char)
28971 vector unsigned short vec_popcnt (vector signed short)
28972 vector unsigned short vec_popcnt (vector unsigned short)
28973 vector unsigned int vec_popcnt (vector signed int)
28974 vector unsigned int vec_popcnt (vector unsigned int)
28975 vector unsigned long long vec_popcnt (vector signed long long)
28976 vector unsigned long long vec_popcnt (vector unsigned long long)
28977 vector signed long long vec_slo (vector signed long long,
28978 vector signed char)
28979 vector signed long long vec_slo (vector signed long long,
28980 vector unsigned char)
28981 vector unsigned long long vec_slo (vector unsigned long long,
28982 vector signed char)
28983 vector unsigned long long vec_slo (vector unsigned long long,
28984 vector unsigned char)
28985 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
28986 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
28987 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
28988 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
28989 * doc/extend.texi: Update the built-in documentation file for the
28990 new built-in functions.
28992 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
28994 * attribs.h (sorted_attr_string): Move machine independent
28995 functions for target clone support from the i386 port to common
28996 code. Rename ix86_function_versions to common_function_versions.
28997 Rename make_name to make_unique_name.
28998 (common_function_versions): Likewise.
28999 (make_unique_name): Likewise.
29000 (make_dispatcher_decl): Likewise.
29001 (is_function_default_version): Likewise.
29002 * attribs.c (attr_strcmp): Likewise.
29003 (sorted_attr_string): Likewise.
29004 (common_function_versions): Likewise.
29005 (make_unique_name): Likewise.
29006 (make_dispatcher_decl): Likewise.
29007 (is_function_default_version): Likewise.
29008 * config/i386/i386.c (attr_strcmp): Likewise.
29009 (sorted_attr_string): Likewise.
29010 (ix86_function_versions): Likewise.
29011 (make_name): Likewise.
29012 (make_dispatcher_decl): Likewise.
29013 (is_function_default_version): Likewise.
29014 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
29016 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29019 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
29020 Account for direct move costs for vec_construct of integer
29023 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
29026 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
29027 (UNSPEC_STX_ATOMIC): Ditto.
29028 (loaddi_via_sse): New insn.
29029 (storedi_via_sse): Ditto.
29030 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
29031 Update corresponding peephole2 patterns.
29032 (atomic_storedi_fpu): Ditto.
29034 2017-05-11 Julia Koval <julia.koval@intel.com>
29036 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
29037 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
29039 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
29040 (__builtin_ia32_rsqrt14ss_mask): New builtins.
29041 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
29043 2017-05-11 Nathan Sidwell <nathan@acm.org>
29045 * graphite-poly.c: Include dumpfile.h.
29047 * dumpfle.h (dump_function): Declare here ...
29048 * tree-dump.h (dump_function): ... not here.
29049 * dumpfile.c: #include tree-cfg.h.
29050 (dump_function): Move here from ...
29051 * tree-dump.c (dump_function): ... here.
29052 * gimplify.c: #include splay-tree.h, not tree-dump.h.
29053 * graphite-poly.c: Don't include tree-dump.h.
29054 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
29055 * print-tree.c: Likewise.
29056 * stor-layout.c: Likewise.
29057 * tree-nested.c: Likewise.
29059 * dumpfile.c (dump_start): Use TDF_FLAGS.
29060 (dump_enable_all): Fix TDF_KIND check thinko.
29062 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
29064 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
29065 array entries to represent two legal parameterizations of the
29066 overloaded __builtin_cmpb function, as represented by the
29067 P6_OV_BUILTIN_CMPB constant.
29068 (altivec_resolve_overloaded_builtin): Add special case handling
29069 for the __builtin_cmpb function, as represented by the
29070 P6_OV_BUILTIN_CMPB constant.
29071 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
29072 (BU_P6_64BIT_2): New macro.
29073 (BU_P6_OVERLOAD_2): New macro
29074 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
29075 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
29076 (CMPB): Add overload support to represent both 32-bit and 64-bit
29077 compare-bytes function.
29078 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
29079 support for TARGET_CMPB.
29080 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
29081 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
29082 documentation of the __builtin_cmpb overloaded built-in function.
29084 2017-05-11 Richard Biener <rguenther@suse.de>
29086 PR tree-optimization/80705
29087 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
29088 bases are not vectorizable.
29090 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29092 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
29093 when counting register pressure.
29095 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29097 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
29098 (iv_ca_more_deps): Renamed to ...
29099 (iv_ca_compare_deps): ... this.
29100 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
29102 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29104 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
29106 (determine_group_iv_costs): ... here.
29107 (find_inv_vars_cb): Record inv var if it's not recorded before.
29109 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29111 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
29112 (get_shiftadd_cost): Ditto.
29114 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29116 * tree-ssa-address.c: Include header file.
29117 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
29119 (add_to_parts): Refactor.
29120 (addr_to_parts): New parameter. Update use of move_hint_to_base.
29121 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
29124 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29126 PR tree-optimization/53090
29127 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
29129 (extract_cond_operands): Detect condition with IV on both sides
29130 and return COMP_IV_EXPR_2.
29131 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
29132 (rewrite_use_compare): Simplify by removing call to function
29133 extract_cond_operands.
29135 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29137 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
29138 (extract_cond_operands): Detect condition comparing against non-
29139 invariant bound and return appropriate enum value.
29140 (find_interesting_uses_cond): Update use of extract_cond_operands.
29141 Handle its return value accordingly.
29142 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
29144 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29146 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
29147 nonlinear iv_use computation in loop invariant sensitive way.
29149 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29151 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
29152 (find_iv_candidates): Call relate_compare_use_with_all_cands.
29154 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29156 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
29157 (dump_cand): Support iv_cand.inv_exprs.
29158 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
29160 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
29163 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29165 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
29167 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
29168 as local function. Include necessary header files.
29169 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
29171 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29173 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
29175 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29177 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
29178 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
29179 RSHIFT_EXPR and BIT_NOT_EXPR.
29181 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29183 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
29184 (adjust_setup_cost): New parameter supporting round up adjustment.
29185 (struct address_cost_data): Delete.
29186 (force_expr_to_var_cost): Don't bound cost with spill_cost.
29187 (split_address_cost, ptr_difference_cost): Delete.
29188 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
29189 (struct ainc_cost_data): New struct.
29190 (get_address_cost_ainc): New function.
29191 (get_address_cost, get_computation_cost): Reimplement.
29192 (determine_group_iv_cost_address): Record inv_expr for all uses of
29194 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
29195 (iv_ca_has_deps): Reimplemented to ...
29196 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
29198 (iv_ca_extend): Call iv_ca_more_deps.
29200 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29202 * tree-ssa-address.c (struct mem_address): Move to header file.
29203 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
29204 * tree-ssa-address.h (struct mem_address): Move from C file.
29205 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
29207 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29209 * tree-affine.h (aff_combination_type): New interface.
29210 (aff_combination_zero_p): Remove static.
29211 (aff_combination_const_p): New interface.
29212 (aff_combination_singleton_var_p): New interfaces.
29214 2017-05-11 Richard Biener <rguenther@suse.de>
29216 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
29217 Skip unreachable blocks and destinations.
29218 (eliminate): Move stmt removal and fixup ...
29219 (fini_eliminate): ... here. Skip inserted exprs.
29220 (pass_pre::execute): Move fini_pre after fini_eliminate.
29221 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
29222 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
29223 PRE to get rid of dead code that has invalid SSA form and
29224 split critical edges again.
29226 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29228 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29230 2017-05-11 Richard Biener <rguenther@suse.de>
29232 * passes.c (execute_function_todo): Verify loops if they are
29233 said to be up-to-date.
29234 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
29235 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
29237 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
29240 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
29241 handle calling assemble_external ourself.
29244 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
29245 modes with zero size. Enhance comment.
29247 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29249 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
29250 built-ins for vec_xl and vec_xst with short and char pointer
29253 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
29255 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
29256 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
29257 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
29258 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
29259 (_mm_maskz_min_round_ss): New intrinsics.
29260 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
29261 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
29262 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
29263 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
29264 (__builtin_ia32_minss_mask_round): New builtins.
29265 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29266 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29267 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
29269 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
29270 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
29272 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
29275 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
29277 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
29278 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
29279 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
29280 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
29281 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
29282 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
29283 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
29284 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29285 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
29286 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
29287 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
29288 (__builtin_ia32_mulss_mask_round): New builtins.
29289 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29290 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29291 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
29293 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
29294 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
29296 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
29299 2017-05-10 Julia Koval <julia.koval@intel.com>
29301 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
29302 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
29303 (_mm256_setr_m128i): New intrinsics.
29305 2017-05-10 Julia Koval <julia.koval@intel.com>
29307 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
29308 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
29309 (_mm_maskz_rcp14_ss): New intrinsics.
29310 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
29311 (__builtin_ia32_rcp14ss_mask): New builtins.
29312 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
29314 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
29316 PR tree-optimization/51513
29317 * tree-cfg.c (gimple_seq_unreachable_p): New function.
29318 (assert_unreachable_fallthru_edge_p): Use it.
29319 (group_case_labels_stmt): Likewise.
29320 * tree-cfg.h: Prototype it.
29321 * stmt.c: Include cfghooks.h and tree-cfg.h.
29322 (emit_case_dispatch_table) <gap_label>: New local variable.
29323 Use it to fill dispatch table gaps.
29324 Test for default_label before updating probabilities.
29325 (expand_case) <default_label>: Remove unneeded initialization.
29326 Test for unreachable default case statement and remove its edge.
29327 Set default_label accordingly.
29328 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
29330 2017-05-10 Carl Love <cel@us.ibm.com>
29332 * config/rs6000/rs6000-c: Add support for built-in functions
29333 vector signed char vec_neg (vector signed char)
29334 vector signed short int vec_neg (vector short int)
29335 vector signed int vec_neg (vector signed int)
29336 vector signed long long vec_neg (vector signed long long)
29337 vector float vec_neg (vector float)
29338 vector double vec_neg (vector double)
29339 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
29341 * config/rs6000/altivec.h: Add define for vec_neg
29342 * doc/extend.texi: Update the built-in documentation for the
29343 new built-in functions.
29345 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29347 PR tree-optimization/77644
29348 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
29350 2017-05-10 Nathan Sidwell <nathan@acm.org>
29352 * dumpfile.h (TDI_lang_all): New.
29353 (TDF_KIND): New. Renumber others
29354 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
29356 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
29358 (get_dump_file_name): Adjust suffix generation.
29359 (dump_enable_all): Use TDF_KIND.
29360 * doc/invoke.texi (-fdump-lang-all): Document.
29362 * dumpfile.h: Tabify.
29364 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
29367 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
29368 Move member access before delete.
29370 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
29372 * tree-inline.c (expand_call_inline): Split block at stmt
29375 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
29378 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
29379 are now unused after splitting mov{sf,sd}_hardfloat.
29380 (f32_lr2): Likewise.
29381 (f32_lm): Likewise.
29382 (f32_lm2): Likewise.
29383 (f32_li): Likewise.
29384 (f32_li2): Likewise.
29385 (f32_lv): Likewise.
29386 (f32_sr): Likewise.
29387 (f32_sr2): Likewise.
29388 (f32_sm): Likewise.
29389 (f32_sm2): Likewise.
29390 (f32_si): Likewise.
29391 (f32_si2): Likewise.
29392 (f32_sv): Likewise.
29393 (f32_dm): Likewise.
29394 (f32_vsx): Likewise.
29395 (f32_av): Likewise.
29396 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
29397 For movsf, order stores so the VSX stores occur before the GPR
29398 store which encourages the register allocator to use a traditional
29399 FPR instead of a GPR. For movsd, order the stores so that the GPR
29400 store comes before the VSX stores to allow the power6 to work.
29401 This is due to the power6 not having a 32-bit integer store
29402 instruction from a FPR.
29403 (movsf_hardfloat): Likewise.
29404 (movsd_hardfloat): Likewise.
29406 2017-05-09 Martin Sebor <msebor@redhat.com>
29408 PR translation/80280
29409 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
29412 PR translation/80280
29413 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
29414 data member added in r247778.
29415 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
29417 2017-05-09 Nathan Sidwell <nathan@acm.org>
29419 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
29421 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
29424 2017-05-09 Marek Polacek <polacek@redhat.com>
29426 * doc/invoke.texi: Fix typo.
29428 2017-05-09 Richard Biener <rguenther@suse.de>
29430 * tree-vrp.c (vrp_val_is_max): Adjust comment.
29431 (vrp_val_is_min): Likewise.
29432 (set_value_range_to_value): Likewise.
29433 (set_value_range_to_nonnegative): Likewise.
29434 (gimple_assign_nonzero_p): Likewise.
29435 (gimple_stmt_nonzero_p): Likewise.
29436 (vrp_int_const_binop): Likewise. Remove unreachable case.
29437 (adjust_range_with_scev): Adjust comments.
29438 (compare_range_with_value): Likewise.
29439 (extract_range_from_phi_node): Likewise.
29440 (test_for_singularity): Likewise.
29442 2017-05-09 Richard Biener <rguenther@suse.de>
29444 * tree-vrp.c (get_single_symbol): Add assert that we don't
29445 get overflowed constants as invariant part.
29446 (compare_values_warnv): Add comment before the TREE_NO_WARNING
29447 checks. Use wi::cmp instead of recursing for integer constants.
29448 (compare_values): Just ignore whether we assumed undefined
29449 overflow instead of failing the compare.
29450 (extract_range_for_var_from_comparison_expr): Add comment before the
29451 TREE_NO_WARNING sets.
29452 (test_for_singularity): Likewise.
29453 (extract_range_from_comparison): Do not disable optimization
29454 when we assumed undefined overflow.
29455 (extract_range_basic): Remove init of unused var.
29457 2017-05-09 Richard Biener <rguenther@suse.de>
29459 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
29460 (extract_range_from_multiplicative_op_1): Adjust.
29461 (extract_range_from_binary_expr_1): Use int_const_binop.
29463 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
29466 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
29467 rs6000_store_data_bypass_p in seven define_bypass directives and
29468 in several comments.
29469 * config/rs6000/rs6000-protos.h: Add prototype for
29470 rs6000_store_data_bypass_p function.
29471 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
29472 function implements slightly different (rs6000-specific) semantics
29473 than store_data_bypass_p, returning false rather than aborting
29474 with assertion error when arguments do not satisfy the
29475 requirements of store data bypass.
29476 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
29477 rs6000_store_data_bypass_p.
29479 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
29481 * config/xtensa/xtensa-protos.h
29482 (xtensa_initial_elimination_offset): New declaration.
29483 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29484 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
29485 macro definition, add case for FRAME_POINTER_REGNUM when
29486 FRAME_GROWS_DOWNWARD.
29487 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
29488 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
29489 xtensa_initial_elimination_offset.
29491 2017-05-08 Nathan Sidwell <nathan@acm.org>
29493 * doc/invoke.texi: Alphabetize -fdump options.
29495 2017-05-08 Martin Sebor <msebor@redhat.com>
29497 PR translation/80280
29498 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
29500 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
29502 * target.def (compute_frame_layout): New optional target hook.
29503 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
29504 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
29505 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
29507 * reload1.c (verify_initial_elim_offsets): Likewise.
29508 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
29509 (use_simple_return_p): Call arm_compute_frame_layout if needed.
29510 (arm_get_frame_offsets): Split up into this ...
29511 (arm_compute_frame_layout): ... and this function.
29513 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
29515 * config/aarch64/constraints.md (Usa): New constraint.
29516 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
29518 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
29520 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
29521 with_multilib_list after it has been checked.
29523 2017-05-08 Richard Biener <rguenther@suse.de>
29525 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
29526 (bitmap_set_subtract_values): Likewise.
29528 2017-05-08 Richard Biener <rguenther@suse.de>
29530 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
29531 (gimple_assign_nonzero): ... this and remove strict_overflow_p
29533 (gimple_stmt_nonzero_warnv_p): Rename to ...
29534 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
29536 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
29537 (extract_range_basic): Adjust, do not disable propagation on
29538 strict overflow sensitive simplification.
29539 (vrp_visit_cond_stmt): Likewise.
29541 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
29543 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
29544 body size unconditionally.
29546 2017-05-07 Jeff Law <law@redhat.com>
29549 2017-05-06 Jeff Law <law@redhat.com>
29550 PR tree-optimization/78496
29551 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
29554 PR tree-optimization/78496
29555 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
29556 (simplify_stmt_using_ranges): Call it.
29557 (vrp_dom_walker::before_dom_children): Extract equivalences
29558 from an ASSERT_EXPR with an equality comparison against a
29561 2017-05-06 Jeff Law <law@redhat.com>
29563 PR tree-optimization/78496
29564 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
29567 PR tree-optimization/78496
29568 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
29569 (simplify_stmt_using_ranges): Call it.
29570 (vrp_dom_walker::before_dom_children): Extract equivalences
29571 from an ASSERT_EXPR with an equality comparison against a
29574 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
29576 * lra-constraints.c (lra_copy_reg_equiv): New function.
29577 (split_reg): Use it to copy equivalence information from the
29578 original register to the spill register.
29580 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
29582 PR rtl-optimization/75964
29583 * simplify-rtx.c (simplify_const_relational_operation): Remove
29584 invalid handling of comparisons of integer ABS.
29586 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
29588 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
29589 initialize to zero.
29590 (init_regs): Remove declaration.
29591 (function_arg_advance_32): Initialize error_p as boolean variable.
29593 2017-05-05 Nathan Sidwell <nathan@acm.org>
29595 * store-motion.c (remove_reachable_equiv_notes): Reformat long
29596 lines. Use for (;;).
29598 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29600 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
29601 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
29602 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
29603 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
29604 VF=2 that require versioning.
29606 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29608 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
29611 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29613 * diagnostic.h (diagnostic_override_option_index): Convert from
29614 macro to inline function.
29616 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29618 * diagnostic.c (last_module_changed_p): New function.
29619 (set_last_module): New function.
29620 (diagnostic_report_current_module): Convert macro usage to
29621 the above functions.
29622 * diagnostic.h (diagnostic_context::last_module): Strengthen
29623 from const line_map * to const line_map_ordinary *.
29624 (diagnostic_last_module_changed): Delete macro.
29625 (diagnostic_set_last_module): Delete macro.
29627 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29629 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
29630 with diagnostic_report_diagnostic.
29631 (diagnostic_n_impl_richloc): Likewise.
29632 * diagnostic.h (report_diagnostic): Delete macro.
29633 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
29634 with diagnostic_report_diagnostic.
29635 * substring-locations.c (format_warning_va): Likewise.
29637 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29639 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
29640 save/restor of format_spec. Move option-printing code to...
29641 (print_option_information): ...this new function, and
29642 reimplement by simply printing to the pretty_printer,
29643 rather than appending to the format string.
29645 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29647 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
29648 handling logic into...
29649 (update_effective_level_from_pragmas): ...this new function.
29651 2017-05-04 Andrew Waterman <andrew@sifive.com>
29653 * config/riscv/riscv.opt (mstrict-align): New option.
29654 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
29655 (SLOW_UNALIGNED_ACCESS): Define.
29656 (riscv_slow_unaligned_access): Declare.
29657 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
29659 (riscv_slow_unaligned_access): New variable.
29660 (rocket_tune_info): Set slow_unaligned_access to true.
29661 (optimize_size_tune_info): Set slow_unaligned_access to false.
29662 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
29663 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
29664 (riscv_option_override): Set riscv_slow_unaligned_access.
29665 * doc/invoke.texi: Add -mstrict-align to RISC-V.
29667 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
29669 * config/riscv/riscv.md: Unify indentation.
29671 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29676 * config/rs6000/rs6000.md (u code attribute): Add FIX and
29678 (extendsi<mode>2): Add support for doing sign extension via
29679 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
29680 don't have ISA 3.0 instructions.
29681 (extendsi<mode>2 splitter): Likewise.
29682 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
29683 generate the normal insns since SImode can now go in vector
29684 registers. Disallow the special UNSPECs needed for previous
29685 machines to hide SImode being used. Add new insns
29686 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
29687 (fix_trunc<mode>si2_stfiwx): Likewise.
29688 (fix_trunc<mode>si2_internal): Likewise.
29689 (fixuns_trunc<mode>si2): Likewise.
29690 (fixuns_trunc<mode>si2_stfiwx): Likewise.
29691 (fctiw<u>z_<mode>_smallint): Likewise.
29692 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
29693 of floating point to 32-bit integer from doing a direct move to
29694 the GPR registers to do a store.
29695 (fctiwz_<mode>): Break long line.
29697 2017-05-05 Bin Cheng <bin.cheng@arm.com>
29699 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
29700 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
29701 (addr_list, addr_offset_valid_p): New.
29702 (split_address_groups): Check offset validity with above function.
29703 (gt-tree-ssa-loop-ivopts.h): Include header file.
29705 2017-05-05 Nathan Sidwell <nathan@acm.org>
29707 * config.gcc (arm*-*-*): Add missing 'fi'.
29709 2017-05-05 Steve Ellcey <sellcey@cavium.com>
29711 * doc/invoke.texi (-fopt-info): Explicitly say order of options
29712 included in -fopt-info does not matter.
29713 * doc/optinfo.texi (-fopt-info): Fix description of default
29714 behavour. Explicitly say order of options included in -fopt-info
29717 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29719 * config.gcc: Allow combinations of aprofile and rmprofile values for
29720 --with-multilib-list.
29721 * config/arm/t-multilib: New file.
29722 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
29723 variables. Remove setting of ISA and floating-point ABI in
29724 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
29725 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
29726 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
29727 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
29729 * config/arm/t-rmprofile: Likewise except for the matches changes.
29730 * doc/install.texi (--with-multilib-list): Document the combination of
29731 aprofile and rmprofile values and warn about pitfalls in doing that.
29733 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29735 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
29736 (movdi_aarch64): Likewise.
29738 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29740 PR tree-optimization/80632
29741 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
29743 (build_arrays): Initialize it for virtual phis.
29744 (fix_phi_nodes): Use it for virtual phis.
29746 PR tree-optimization/80558
29747 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
29748 [x, y] op z into [x op, y op z] for op & or | if conditions
29751 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
29752 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29755 * config/arm/arm.md (use_literal_pool): Remove.
29756 (64-bit immediate split): No longer takes cost into consideration
29757 if arm_disable_literal_pool is enabled.
29758 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
29759 used when arm_disable_literal_pool is enabled.
29760 (arm_max_const_double_inline_cost): Remove use of
29761 arm_disable_literal_pool.
29762 (push_minipool_fix): Add assert.
29763 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
29764 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
29765 (no_literal_pool_sf_immediate): New.
29767 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29769 PR tree-optimization/80613
29770 * tree-ssa-dce.c (propagate_necessity): Remove cases for
29771 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
29773 2017-05-05 Richard Biener <rguenther@suse.de>
29775 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
29777 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
29779 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
29780 of this flag from insn conditions due to removal from r247495.
29782 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29784 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
29786 (arm_early_store_addr_dep_ptr): Likewise.
29787 * config/arm/aarch-common-protos.h
29788 (arm_early_load_addr_dep_ptr): Add prototype.
29789 (arm_early_store_addr_dep_ptr): Likewise.
29790 * config/arm/cortex-a53.md: Add new bypasses.
29792 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29794 * tree.c (next_type_uid): Change type to unsigned.
29795 (type_hash_canon): Decrement back next_type_uid if
29796 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
29797 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
29800 2017-05-04 Martin Sebor <msebor@redhat.com>
29802 * builtins.c: Fix a trivial typo in a comment.
29804 PR middle-end/79234
29805 * builtins.c (check_sizes): Adjust to handle reading past the end.
29806 Avoid printing excessive upper bound of ranges. Use %E to print
29807 tree nodes instead of converting them to %wu.
29808 (expand_builtin_memchr): New function.
29809 (compute_dest_size): Rename...
29810 (compute_objsize): ...to this.
29811 (expand_builtin_memcpy): Adjust.
29812 (expand_builtin_mempcpy): Adjust.
29813 (expand_builtin_strcat): Adjust.
29814 (expand_builtin_strcpy): Adjust.
29815 (check_strncat_sizes): Adjust.
29816 (expand_builtin_strncat): Adjust.
29817 (expand_builtin_strncpy): Adjust and simplify.
29818 (expand_builtin_memset): Adjust.
29819 (expand_builtin_bzero): Adjust.
29820 (expand_builtin_memcmp): Adjust.
29821 (expand_builtin): Handle memcmp.
29822 (maybe_emit_chk_warning): Check strncat just once.
29824 2017-05-04 Martin Sebor <msebor@redhat.com>
29826 PR preprocessor/79214
29827 PR middle-end/79222
29828 PR middle-end/79223
29829 * builtins.c (check_sizes): Add inlining context and issue
29830 warnings even when -Wno-system-headers is set.
29831 (check_strncat_sizes): Same.
29832 (expand_builtin_strncat): Same.
29833 (expand_builtin_memmove): New function.
29834 (expand_builtin_stpncpy): Same.
29835 (expand_builtin): Handle memmove and stpncpy.
29837 2017-05-04 Bin Cheng <bin.cheng@arm.com>
29839 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
29840 which is not used any more.
29842 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29844 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
29846 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29848 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
29849 (cortexa53_tunings): Likewise.
29850 (cortexa57_tunings): Likewise.
29851 (cortexa72_tunings): Likewise.
29852 (cortexa73_tunings): Likewise.
29854 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29856 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
29857 Set loop alignment to 8.
29859 2017-05-04 Martin Sebor <msebor@redhat.com>
29861 PR translation/80280
29862 * builtins.c (expand_builtin_object_size): Add missing quoting to
29863 %D and like directives.
29864 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
29865 (hsa_type_for_tree_type): Same.
29866 (verify_function_arguments): Same.
29867 * symtab.c (symbol_table::change_decl_assembler_name): Same.
29868 * varasm.c (get_section): Same.
29871 2017-05-04 Martin Sebor <msebor@redhat.com>
29873 PR translation/80280
29874 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
29876 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29878 * config/aarch64/aarch64.c (generic_addrcost_table):
29879 Change HI/TI mode setting.
29881 2017-05-04 Martin Jambor <mjambor@suse.cz>
29883 PR tree-optimization/80622
29884 * tree-sra.c (comes_initialized_p): New function.
29885 (build_accesses_from_assign): Only set write lazily when
29886 comes_initialized_p is false.
29887 (analyze_access_subtree): Use comes_initialized_p.
29888 (propagate_subaccesses_across_link): Assert !comes_initialized_p
29889 instead of testing for PARM_DECL.
29891 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29893 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
29894 constraint on operand 0 to allow more general addressing modes.
29895 Adjust output template.
29896 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
29898 * config/aarch64/aarch64-protos.h
29899 (aarch64_address_valid_for_prefetch_p): Declare prototype.
29900 * config/aarch64/constraints.md (Dp): New address constraint.
29901 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
29904 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
29906 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
29907 update use of estimate_ipcp_clone_size_and_time.
29908 (estimate_local_effects): Update use of
29909 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
29910 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
29911 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
29912 Return nonspecialized time.
29914 2017-05-04 Richard Biener <rguenther@suse.de>
29916 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
29917 for the last VUSE which def dominates the PHI. Directly call
29919 (get_continuation_for_phi_1): Remove.
29921 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
29923 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
29924 to explain the use of truncating division. Cap the number of
29925 iterations to the maximum given by nb_iterations_upper_bound,
29928 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
29930 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
29931 * configure: Regenerate.
29932 * config.in: Regenerate.
29933 * config/i386/driver-mingw32.c: new file.
29934 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
29935 * config.host: Link driver-mingw32.o on MinGW host.
29936 * doc/install.texi: Document new --enable-mingw-wildcard configure
29939 2017-05-04 Marek Polacek <polacek@redhat.com>
29941 PR tree-optimization/80612
29942 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
29944 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29945 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
29947 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
29948 (movt splitter): Likewise.
29949 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
29950 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
29951 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
29952 block for Thumb-1 with MOVT.
29953 (thumb2_legitimate_address_p): Move code block ...
29954 (can_avoid_literal_pool_for_label_p): ... into this new function.
29955 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
29957 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
29958 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
29959 "M-profile targets with the MOVT instruction".
29961 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29963 * config/arm/arm-builtins.c (arm_init_builtins): Rename
29964 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
29965 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
29967 2017-05-04 Martin Liska <mliska@suse.cz>
29969 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
29970 variable cond_code.
29972 2017-05-04 Richard Biener <rguenther@suse.de>
29974 * tree.c (array_at_struct_end_p): Handle arrays at struct
29975 end with flexarrays more conservatively. Refactor and treat
29976 arrays of arrays or aggregates more strict. Fix
29977 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
29978 * tree.c (array_at_struct_end_p): Adjust prototype.
29979 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
29980 * gimple-fold.c (get_range_strlen): Likewise.
29981 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
29983 2017-05-04 Richard Biener <rguenther@suse.de>
29985 PR tree-optimization/31130
29986 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
29988 (supports_overflow_infinity): Likewise.
29989 (is_negative_overflow_infinity): Likewise.
29990 (is_positive_overflow_infinity): Likewise.
29991 (is_overflow_infinity): Likewise.
29992 (stmt_overflow_infinity): Likewise.
29993 (overflow_infinity_range_p): Likewise.
29994 (usable_range_p): Remove as always returning true.
29995 (make_overflow_infinity): Remove.
29996 (negative_overflow_infinity): Likewise.
29997 (positive_overflow_infinity): Likewise.
29998 (avoid_overflow_infinity): Likewise.
29999 (set_value_range): Adjust accordingly.
30000 (set_value_range_to_nonnegative): Likewise, remove now unused
30001 overflow_infinity arg.
30002 (vrp_operand_equal_p): Adjust.
30003 (update_value_range): Likewise.
30004 (range_int_cst_singleton_p): Likewise.
30005 (operand_less_p): Likewise.
30006 (compare_values_warnv): Likewise.
30007 (extract_range_for_var_from_comparison_expr): Likewise.
30008 (vrp_int_const_binop): Likewise.
30009 (zero_nonzero_bits_from_vr): Likewise.
30010 (extract_range_from_multiplicative_op_1): Likewise.
30011 (extract_range_from_binary_expr_1): Likewise.
30012 (extract_range_from_unary_expr): Likewise.
30013 (extract_range_from_comparison): Likewise.
30014 (extract_range_basic): Likewise.
30015 (adjust_range_with_scev): Likewise.
30016 (compare_ranges): Likewise.
30017 (compare_range_with_value): Likewise.
30018 (dump_value_range): Likewise.
30019 (test_for_singularity): Likewise, remove strict_overflow_p parameter
30021 (simplify_cond_using_ranges): Adjust.
30023 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
30025 * brig-builtins.def: Added a builtin for class_f64.
30026 * builtin-types.def: Added a builtin type needed by class_f64.
30028 2017-05-03 Jason Merrill <jason@redhat.com>
30030 * timevar.def: Add TV_CONSTEXPR.
30032 2017-05-03 David Malcolm <dmalcolm@redhat.com>
30034 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
30036 2017-05-03 Martin Jambor <mjambor@suse.cz>
30038 * ipa-prop.c (ipa_update_after_lto_read): Removed.
30039 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
30040 * ipa-cp.c (ipcp_propagate_stage): Do not call
30041 ipa_update_after_lto_read.
30042 * ipa-inline.c (ipa_inline): Likewise.
30044 2017-05-03 Martin Jambor <mjambor@suse.cz>
30046 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
30047 tag. Added a default constructor and a destructor.
30048 (ipa_edge_args_sum_t): New class;
30049 (ipa_edge_args_sum): Declare.
30050 (ipa_edge_args_vector): Remove declaration.
30051 (IPA_EDGE_REF): Use ipa_edge_args_sum.
30052 (ipa_free_edge_args_substructures): Remove declaration.
30053 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
30054 (ipa_edge_args_info_available_for_edge_p): Likewise.
30055 * ipa-prop.c (ipa_edge_args_vector): Removed.
30056 (edge_removal_hook_holder): Likewise.
30057 (edge_duplication_hook_holder): Likewise.
30058 (ipa_edge_args_sum): New variable.
30059 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
30060 ipa_edge_args_vector.
30061 (ipa_free_edge_args_substructures): Likewise.
30062 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
30063 ipa_edge_args_vector.
30064 (ipa_edge_removal_hook): Turned into method
30065 ipa_edge_args_sum_t::remove.
30066 (ipa_edge_duplication_hook): Turned into method
30067 ipa_edge_args_sum_t::duplicate.
30068 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
30069 registering edge hooks.
30070 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
30071 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
30072 ipa_edge_args_sum instead of ipa_edge_args_vector.
30073 * ipa-profile.c (ipa_profile): Likewise.
30075 2017-05-03 Martin Jambor <mjambor@suse.cz>
30077 * symbol-summary.h (function_summary): New method exists.
30078 (function_summary::symtab_removal): Deallocate through release.
30079 (call_summary): New class.
30080 (gt_ggc_mx): New overload.
30081 (gt_pch_nx): Likewise.
30082 (gt_pch_nx): Likewise.
30084 2017-05-03 Jeff Law <law@redhat.com>
30086 PR tree-optimization/78496
30087 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
30088 from simplify_cond_using_ranges. Split off code to walk
30089 backwards through casts into ...
30090 (simplify_cond_using_ranges_2): New function.
30091 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
30092 (execute_vrp): After identifying jump threads, call
30093 simplify_cond_using_ranges_2.
30095 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
30098 * ipa-inline.h (inline_summary): Add ctor.
30099 (create_ggc): Do not use ggc_cleared_alloc.
30101 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
30102 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
30104 * gcc.c (handle_braces): Support escaping in switch matching
30106 * doc/invoke.texi (Spec Files): Document it.
30107 Remove superfluous @code markup in items.
30109 2017-05-03 David Malcolm <dmalcolm@redhat.com>
30111 * diagnostic-show-locus.c (struct column_range): New struct.
30112 (get_affected_columns): New function.
30113 (get_printed_columns): New function.
30114 (struct correction): New struct.
30115 (correction::ensure_capacity): New function.
30116 (correction::ensure_terminated): New function.
30117 (struct line_corrections): New struct.
30118 (line_corrections::~line_corrections): New dtor.
30119 (line_corrections::add_hint): New function.
30120 (layout::print_trailing_fixits): Reimplement in terms of the new
30122 (selftest::test_overlapped_fixit_printing): New function.
30123 (selftest::diagnostic_show_locus_c_tests): Call it.
30125 2017-05-03 Nathan Sidwell <nathan@acm.org>
30127 Canonicalize canonical type hashing
30128 * tree.h (type_hash_canon_hash): Declare.
30129 * tree.c (type_hash_list, attribute_hash_list): Move into
30130 type_hash_canon_hash.
30131 (build_type_attribute_qual_variant): Break out hash code calc into
30132 type_hash_canon_hash.
30133 (type_hash_canon_hash): New. Generic type hash computation.
30134 (build_range_type_1, build_array_type_1, build_function_type,
30135 build_method_type_directly, build_offset_type, build_complex_type,
30136 make_vector_type): Call it.
30138 2017-05-03 Richard Biener <rguenther@suse.de>
30140 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
30141 When all DRs have unknown misaligned do not always peel
30142 when there is a store but apply the same costing model as if
30143 there were only loads.
30145 2017-05-03 Richard Biener <rguenther@suse.de>
30148 PR tree-optimization/80492
30149 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30150 compare_base_decls returning dont-know properly.
30152 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
30154 * config/arm/iterators.md (CCSI): New mode iterator.
30155 (arch): New mode attribute.
30156 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
30157 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
30158 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
30159 code iterator for success result mode.
30160 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
30161 the corresponding new insn generators.
30163 2017-05-03 Bin Cheng <bin.cheng@arm.com>
30166 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30167 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30169 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
30171 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
30172 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
30173 (DDR_A): Wrap DDR argument in brackets.
30174 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
30175 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
30176 (DDR_REVERSED_P): Likewise.
30178 2017-05-03 Jakub Jelinek <jakub@redhat.com>
30180 PR tree-optimization/79472
30181 * tree-switch-conversion.c (struct switch_conv_info): Add
30182 contiguous_range and default_case_nonstandard fields.
30183 (collect_switch_conv_info): Compute contiguous_range and
30184 default_case_nonstandard fields, don't clear final_bb if
30185 contiguous_range and only the default case doesn't have the required
30187 (check_all_empty_except_final): Set default_case_nonstandard instead
30188 of failing if contiguous_range and the default case doesn't have empty
30190 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
30191 and only the default case doesn't have the required constants. Skip
30193 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
30194 if default_case_nonstandard.
30195 (build_constructors): Build constant 1 just once. Assert that default
30196 values aren't inserted in between cases if contiguous_range. Skip
30198 (build_arrays): Skip virtual phis.
30199 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
30200 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
30201 Handle virtual phis.
30202 (gen_inbound_check): Handle default_case_nonstandard case.
30203 (process_switch): Adjust check_final_bb caller. Call
30204 gather_default_values with the first non-default case instead of
30205 default case if default_case_nonstandard.
30207 2017-05-02 Nathan Sidwell <nathan@acm.org>
30209 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
30210 check. Fix formatting.
30212 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
30214 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
30215 errors when comparing specialized and unspecialized times.
30217 2017-05-02 David Malcolm <dmalcolm@redhat.com>
30219 * diagnostic-show-locus.c
30220 (layout::should_print_annotation_line_p): Make private.
30221 (layout::print_annotation_line): Make private.
30222 (layout::annotation_line_showed_range_p): Make private.
30223 (layout::show_ruler): Make private.
30224 (layout::print_source_line): Make private. Pass in line and
30225 line_width, rather than calling location_get_source_line. Drop
30227 (layout::print_leading_fixits): New method.
30228 (layout::print_any_fixits): Rename to...
30229 (layout::print_trailing_fixits): ...this, and make private.
30230 Don't print newline fixits.
30231 (diagnostic_show_locus): Move logic for printing one row into...
30232 (layout::print_line): ...this new function. Move the
30233 location_get_source_line call and error-handling from
30234 print_source_line to here. Call print_leading_fixits, and rename
30235 print_any_fixits to print_trailing_fixits.
30236 (selftest::test_fixit_insert_containing_newline): Update now that
30237 newlines are partially supported.
30238 (selftest::test_fixit_insert_containing_newline_2): New test.
30239 (selftest::test_fixit_replace_containing_newline): Update comments.
30240 (selftest::diagnostic_show_locus_c_tests): Call the new test.
30241 * edit-context.c (class added_line): New class.
30242 (class edited_line): Describe newline handling in comment.
30243 (edited_line::actually_edited_p): New method.
30244 (edited_line::print_content): Delete redundant decl.
30245 (edited_line::m_predecessors): New field.
30246 (edited_file::print_content): Call edited_line::print_content.
30247 (edited_file::print_diff): Update to support newlines.
30248 (edited_file::print_diff_hunk): Likewise.
30249 (edited_file::print_run_of_changed_lines): New function.
30250 (edited_file::print_diff_line): Convert to...
30251 (print_diff_line): ...this.
30252 (edited_file::get_effective_line_count): New function.
30253 (edited_line::edited_line): Initialize new field m_predecessors.
30254 (edited_line::~edited_line): Clean up m_predecessors.
30255 (edited_line::apply_fixit): Handle newlines.
30256 (edited_line::get_effective_line_count): New function.
30257 (edited_line::print_content): New function.
30258 (edited_line::print_diff_lines): New function.
30259 (selftest::test_applying_fixits_insert_containing_newline): New
30261 (selftest::test_applying_fixits_replace_containing_newline): New
30263 (selftest::insert_line): New function.
30264 (selftest::test_applying_fixits_multiple_lines): Add example of
30266 (selftest::edit_context_c_tests): Call the new tests.
30268 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30270 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
30271 parameter cand. Update dump information.
30272 (get_computation_cost): Update uses.
30274 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30276 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
30277 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
30278 (get_computation_at, rewrite_use_address): Update use of
30279 get_computation_aff.
30281 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30283 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
30284 (get_computation): Delete.
30285 (get_computation_cost): Implement like get_computation_cost_at.
30286 Use get_computation_at.
30287 (get_computation_cost_at): Delete.
30288 (rewrite_use_nonlinear_expr): Use get_computation_at.
30289 (rewrite_use_compare, remove_unused_ivs): Ditto.
30291 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30293 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
30295 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30297 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
30298 (ivopts_global_cost_for_size): Rename parameter and update uses.
30299 (iv_ca_recount_cost): Update uses.
30300 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
30301 candidates seperately in n_invs and n_cands.
30302 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
30304 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30306 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
30307 (find_inv_vars_cb): New.
30308 (find_depends): Renamed to ...
30309 (find_inv_vars): ... this.
30310 (add_candidate_1, force_var_cost): Call find_inv_vars.
30311 (split_address_cost, determine_group_iv_cost_cond): Ditto.
30313 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30315 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
30316 inv_vars. Add inv_exprs.
30317 (struct iv_cand): Rename depends_on to inv_vars.
30318 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
30319 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
30320 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
30321 (dump_cand): Dump inv_vars.
30322 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
30323 (record_invariant, find_depends, add_candidate_1): Ditto.
30324 (set_group_iv_cost, force_var_cost): Ditto.
30325 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
30326 (get_computation_cost_at, get_computation_cost): Ditto.
30327 (determine_group_iv_cost_generic): Ditto.
30328 (determine_group_iv_cost_address): Ditto.
30329 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
30330 (determine_group_iv_costs): Ditto.
30331 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
30332 (iv_ca_set_remove_invariants): Renamed to ...
30333 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
30334 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
30335 (iv_ca_set_add_invariants): Renamed to ...
30336 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
30337 (iv_ca_set_cp): Use iv_ca_set_add_invs.
30338 (iv_ca_has_deps): Support inv_vars and inv_exprs.
30339 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
30340 (create_new_ivs): Remove useless dump.
30342 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30344 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
30346 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
30347 (iv_ca_set_no_cp, create_new_iv): Ditto.
30349 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30351 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30353 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
30355 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
30356 function tree_check2.
30358 2017-05-02 Martin Liska <mliska@suse.cz>
30360 * doc/gcov.texi: Add missing preposition.
30361 * gcov.c (function_info::function_info): Properly fill up
30362 all member variables.
30364 2017-05-02 Tamar Christina <tamar.christina@arm.com>
30366 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
30368 2017-05-02 Tamar Christina <tamar.christina@arm.com>
30370 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
30372 2017-05-02 Martin Liska <mliska@suse.cz>
30375 * lto-streamer-in.c (lto_read_tree_1): Remove
30376 LTO_STREAMER_DEBUG.
30377 * lto-streamer.c (struct tree_hash_entry): Likewise.
30378 (struct tree_entry_hasher): Likewise.
30379 (tree_entry_hasher::hash): Likewise.
30380 (tree_entry_hasher::equal): Likewise.
30381 (lto_streamer_init): Likewise.
30382 (lto_orig_address_map): Likewise.
30383 (lto_orig_address_get): Likewise.
30384 (lto_orig_address_remove): Likewise.
30385 * lto-streamer.h: Likewise.
30386 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
30387 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
30389 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
30391 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
30392 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
30393 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
30394 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
30395 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
30396 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
30397 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
30398 (mm_maskz_sub_ss): New intrinsics.
30399 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30400 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
30401 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
30402 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
30403 (__builtin_ia32_subss_mask_round): New builtins.
30404 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30405 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30406 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
30408 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
30409 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
30411 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
30414 2017-05-02 Martin Jambor <mjambor@suse.cz>
30416 PR tree-optimization/78687
30417 * tree-sra.c (access): New field parent.
30418 (process_subtree_disqualification): New function.
30419 (disqualify_candidate): Call it.
30420 (build_accesses_from_assign): Reset write flag if creating an
30422 (build_access_subtree): Fill in parent field and also prpagate
30423 down grp_write flag.
30424 (create_artificial_child_access): New parameter set_grp_write, set
30425 grp_write to its value.
30426 (propagate_subaccesses_across_link): Also propagate grp_write flag
30428 (propagate_all_subaccesses): Push the closest parent back to work
30429 queue if add_access_to_work_queue returned true.
30431 2017-05-02 Richard Biener <rguenther@suse.de>
30433 * common.opt (fstrict-overflow): Alias negative to fwrapv.
30434 * doc/invoke.texi (fstrict-overflow): Remove all traces of
30435 -fstrict-overflow documentation.
30436 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
30437 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
30438 flag_strict_overflow.
30439 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
30440 * lto-opts.c (lto_write_options): Do not stream it.
30441 * lto-wrapper.c (merge_and_complain): Do not handle it.
30442 * opts.c (default_options_table): Do not set -fstrict-overflow.
30443 (finish_options): Likewise do not clear it when sanitizing.
30444 * simplify-rtx.c (simplify_const_relational_operation): Do not
30445 test flag_strict_overflow.
30447 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
30449 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
30450 using enabled attribute.
30451 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
30452 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
30453 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
30454 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
30455 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
30456 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
30457 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
30458 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
30459 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
30460 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
30462 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
30464 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
30466 2017-05-02 Richard Biener <rguenther@suse.de>
30468 PR tree-optimization/80591
30470 2017-04-10 Richard Biener <rguenther@suse.de>
30472 * tree-ssa-structalias.c (find_func_aliases): Properly handle
30475 2017-05-02 Richard Biener <rguenther@suse.de>
30477 PR tree-optimization/80549
30478 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
30479 (cleanup_tree_cfg_noloop): Create forwarders to known loop
30480 headers if they do not have a preheader.
30482 2017-05-02 Martin Liska <mliska@suse.cz>
30485 * common.opt: Fix typo.
30486 * doc/invoke.texi: Likewise.
30488 2017-05-01 Jan Beulich <jbeulich@suse.com>
30490 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
30491 swapping, add (x,x,m,x,n) alternative.
30493 2017-05-01 Nathan Sidwell <nathan@acm.org>
30495 * calls.c (combine_pending_stack_adjustment_and_call): Remove
30496 unnecessary unadjusted_alignment check.
30498 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
30501 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
30503 * gimplify.c (gimplify_cilk_detach): New function.
30504 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
30505 * tree-core.h: Document EXPR_CILK_SPAWN.
30506 * tree.h (EXPR_CILK_SPAWN): Define.
30508 2017-05-01 David Malcolm <dmalcolm@redhat.com>
30510 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
30511 to use new fixit_hint representation, using the "replace" logic.
30512 (get_line_span_for_fixit_hint): Likewise.
30513 (layout::print_any_fixits): Likewise.
30514 (selftest::test_one_liner_many_fixits): Rename to...
30515 (selftest::test_one_liner_many_fixits_1): ...this, and update
30516 comment and expected output to reflect that the multiple fix-it
30517 hints are now consolidated into one insertion.
30518 (selftest::test_one_liner_many_fixits_2): New test.
30519 (selftest::test_diagnostic_show_locus_one_liner): Update for
30521 (selftest::test_fixit_consolidation): Update for fix-it API
30523 * diagnostic.c (print_parseable_fixits): Likewise.
30524 * edit-context.c (edited_line::m_line_events): Convert from
30525 auto_vec <line_event *> to auto_vec <line_event>.
30526 (class line_event): Convert from abstract base class to a concrete
30527 class, taking over the role of replace_event.
30528 (class insert_event): Delete.
30529 (class replace_event): Rename to class line_event. Convert to
30531 (edit_context::add_fixits): Reimplement.
30532 (edit_context::apply_insert): Delete.
30533 (edit_context::apply_replace): Rename to...
30534 (edit_context::apply_fixit): ...this. Convert to half-open range.
30535 (edited_file::apply_insert): Delete.
30536 (edited_file::apply_replace): Rename to...
30537 (edited_file::apply_fixit): ...this.
30538 (edited_line::~edited_line): Drop deletion of events.
30539 (edited_line::apply_insert): Delete.
30540 (edited_line::apply_replace): Rename to...
30541 (edited_line::apply_fixit): ...this. Convert to half-open range.
30542 Update for change to type of m_line_events.
30543 * edit-context.h (edit_context::apply_insert): Delete.
30544 (edit_context::apply_replace): Rename to...
30545 (edit_context::apply_fixit): ...this.
30547 2017-05-01 Martin Sebor <msebor@redhat.com>
30549 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
30552 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
30555 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
30556 __get_cpuid_max returns 0.
30557 (__get_cpuid_count): Ditto.
30559 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
30561 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
30562 replacement expression is another instance of one of its arguments.
30564 2017-05-01 Jakub Jelinek <jakub@redhat.com>
30567 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
30568 check for stack push/pop autoinc.
30569 * config/i386/i386.c (ix86_agi_dependent): Return false
30570 if the only reason why modified_in_p returned true is that
30571 addr is SP based and set_insn is a push or pop.
30573 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
30575 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
30578 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
30581 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
30582 (account_size_time): Use two predicates - exec_pred and
30584 (evaluate_conditions_for_known_args): Compute both clause and
30586 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
30587 (inline_summary_t::duplicate): Update.
30588 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
30590 (compute_inline_parameters): Likewise.
30591 (estimate_edge_size_and_time): Update caluclation of time.
30592 (estimate_node_size_and_time): Compute both time and nonspecialized
30594 (estimate_ipcp_clone_size_and_time): Update.
30595 (inline_merge_summary): Update.
30596 (do_estimate_edge_time): Update.
30597 (do_estimate_edge_size): Update.
30598 (do_estimate_edge_hints): Update.
30599 (inline_read_section, inline_write_summary): Stream both new predicates.
30600 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
30602 (compute_inlined_call_time): Cleanup.
30603 (big_speedup_p): Update.
30604 (edge_badness): Update.
30605 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
30606 (size_time_entry): Replace predicate by exec_predicate and
30607 nonconst_predicate.
30608 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
30609 (estimate_edge_time): Return also nonspec_time.
30610 (reset_edge_growth_cache): Update.
30612 2017-04-29 Jakub Jelinek <jakub@redhat.com>
30614 PR rtl-optimization/80491
30615 * ifcvt.c (noce_process_if_block): When looking for x setter
30616 with missing else_bb, don't check only the insn right before
30617 cond_earliest, but look for the last insn that x is modified in
30618 within the same bb.
30620 PR rtl-optimization/80491
30621 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
30623 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
30625 PR tree-optimization/80487
30626 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
30628 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30630 PR tree-optimization/79697
30631 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
30632 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
30633 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
30635 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
30636 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
30638 2017-04-28 Martin Sebor <msebor@redhat.com>
30640 PR tree-optimization/80523
30641 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
30642 (init_target_to_host_charmap, target_to_host, target_strtol10): New
30644 (maybe_warn, format_directive, parse_directive): Use new functions.
30645 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
30647 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
30649 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
30651 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
30653 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
30654 target_header_dir): Set correctly.
30655 * configure: Regenerated.
30656 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30657 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30658 instead of SYSTEM_HEADER_DIR.
30660 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
30662 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
30663 (estimate_local_effects): Likewise.
30664 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
30665 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
30666 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
30667 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
30668 do_estimate_edge_time, estimate_edge_time): Likewise.
30669 * ipa-inline-analysis.c (estimate_node_size_and_time,
30670 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
30671 (estimate_time_after_inlining): Remove.
30673 2017-04-28 Martin Liska <mliska@suse.cz>
30675 * doc/gcov.texi: Enhance documentation of gcov.
30677 2017-04-28 Martin Liska <mliska@suse.cz>
30679 * doc/gcov.texi: Sort options in alphabetic order.
30680 * doc/gcov-dump.texi: Likewise.
30681 * doc/gcov-tool.texi: Likewise.
30682 * gcov.c (print_usage): Likewise.
30683 * gcov-dump.c (print_usage): Likewise.
30684 * gcov-tool.c (print_merge_usage_message): Likewise.
30685 (print_rewrite_usage_message): Likewise.
30686 (print_overlap_usage_message): Likewise.
30688 2017-04-28 Martin Liska <mliska@suse.cz>
30690 PR gcov-profile/53915
30691 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
30693 2017-04-28 Martin Liska <mliska@suse.cz>
30695 PR gcov-profile/79891
30696 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
30697 is marked by compiler as living on a line.
30698 (get_cycles_count): Remove usage of the union.
30699 (output_intermediate_file): Likewise.
30700 (find_source): Fix GNU coding style.
30701 (accumulate_line_counts): Remove old non-all block mode.
30702 (output_lines): Remove usage of the union.
30703 * profile.c (output_location): Include all BBs, even if
30704 belonging to a same line (and file) as a previous BB.
30706 2017-04-28 Martin Liska <mliska@suse.cz>
30708 * gcov.c (process_args): Handle new argument 'w'.
30709 (read_graph_file): Assign ID to BBs.
30710 (output_branch_count): Display BB # if verbose flag is set.
30711 (output_lines): Likewise for arcs.
30712 (print_usage): Add '--verbose' option help.
30713 * doc/gcov.texi: Document --verbose (-w) option.
30715 2017-04-28 Martin Liska <mliska@suse.cz>
30717 * gcov.c (struct block_location_info): New struct.
30718 (process_file): Fill up the new structure.
30719 (read_graph_file): Replace usage of encoding by the newly added
30721 (add_line_counts): Likewise.
30722 (accumulate_line_counts): Remove usage of the union.
30723 (function_info::function_info): New function.
30724 (function_info::~function_info): Likewise.
30725 (process_file): Call delete instead of release_function.
30726 (release_function): Release the function.
30727 (release_structures): Call delete instead of release_function.
30728 (solve_flow_graph): Replace usage of num_blocks.
30729 (find_exception_blocks): Likewise.
30730 (output_lines): Fix GNU coding style.
30732 2017-04-28 Martin Liska <mliska@suse.cz>
30735 * coverage.c (coverage_remove_note_file): New function.
30736 * coverage.h: Declare the function.
30737 * toplev.c (finalize): Clean if an error has been seen.
30739 2017-04-28 Martin Liska <mliska@suse.cz>
30741 PR gcov-profile/80031
30742 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
30743 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
30744 * gcov.c (read_graph_file): Read just number of blocks.
30745 * profile.c (branch_prob): Do not stream 0 flags per a basic
30748 2017-04-28 Martin Liska <mliska@suse.cz>
30750 * gcov-dump.c (tag_*): Add new argument to declarations.
30751 (dump_gcov_file): Likewise.
30752 (tag_blocks): Add and use new argument depth.
30753 (tag_arcs): Likewise.
30754 (tag_lines): Likewise.
30755 (tag_counters): Likewise.
30756 (tag_summary): Likewise.
30757 (dump_working_sets): Use depth to do a proper indentation.
30759 2017-04-28 Jakub Jelinek <jakub@redhat.com>
30762 * cgraph.h (symtab_node::debug_symtab): No longer inline.
30763 * symtab.c (symtab_node::debug_symtab): Move definition here.
30765 2017-04-28 Richard Biener <rguenther@suse.de>
30767 * lto-streamer.h (LTO_major_version): Bump to 7.
30769 2017-04-28 Richard Biener <rguenther@suse.de>
30771 * tree-vrp.c (assert_info): New struct.
30772 (add_assert_info): New helper.
30773 (register_edge_assert_for_2): Refactor to add asserts to a vector
30775 (register_edge_assert_for_1): Likewise.
30776 (register_edge_assert_for): Likewise.
30777 (finish_register_edge_assert_for): New helper actually registering
30778 asserts where live on edge.
30779 (find_conditional_asserts): Adjust.
30780 (find_switch_asserts): Likewise.
30781 (evrp_dom_walker::try_find_new_range): Generalize.
30782 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
30784 2017-04-27 Marek Polacek <polacek@redhat.com>
30787 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
30788 arg10 and arg11 to itype.
30790 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
30792 * doc/extend.texi (Object Size Checking): Improve grammar.
30794 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
30797 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
30798 that the logic for permitting reciprocal estimates matches that
30801 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30804 * tree.c (type_cache_hasher::equal): Only compare
30805 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
30806 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
30807 non-aggregate element types.
30808 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
30809 about the flag on ARRAY_TYPEs in the comment, formatting fix.
30811 2017-04-27 Richard Biener <rguenther@suse.de>
30813 PR middle-end/80533
30814 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
30815 stripping ARRAY_REFs from MEM_EXPR make sure we're not
30816 keeping a reference to a trailing array.
30818 2017-04-27 Richard Biener <rguenther@suse.de>
30820 PR middle-end/80539
30821 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
30822 being in loop-closed SSA form conservatively.
30823 (chrec_fold_multiply_poly_poly): Likewise.
30825 2017-04-27 Tamar Christina <tamar.christina@arm.com>
30827 PR middle-end/79665
30828 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
30829 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
30831 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30834 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
30835 (aarch64_function_arg_alignment): Return unsigned int again, but still
30836 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
30837 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30838 Don't emit -Wpsabi note.
30839 (aarch64_function_arg_boundary): Likewise.
30840 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30843 2017-04-26 Nathan Sidwell <nathan@acm.org>
30845 * tree.h (crc32_unsigned_n): Declare.
30846 (crc32_unsigned, crc32_unsigned): Make inline.
30847 * tree.c (crc32_unsigned_bits): Replace with ...
30848 (crc32_unsigned_n): ... this.
30849 (crc32_unsigned, crc32_byte): Remove.
30850 (crc32_string): Remove unnecessary braces.
30852 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30854 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
30855 * ipa-inline-analysis.c (MAX_TIME): Remove.
30856 (account_size_time): Use sreal for time.
30857 (dump_inline_summary): Update.
30858 (estimate_function_body_sizes): Update.
30859 (estimate_edge_size_and_time): Update.
30860 (estimate_calls_size_and_time): Update.
30861 (estimate_node_size_and_time): Update.
30862 (inline_merge_summary): Update.
30863 (inline_update_overall_summary): Update.
30864 (estimate_time_after_inlining): Update.
30865 (inline_read_section): Update.
30866 (inline_write_summary): Update.
30867 * ipa-inline.c (compute_uninlined_call_time): Update.
30868 (compute_inlined_call_time): Update.
30869 (recursive_inlining): Update.
30870 (inline_small_functions): Update.
30871 (dump_overall_stats): Update.
30872 * ipa-inline.h: Include sreal.h.
30873 (size_time_entry): Turn time to sreal.
30874 (inline_summary): Turn self_time nad time to sreal.
30876 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30878 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
30880 (sreal::stream_out, sreal::stream_in): New.
30881 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
30883 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30885 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
30888 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
30891 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
30893 (dimode_scalar_chain::compute_convert_gain): Ditto.
30894 (dimode_scalar_chain::make_vector_copies): Ditto.
30895 (dimode_scalar_chain::convert_reg): Ditto.
30896 (dimode_scalar_chain::convert_insn): Ditto.
30897 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
30898 (VI248_AVX512BW_1): New mode iterator.
30899 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
30900 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
30903 2017-04-25 Martin Sebor <msebor@redhat.com>
30905 PR tree-optimization/80497
30906 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
30907 constants are representable in HOST_WIDE_INT.
30908 (parse_directive): Ditto.
30910 2017-04-25 Martin Sebor <msebor@redhat.com>
30913 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
30914 (new_zero_array): Adjust signature.
30915 (dom_info::dom_init): Used unsigned rather that size_t.
30916 (dom_info::dom_info): Same.
30918 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30919 Jakub Jelinek <jakub@redhat.com>
30922 * config/arm/arm.c: Include gimple.h.
30923 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30924 returns negative, increment ncrn only if it returned positive.
30925 (arm_needs_doubleword_align): Return int instead of bool,
30926 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
30927 members, but if there is any such non-FIELD_DECL
30928 > PARM_BOUNDARY aligned decl, return -1 instead of false.
30929 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30930 returns negative, increment nregs only if it returned positive.
30931 (arm_setup_incoming_varargs): Likewise.
30932 (arm_function_arg_boundary): Emit -Wpsabi note if
30933 arm_needs_doubleword_align returns negative, return
30934 DOUBLEWORD_ALIGNMENT only if it returned positive.
30936 2017-04-25 Marek Polacek <polacek@redhat.com>
30939 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30940 first argument to type.
30942 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
30945 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
30946 type checks to test for compatibility instead of equality.
30948 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30949 Jakub Jelinek <jakub@redhat.com>
30952 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
30954 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
30955 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
30956 the alignment computation, but return their maximum in warn_alignment.
30957 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30958 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
30960 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
30961 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30964 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30966 * config/arc/simdext.md (dmpyh): Fix typo.
30968 2017-04-25 Richard Biener <rguenther@suse.de>
30970 PR tree-optimization/80492
30971 * alias.c (compare_base_decls): Handle registers with asm
30972 specification conservatively.
30973 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30974 compare_base_decls returning dont-know properly.
30976 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30978 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
30979 (legitimate_offset_address_p): New function.
30980 (arc_legitimate_address_p): Use above function.
30982 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30984 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
30986 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30988 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
30989 ACCH registers whenever they are available.
30991 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30993 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
30994 double regs fix when not used.
30996 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30998 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
31000 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
31001 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
31003 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31005 * config/arc/arc.c (arc_output_addsi): Check for h-register class
31006 when emitting short ADD instructions.
31008 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31010 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
31012 (cmpsi_cc_c_insn): Likewise.
31013 (cbranchsi4_scratch): Compute proper instruction length using
31014 compact_hreg_operand.
31015 * config/arc/predicates.md (compact_hreg_operand): New predicate.
31017 2017-04-25 Richard Biener <rguenther@suse.de>
31019 PR middle-end/80509
31020 * passes.c (pass_manager::pass_manager): Initialize
31021 m_name_to_pass_map.
31023 2017-04-25 Richard Biener <rguenther@suse.de>
31025 PR tree-optimization/79201
31026 * tree-ssa-sink.c (statement_sink_location): Handle calls.
31028 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31031 * config/s390/vector.md: Split MEM->GPR vector moves for
31032 non-s_operand addresses.
31034 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31037 * config/s390/predicates.md (reload_const_wide_int_operand): New
31039 * config/s390/s390.md ("movti"): Remove d/P alternative.
31040 ("movti_bigconst"): New pattern definition.
31042 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31045 * s390-protos.h (s390_expand_cs_hqi): Removed.
31046 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
31047 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
31048 modes as well as CCZ1mode and CCZmode.
31049 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
31050 signature of s390_emit_compare_and_swap.
31051 (s390_expand_cs_hqi): Likewise, make static.
31052 (s390_expand_cs_tdsi): Generate an explicit compare before trying
31053 compare-and-swap, in some cases.
31054 (s390_expand_cs): Wrapper function.
31055 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
31057 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
31058 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
31059 patterns for small and large integers. Forbid symref memory operands.
31060 Move expander to s390.c. Require cc register.
31061 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
31062 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
31063 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
31064 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
31065 symref memory operands. Remove CC mode and call s390_match_ccmode
31067 ("atomic_exchange<mode>"): Allow and implement all integer modes.
31069 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31071 * config/s390/s390.md (define_peephole2): New peephole to help
31072 combining the load-and-test pattern with volatile memory.
31074 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31076 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
31077 with CCZmode for TARGET_Z196.
31079 2017-04-25 Jakub Jelinek <jakub@redhat.com>
31081 PR rtl-optimization/80501
31082 * combine.c (make_compound_operation_int): Set subreg_code to SET
31083 even for AND with mask of the sign bit of mode.
31085 PR rtl-optimization/80500
31086 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
31087 sum's initial value.
31089 2017-04-25 Julian Brown <julian@codesourcery.com>
31090 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31092 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
31094 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
31096 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
31098 2017-04-25 Julian Brown <julian@codesourcery.com>
31099 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31101 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
31102 (thunderx2t99_sha): New Reservation.
31104 2017-04-25 Julian Brown <julian@codesourcery.com>
31105 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31107 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
31108 type for 1-element load.
31110 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
31112 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
31114 2017-04-24 Martin Jambor <mjambor@suse.cz>
31116 PR tree-optimization/80293
31117 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
31118 char arrays not totally scalarizable if it is false.
31119 (analyze_all_variable_accesses): Pass correct value in the new
31120 parameter. Add a statistics counter.
31122 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
31124 PR middle-end/79931
31125 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
31127 2017-04-24 Richard Biener <rguenther@suse.de>
31129 PR tree-optimization/80494
31130 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
31131 out for complex types.
31133 2017-04-24 Richard Biener <rguenther@suse.de>
31135 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
31136 * tree-ssa-sccvn.c (print_scc): Print SCC size.
31137 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
31138 (DFS): Adjust and never fail.
31139 (sccvn_dom_walker::fail): Remove.
31140 (sccvn_dom_walker::before_dom_children): Adjust.
31141 (run_scc_vn): Likewise and never fail.
31142 * tree-ssa-pre.c (pass_pre::execute): Adjust.
31143 (pass_fre::execute): Likewise.
31145 2017-04-24 Richard Biener <rguenther@suse.de>
31147 PR tree-optimization/79725
31148 * tree-ssa-sink.c (statement_sink_location): Return whether
31149 failure reason was zero uses. Move that check later.
31150 (sink_code_in_bb): Deal with zero uses by removing the stmt
31153 2017-04-24 Richard Biener <rguenther@suse.de>
31156 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
31157 pointer-based references.
31159 2017-04-24 Richard Biener <rguenther@suse.de>
31162 * pass_manager.h (pass_manager::operator new): Remove.
31163 (pass_manager::operator delete): Likewise.
31164 * passes.c (pass_manager::operator new): Remove.
31165 (pass_manager::operator delete): Likewise.
31166 (pass_manager::pass_manager): Zero individual pass members.
31168 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
31171 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
31172 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
31173 Check "XEXP (src, 1)" operand here.
31174 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
31175 Check "XEXP (src, 1)" operand here.
31176 (dimode_scalar_chain::make_vector_copies): Detect count register
31177 of a shift instruction. Zero extend count register from QImode
31178 to DImode to satisfy vector shift pattern count operand predicate.
31179 Substitute vector shift count operand with a DImode copy.
31180 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
31183 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
31185 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
31186 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
31187 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
31188 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
31189 (UNSPEC_NOREX_MEM): Remove definition.
31191 2017-04-21 Richard Biener <rguenther@suse.de>
31193 PR tree-optimization/79547
31194 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
31195 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
31196 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
31197 without any constraints.
31199 2017-04-21 Richard Biener <rguenther@suse.de>
31201 PR tree-optimization/78847
31202 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
31204 2017-04-21 Richard Biener <rguenther@suse.de>
31206 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
31207 (build_distinct_type_copy): Likewise.
31208 (build_variant_type_copy): Likewise.
31209 * tree.c (build_qualified_type): Pass down mem-stat info.
31210 (build_distinct_type_copy): Likewise.
31211 (build_variant_type_copy): Likewise.
31213 2017-04-21 Richard Biener <rguenther@suse.de>
31215 PR tree-optimization/80237
31216 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
31218 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
31219 for a simplified result.
31221 2016-04-21 Richard Biener <rguenther@suse.de>
31223 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
31224 sth as strict as a simple_iv but a chrec without symbols and an
31225 operand defined in the loop we are peeling (and not some subloop).
31226 (propagate_constants_for_unrolling): Propagate all constants.
31228 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
31231 * config/i386/i386.c (print_reg): Remove assert for disalowed
31232 regno values, call output_operand_lossage instead.
31234 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
31237 * config/i386/constraints.md (Yc): New register constraint.
31238 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
31239 Use Yc constraint for alternative 2 of operand 0. Remove
31240 preferred_for_speed attribute.
31242 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
31244 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
31245 lastprivate clauses in SIMT case.
31247 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
31249 * doc/invoke.texi (-Wextra-semi): Document new warning option.
31251 2017-04-20 Richard Biener <rguenther@suse.de>
31253 PR tree-optimization/57796
31254 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
31255 as N scalar stores.
31256 (vect_model_load_cost): Cost gathers as N scalar loads.
31258 2017-04-20 Richard Biener <rguenther@suse.de>
31260 * ggc-page.c (ggc_allocated_p): Rename to ...
31261 (safe_lookup_page_table_entry): ... this and return the lookup
31263 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
31265 2017-04-20 Richard Biener <rguenther@suse.de>
31267 PR tree-optimization/80453
31268 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
31269 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
31270 from the conditions.
31271 (vn_phi_eq): Pass them down.
31272 (vn_phi_lookup): Record them.
31273 (vn_phi_insert): Likewise.
31275 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
31277 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
31278 uninitialized variable warning to avoid buffer overrun.
31280 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
31283 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
31284 is suppressed for '{ 0 }' in C.
31286 2017-04-20 Jakub Jelinek <jakub@redhat.com>
31288 * BASE-VER: Set to 8.0.0.
31290 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
31292 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
31293 priority .init_array and .fini_array section with SECTION_NOTYPE
31296 2017-04-20 Jakub Jelinek <jakub@redhat.com>
31298 PR middle-end/80423
31299 * tree.h (build_array_type): Add typeless_storage default argument.
31300 * tree.c (type_cache_hasher::equal): Also compare
31301 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
31302 (build_array_type): Add typeless_storage argument, set
31303 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
31305 (build_nonshared_array_type): Adjust build_array_type_1 caller.
31306 (build_array_type): Likewise. Add typeless_storage argument.
31308 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
31309 Jakub Jelinek <jakub@redhat.com>
31311 PR tree-optimization/80426
31312 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
31313 operation on symbolic operands, also compute the overflow for the
31314 invariant part when the operation degenerates into a negation.
31316 2017-04-19 Jakub Jelinek <jakub@redhat.com>
31319 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
31320 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
31323 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
31325 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
31328 * config/avr/avr.c (tree.h): Include it.
31329 (cgraph.h): Include it.
31330 (avr_encode_section_info): Don't warn for uninitialized progmem
31331 variable if it's just an alias.
31333 2017-04-19 Richard Biener <rguenther@suse.de>
31336 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
31337 when needed by AutoPGO.
31339 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
31342 * doc/lto.texi: Remove an extra 'that'.
31344 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
31346 PR rtl-optimization/80429
31347 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
31348 are only used in debug insns.
31350 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
31351 Vladimir Makarov <vmakarov@redhat.com>
31353 * config/sparc/predicates.md (input_operand): Add comment. Return
31354 true for any memory operand when LRA is in progress.
31355 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
31357 2017-04-18 Jeff Law <law@redhat.com>
31360 * mips.md ({return,simple_return}_internal): Do not overwrite
31363 2017-04-18 Jakub Jelinek <jakub@redhat.com>
31365 PR tree-optimization/80443
31366 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
31367 instead of adding 1, subtract -1 and similarly instead of subtracting
31370 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
31372 PR rtl-optimization/80357
31373 * haifa-sched.c (tmp_bitmap): New variable.
31374 (model_recompute): Handle duplicate use records.
31375 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
31376 (free_global_sched_pressure_data): Free it.
31378 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
31381 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31382 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31383 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31384 instead of SYSTEM_HEADER_DIR.
31386 2017-04-18 Jeff Law <law@redhat.com>
31388 PR middle-end/80422
31389 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
31390 predecessors after walking up the insn chain.
31392 2017-04-18 Jakub Jelinek <jakub@redhat.com>
31395 * dwarf2out.c (modified_type_die): Try harder not to emit internal
31396 sizetype type into debug info.
31398 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
31401 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
31402 unneeded test for TARGET_UPPER_REGS_SF.
31403 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
31405 2017-04-18 Jakub Jelinek <jakub@redhat.com>
31408 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
31409 instead of gsi_after_labels.
31411 2017-04-18 Jeff Law <law@redhat.com>
31413 * regcprop.c (maybe_mode_change): Avoid creating copies of the
31417 2017-04-13 Jeff Law <law@redhat.com>
31418 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
31419 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
31421 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
31424 * config/avr/avr.c (intl.h): Include it.
31425 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
31427 2017-04-18 Martin Liska <mliska@suse.cz>
31429 PR gcov-profile/78783
31430 * gcov-tool.c (gcov_output_files): Validate that destination
31431 file is either removed by the tool or by a user.
31433 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
31434 Guy Benyei <guybe@mellanox.com>
31436 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
31437 block, and do not negate it, the stored id is already negative.
31439 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
31441 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
31443 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
31446 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
31447 masks of options that should be turned off if the VSX vector
31448 options are turned off.
31449 (OTHER_P8_VECTOR_MASKS): Likewise.
31450 (OTHER_VSX_VECTOR_MASKS): Likewise.
31451 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
31452 rs6000_disable_incompatible_switches to validate no type switches
31454 (rs6000_incompatible_switch): New function to disallow turning on
31455 other vector options if -mno-vsx, -mno-power8-vector, or
31456 -mno-power9-vector are specified.
31458 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
31460 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
31462 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
31464 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
31465 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
31466 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
31467 (ARG_POINTER_CFA_OFFSET): Likewise.
31469 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
31471 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
31472 conditions to take advantage of various optimizations.
31474 2017-04-13 Jeff Law <law@redhat.com>
31476 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
31477 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
31478 (zero_extendsidi2_dext): Likewise.
31480 2017-04-13 Jakub Jelinek <jakub@redhat.com>
31483 * fold-const.c (fold_ternary_loc): Revert
31484 use op0 instead of fold_convert_loc (loc, type, arg0) part of
31487 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
31489 PR rtl-optimization/80343
31490 * lra-remat.c (update_scratch_ops): Assign original hard reg to
31491 new scratch pseudo.
31493 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
31496 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
31497 to ubsan_encode_value.
31499 2017-04-13 Jeff Law <law@redhat.com>
31501 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
31502 appearing in DEBUG_INSNs.
31504 2017-04-13 Martin Liska <mliska@suse.cz>
31506 PR gcov-profile/80413
31507 * gcov-io.c (gcov_write_string): Copy to buffer just when
31508 allocated size is greater than zero.
31510 2017-04-13 Jakub Jelinek <jakub@redhat.com>
31513 * dwarf2out.c (decls_for_scope): Ignore declarations of
31514 current_function_decl in BLOCK_NONLOCALIZED_VARS.
31516 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
31519 * ipa-visibility.c (non_local_p): Fix typos.
31520 (localize_node): When localizing symbol in same comdat group,
31521 dissolve the group only when we know external symbols are going
31523 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
31525 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31527 PR tree-optimization/79390
31528 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
31529 order does not result in usable sequence, retry with reversed operand
31535 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
31536 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
31537 op0 instead of fold_convert_loc (loc, type, arg0).
31539 2017-04-12 Jeff Law <law@redhat.com>
31541 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
31542 has a delay slot in the generated code.
31544 * config/cris/cris.md (cris_preferred_reload_class): Return
31545 GENNONACR_REGS rather than GENERAL_REGS.
31547 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31550 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
31551 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
31552 signedness of the result type.
31554 2017-04-12 Richard Biener <rguenther@suse.de>
31555 Jeff Law <law@redhat.com>
31557 PR tree-optimization/80359
31558 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
31559 trim stores to TARGET_MEM_REFs.
31561 2017-04-12 Richard Biener <rguenther@suse.de>
31563 PR tree-optimization/79390
31564 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
31565 threading case even more.
31567 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
31570 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
31571 for quad_address_p for TImode, instead of just not indexed_address.
31573 2017-04-12 Richard Biener <rguenther@suse.de>
31574 Bernd Edlinger <bernd.edlinger@hotmail.de>
31576 PR middle-end/79671
31577 * alias.c (component_uses_parent_alias_set_from): Handle
31578 TYPE_TYPELESS_STORAGE.
31579 (get_alias_set): Likewise.
31580 * tree-core.h (tree_type_common): Add typeless_storage flag.
31581 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
31582 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
31583 for types containing members with TYPE_TYPELESS_STORAGE.
31584 (place_field): Likewise.
31585 (layout_type): Likewise for ARRAY_TYPE.
31586 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
31587 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
31588 TYPE_TYPELESS_STORAGE.
31589 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
31591 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31594 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
31595 first argument to type.
31597 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31601 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
31602 CONST0_RTX (mode) rather than const0_rtx where appropriate.
31603 (rs6000_expand_binop_builtin): Likewise.
31604 (rs6000_expand_ternop_builtin): Likewise; also add missing
31605 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
31606 vshasigma built-ins.
31607 * doc/extend.texi: Document that vec_xxpermdi's third argument
31608 must be a constant.
31610 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
31612 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
31613 Use shift_const cost parameter when calculating gain of STV shifts.
31615 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
31617 PR rtl-optimization/70478
31618 * lra-constraints.c (process_alt_operands): Check memory for
31619 disfavoring memory insn operand.
31621 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31623 PR middle-end/80100
31624 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
31625 left shift in unsigned HOST_WIDE_INT type.
31627 PR rtl-optimization/80385
31628 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
31629 (not (neg X)) into (plus X -1) for complex or non-integral modes.
31632 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
31633 if they have any depend clauses.
31635 2017-04-11 Martin Liska <mliska@suse.cz>
31638 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
31639 * ipa-split.c (split_function): Create a local comdat symbol
31640 if caller is in a comdat group.
31642 2017-04-11 Martin Liska <mliska@suse.cz>
31645 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
31648 2017-04-11 Martin Sebor <msebor@redhat.com>
31650 PR middle-end/80364
31651 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
31652 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
31654 (directive::set_width, directive::set_precision, format_character):
31656 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
31659 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
31662 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
31663 conflict, set target->arch_name instead of target->cpu_name.
31665 2017-04-11 Richard Biener <rguenther@suse.de>
31667 PR tree-optimization/80374
31668 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
31669 build_zero_cst, remove fold_convertible_p check again.
31671 2017-04-11 Martin Liska <mliska@suse.cz>
31674 * ubsan.c (instrument_object_size): Do not instrument register
31677 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31680 * config/i386/i386-builtin-types.def
31681 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
31682 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
31683 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
31684 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
31685 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
31686 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
31687 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
31688 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
31689 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
31690 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
31691 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
31692 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
31693 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
31694 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
31695 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
31696 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
31697 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
31698 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
31699 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
31700 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
31701 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
31702 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
31703 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
31704 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
31705 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
31706 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
31707 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
31708 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
31709 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
31710 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
31711 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
31712 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
31713 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
31714 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
31715 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
31716 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
31717 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
31718 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
31719 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
31720 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
31721 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
31722 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
31723 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
31724 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
31725 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
31726 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
31728 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
31729 flag to second_arg_count, handle 4 argument function type _COUNT
31730 aliases, handle second_arg_count on second argument rather than last.
31732 2017-04-10 Jeff Law <law@redhat.com>
31734 PR tree-optimization/80374
31735 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
31736 record anything if we can not convert integer_zero_node to the
31739 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
31742 * config/rs6000/rs6000.c (rs6000_option_override_internal):
31743 Enhance special handling given to the TARGET_P9_MINMAX option in
31744 relation to certain other options.
31746 2017-04-10 Bin Cheng <bin.cheng@arm.com>
31748 PR tree-optimization/80153
31749 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
31750 remove POINTER_PLUS_EXPR's base part directly, rather than through
31753 2017-04-10 Richard Biener <rguenther@suse.de>
31754 Bin Cheng <bin.cheng@arm.com>
31756 PR tree-optimization/80153
31757 * tree-affine.c (aff_combination_to_tree): Get base pointer from
31758 the first element of pointer type aff_tree. Build result expr in
31760 (add_elt_to_tree): Convert to type unconditionally. Remove other
31761 fold_convert calls.
31762 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
31763 (rewrite_use_nonlinear_expr): Check invariant using iv information.
31765 2017-04-10 Richard Biener <rguenther@suse.de>
31767 * tree-ssa-structalias.c (find_func_aliases): Properly handle
31770 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
31772 PR rtl-optimization/70478
31773 * lra-constraints.c (curr_small_class_check): New.
31774 (update_and_check_small_class_inputs): New.
31775 (process_alt_operands): Update curr_small_class_check. Disfavor
31776 alternative insn memory operands. Check available regs for small
31779 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31782 * config/mips/mips.opt (-mvirt): Update description.
31783 * doc/invoke.texi (-mvirt): Likewise.
31785 2017-04-10 Richard Biener <rguenther@suse.de>
31787 PR middle-end/80362
31788 * fold-const.c (fold_binary_loc): Look at unstripped ops when
31789 looking for NEGATE_EXPR in -A / -B to A / B folding.
31791 2017-04-10 Martin Liska <mliska@suse.cz>
31793 PR gcov-profile/80224
31794 * gcov.c (print_usage): Fix usage string.
31795 (get_gcov_intermediate_filename): Remove.
31796 (output_gcov_file): Use both for normal and intermediate format.
31797 (generate_results): Do not initialize special file for
31798 intermediate format.
31800 2017-04-10 Richard Biener <rguenther@suse.de>
31802 PR tree-optimization/80304
31803 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
31806 2017-04-10 Nathan Sidwell <nathan@acm.org>
31809 * config/rs6000/rs6000.c (rs6000_vector_type): New.
31810 (rs6000_init_builtins): Use it.
31812 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31814 * config/arm/arm.md (<mrc>): Add mode to SET source.
31815 (<mrrc>): Likewise.
31817 2017-04-10 Richard Biener <rguenther@suse.de>
31819 PR middle-end/80344
31820 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
31822 2017-04-10 Jakub Jelinek <jakub@redhat.com>
31825 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
31826 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
31827 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
31828 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
31829 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
31830 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
31831 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
31832 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
31833 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
31834 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
31835 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
31836 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
31837 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
31838 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
31839 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
31840 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
31841 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
31842 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
31843 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
31844 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
31845 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
31846 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
31847 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
31849 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
31851 PR rtl-optimization/70478
31852 * lra-constraints.c: Reverse the last patch.
31854 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
31856 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
31857 Add comment for WCHAR_T.
31859 2017-04-08 Martin Liska <mliska@suse.cz>
31862 2017-04-07 Martin Liska <mliska@suse.cz>
31865 * ipa-split.c (split_function): Add function part to a same comdat
31868 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31871 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
31873 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
31875 * rs6000/rs6000.c (vec_load_pendulum): Rename...
31876 (vec_pairing): ...to this.
31877 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
31878 (rs6000_sched_init): Adjust for name change.
31879 (struct rs6000_sched_context): Likewise.
31880 (rs6000_init_sched_context): Likewise.
31881 (rs6000_set_sched_context): Likewise.
31883 2017-04-07 Jakub Jelinek <jakub@redhat.com>
31889 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
31891 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
31892 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
31893 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
31895 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
31897 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
31899 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31901 PR rtl-optimization/70703
31902 * ira-color.c (update_conflict_hard_regno_costs): Use
31903 int64_t instead of HOST_WIDE_INT.
31905 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31907 PR rtl-optimization/70478
31908 * lra-constraints.c (process_alt_operands): Disfavor alternative
31909 insn memory operands.
31911 2017-04-07 Jeff Law <law@redhat.com>
31913 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
31914 CALL and NOTE_INSN_CALL_ARG_LOCATION.
31916 2017-04-07 Martin Liska <mliska@suse.cz>
31919 * config/aarch64/aarch64.c (aarch64_process_target_attr):
31920 Show error message instead of an ICE.
31922 2017-04-07 Martin Liska <mliska@suse.cz>
31925 * ipa-split.c (split_function): Add function part to a same comdat
31928 2017-04-07 Richard Biener <rguenther@suse.de>
31930 PR middle-end/80341
31931 * tree.c (get_unwidened): Also handle ! for_type case for
31933 * convert.c (do_narrow): Split out from ...
31934 (convert_to_integer_1): ... here. Do not pass final truncation
31935 type to get_unwidened for TRUNC_DIV_EXPR.
31937 2017-04-07 Richard Biener <rguenther@suse.de>
31939 * tree-affine.c (wide_int_ext_for_comb): Take type rather
31941 (aff_combination_const): Adjust.
31942 (aff_combination_scale): Likewise.
31943 (aff_combination_add_elt): Likewise.
31944 (aff_combination_add_cst): Likewise.
31945 (aff_combination_convert): Likewise.
31946 (add_elt_to_tree): Likewise. Remove unused argument.
31947 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
31949 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
31951 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
31953 * config/arm/arm.c (arm_default_short_enums): Use
31954 ARM_DEFAULT_SHORT_ENUMS.
31955 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
31957 2017-04-06 Jakub Jelinek <jakub@redhat.com>
31960 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
31961 members with redundant out-of-class redeclaration.
31963 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31966 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
31967 * config/i386/i386.md (*zero_extendsidi2):
31968 Add (?*x,*x) and (?*v,*v) alternatives.
31970 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31973 * config/i386/i386.c (ix86_expand_builtin)
31974 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
31975 mode from insn data. Convert operands to insn operand mode.
31976 Copy operands that don't satisfy insn predicate to a register.
31978 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
31980 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
31983 2017-04-06 Richard Biener <rguenther@suse.de>
31985 PR tree-optimization/80334
31986 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
31987 preserve alignment of accesses.
31989 2017-04-06 Richard Biener <rguenther@suse.de>
31991 PR tree-optimization/80262
31992 * tree-sra.c (build_ref_for_offset): Preserve address-space
31994 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
31995 Drop useless address-space information on MEM_REF offsets.
31997 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
31999 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
32001 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
32003 PR rtl-optimization/70703
32004 * ira-color.c (update_conflict_hard_regno_costs): Use
32005 HOST_WIDE_INT instead of long.
32007 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
32010 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
32011 not defined for x86_64 target. Add -mmmx target option when __SSE2__
32013 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
32014 for x86_64 target. Handle -m3dnowa option.
32016 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
32018 PR rtl-optimization/70703
32019 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
32020 (update_conflict_hard_regno_costs): Use long instead of unsigned
32021 arithmetic for cost calculation.
32023 2017-04-05 Jakub Jelinek <jakub@redhat.com>
32024 Bernd Edlinger <bernd.edlinger@hotmail.de>
32027 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
32030 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
32033 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
32034 ptr_mode with Pmode throughout.
32035 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
32036 into probe_stack_range and use DImode.
32038 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
32041 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
32042 call_eh_return is true.
32044 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32046 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
32047 Initialize last_match_fntype_index.
32049 2017-04-05 Jakub Jelinek <jakub@redhat.com>
32052 * tree-nvr.c: Include internal-fn.h.
32053 (pass_return_slot::execute): Ignore internal calls without
32056 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32057 Richard Biener <rguenther@suse.de>
32060 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
32061 captures used multiple times, except for the last use.
32062 * generic-match-head.c: Include gimplify.h.
32064 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32066 PR tree-optimization/79390
32067 * target.h (struct noce_if_info): Declare.
32068 * targhooks.h (default_noce_conversion_profitable_p): Declare.
32069 * target.def (noce_conversion_profitable_p): New target hook.
32070 * ifcvt.h (struct noce_if_info): New type, moved from ...
32071 * ifcvt.c (struct noce_if_info): ... here.
32072 (noce_conversion_profitable_p): Renamed to ...
32073 (default_noce_conversion_profitable_p): ... this. No longer
32075 (noce_try_store_flag_constants, noce_try_addcc,
32076 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
32077 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
32078 instead of noce_conversion_profitable_p.
32079 * config/i386/i386.c: Include ifcvt.h.
32080 (ix86_option_override_internal): Don't override
32081 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
32082 (ix86_noce_conversion_profitable_p): New function.
32083 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
32084 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
32085 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
32086 * doc/tm.texi: Regenerated.
32088 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32090 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
32093 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32096 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
32097 instructions for small multiply cores.
32099 2017-04-04 Jeff Law <law@redhat.com>
32101 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
32103 (mips_expand_vec_perm_const): Initialize elements in orig_perm
32104 that are not set by the loop over the elements.
32106 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32109 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
32110 int mode, convert_modes it to mode as unsigned, otherwise use
32111 lowpart_subreg to mode rather than SImode.
32112 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
32113 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
32114 Use DImode instead of SImode for the shift count operand.
32115 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
32118 2017-04-04 Richard Biener <rguenther@suse.de>
32120 PR middle-end/80281
32121 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
32122 arithmetic done for the negate or the plus. Simplify.
32123 (A - (-B) -> A + B): Likewise.
32124 * fold-const.c (split_tree): Make sure to not negate pointers.
32126 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
32128 PR rtl-optimization/60818
32129 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
32130 a compare of comparisons with the thing compared if this results
32131 in a different machine mode.
32133 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
32135 * alias.c (base_alias_check): Fix typo in comment.
32136 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
32137 * cgraphunit.c (symbol_table::compile): Likewise.
32138 * collect2.c (maybe_run_lto_and_relink): Likewise.
32139 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
32140 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
32141 * config/avr/avr.c (avr_map_op_t): Likewise.
32142 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
32143 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
32144 * config/epiphany/epiphany.md (movcc): Likewise.
32145 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
32146 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
32148 * config/mips/mips.c (mips_save_restore_reg): Likewise.
32149 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
32150 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
32151 * config/sh/sh.c (sh_rtx_costs): Likewise.
32152 * fold-const.c (fold_truth_andor): Likewise.
32153 * genautomata.c (collapse_flag): Likewise.
32154 * gengtype.h (struct type::u::s): Likewise.
32155 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
32156 * input.c (FORMAT_AMOUNT): Likewise.
32157 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
32158 (known_aggs_to_agg_replacement_list): Likewise.
32159 * ipa-inline-analysis.c: Likewise.
32160 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
32161 * ipa-polymorphic-call.c
32162 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
32163 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
32164 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
32166 * modulo-sched.c (apply_reg_moves): Likewise.
32167 * omp-expand.c (build_omp_regions_1): Likewise.
32168 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
32169 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
32170 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
32171 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
32172 * value-prof.c: Likewise.
32173 * var-tracking.c (val_reset): Likewise.
32175 2017-04-03 Richard Biener <rguenther@suse.de>
32177 PR tree-optimization/80275
32178 * fold-const.c (split_address_to_core_and_offset): Handle
32181 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
32183 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
32184 descriptors is at least equal to that of functions.
32186 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
32188 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
32190 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
32193 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
32194 (mov<IMOD4:mode>): New expander.
32195 (*mov<IMOD4:mode>_internal): New insn and split pattern.
32197 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
32199 PR rtl-optimization/79405
32200 * fwprop.c (propagations_left): New variable.
32201 (forward_propagate_into): Decrement it.
32202 (fwprop_init): Initialize it.
32203 (fw_prop): If the variable has reached zero, stop propagating.
32204 (fwprop_addr): Ditto.
32206 2017-03-31 Jakub Jelinek <jakub@redhat.com>
32209 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
32210 a FUNCTION_DECL, pass it as decl instead of origin to
32213 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
32215 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
32218 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
32221 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
32222 TARGET_VSX_SMALL_INTEGER.
32224 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32226 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
32227 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
32229 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
32231 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
32232 extraction from odd-numbered MSA register.
32234 2017-03-31 Jakub Jelinek <jakub@redhat.com>
32236 PR middle-end/80173
32237 * expmed.c (store_bit_field_1): Don't attempt to create
32238 a word subreg out of hard registers wider than word if they
32239 have HARD_REGNO_NREGS of 1 for their mode.
32241 PR middle-end/80163
32242 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
32243 conversions to integer types wider than word and pointer.
32246 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
32247 (rtx_equal_for_cselib_p): Pass 0 to it.
32248 * cselib.c (cselib_hasher::equal): Likewise.
32249 (rtx_equal_for_cselib_1): Add depth argument. If depth
32250 is 128, don't look up VALUE locs and punt. Increment
32251 depth in recursive calls when walking VALUE locs.
32253 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
32255 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
32256 (make_gcov_file_name): Use the canonical path name for generating
32258 (read_line): Fix handling of files with ascii null bytes.
32260 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
32262 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
32263 to initialise a vector register instead
32264 of using a const_int.
32266 2017-03-30 Jakub Jelinek <jakub@redhat.com>
32268 PR translation/80189
32269 * gimplify.c (omp_default_clause): Use %qs instead of %s in
32270 diagnostic messages.
32272 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
32275 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
32276 (dfp_diex_<mode>): Update mode of operand 1.
32277 * doc/extend.texi (dxex, dxexq): Document change to return type.
32278 (diex, diexq): Document change to argument type.
32280 2017-03-30 Martin Jambor <mjambor@suse.cz>
32283 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
32284 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
32285 it reflects the signature changes performed at the callee side.
32286 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
32287 to cgraph_build_function_type_skip_args.
32288 (build_function_decl_skip_args): Adjust call to the above function.
32290 2017-03-30 Jakub Jelinek <jakub@redhat.com>
32293 * config/i386/sse.md
32294 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
32295 register as dest whenever it is a MEM not rtx_equal_p to the
32296 corresponding dup operand, and when forcing into reg move the
32297 reg into the memory afterwards.
32298 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
32299 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
32300 for the force_reg mode.
32301 (avx512vl_vextractf128<mode>): Use register as dest either
32302 always when a MEM, or when it is a MEM not rtx_equal_p to the
32303 corresponding dup operand, or even not when it is a CONST_VECTOR
32304 depending on the mode and lo vs. hi.
32305 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
32307 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
32308 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
32309 Likewise. Require that operands[2] is even.
32310 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
32311 Remove extraneous parens. Require that operands[2] is a multiple
32313 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
32314 operands[0] is a MEM if <mask_applied>, the predicates/constraints
32315 disallow memory then.
32317 2017-03-30 Richard Biener <rguenther@suse.de>
32319 PR tree-optimization/77498
32320 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
32321 to non-constants over backedges.
32323 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
32325 PR rtl-optimization/80233
32326 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
32327 as last_combined_insn. Do not test for BARRIER_P separately.
32329 2017-03-29 Andreas Schwab <schwab@suse.de>
32332 * calls.c (prepare_call_address): Convert funexp to Pmode before
32333 copying to temp reg.
32335 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32337 PR tree-optimization/80158
32338 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
32339 Handle possible future case of more than one alternate
32341 (replace_rhs_if_not_dup): Likewise.
32342 (replace_one_candidate): Likewise.
32344 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
32346 PR rtl-optimization/80193
32347 * ira.c (ira): Do not check allocation for LRA.
32349 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
32351 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
32352 (nvptx_output_simt_exit): Declare.
32353 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
32354 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
32355 (init_softstack_frame): Move initialization of crtl->is_leaf to...
32356 (nvptx_declare_function_name): ...here. Emit declaration of local
32357 memory space buffer for omp_simt_enter insn.
32358 (nvptx_output_unisimt_switch): New.
32359 (nvptx_output_softstack_switch): New.
32360 (nvptx_output_simt_enter): New.
32361 (nvptx_output_simt_exit): New.
32362 * config/nvptx/nvptx.h (struct machine_function): New fields
32363 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
32364 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
32365 (UNSPECV_SIMT_EXIT): Ditto.
32366 (omp_simt_enter_insn): New insn.
32367 (omp_simt_enter): New expansion.
32368 (omp_simt_exit): New insn.
32369 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
32371 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
32372 (expand_GOMP_SIMT_ENTER_ALLOC): New.
32373 (expand_GOMP_SIMT_EXIT): New.
32374 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
32375 (GOMP_SIMT_ENTER_ALLOC): Ditto.
32376 (GOMP_SIMT_EXIT): Ditto.
32377 * target-insns.def (omp_simt_enter): New insn.
32378 (omp_simt_exit): Ditto.
32379 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
32381 (lower_rec_simd_input_clauses): Implement SIMT privatization.
32382 (lower_rec_input_clauses): Likewise.
32383 (lower_lastprivate_clauses): Handle SIMT privatization.
32385 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
32386 (ompdevlow_adjust_simt_enter): New.
32387 (find_simtpriv_var_op): New.
32388 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
32389 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
32391 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
32392 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
32393 (copy_decl_for_dup_finish): Ditto.
32395 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
32397 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
32400 * config/i386/i386.c (ix86_option_override_internal): Always
32401 allow -mpreferred-stack-boundary=3 for 64-bit targets.
32403 2017-03-28 Bin Cheng <bin.cheng@arm.com>
32405 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
32407 2017-03-28 Bin Cheng <bin.cheng@arm.com>
32409 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
32410 mark new edge's irreducible flag accordign to it.
32411 (vect_do_peeling): Check loop preheader edge's irreducible flag
32412 and pass it to function slpeel_add_loop_guard.
32414 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
32416 PR tree-optimization/80218
32417 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
32418 Update block frequencies and counts.
32420 2017-03-28 Richard Biener <rguenther@suse.de>
32422 PR tree-optimization/78644
32423 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
32424 of a simplification result we may not use it at all.
32426 2017-03-28 Richard Biener <rguenther@suse.de>
32429 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
32430 without arguments, generate default definition of a SSA name.
32432 2017-03-28 Richard Biener <rguenther@suse.de>
32434 PR middle-end/80222
32435 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
32436 TYPE_REF_CAN_ALIAS_ALL references.
32437 * fold-const.c (fold_indirect_ref_1): Likewise.
32439 2017-03-28 Martin Liska <mliska@suse.cz>
32442 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
32443 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
32445 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
32446 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
32448 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
32449 (EXTRA_SPECS): Define.
32450 (SUBTARGET_EXTRA_SPECS): Likewise.
32451 (SUBTARGET_CPP_SPEC): Likewise.
32452 * config/arc/elf.h (EXTRA_SPECS): Renamed to
32453 SUBTARGET_EXTRA_SPECS.
32454 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
32456 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
32458 * config/arc/simdext.md (vst64_insn): Update pattern.
32459 (vld32wh_insn): Likewise.
32460 (vld32wl_insn): Likewise.
32461 (vld64_insn): Likewise.
32462 (vld32_insn): Likewise.
32464 2017-03-28 Marek Polacek <polacek@redhat.com>
32467 * fold-const.c (fold_comparison): Use protected_set_expr_location
32468 instead of SET_EXPR_LOCATION.
32470 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
32472 * tree.c (add_expr): Avoid name lookup warning.
32474 2017-03-27 Jeff Law <law@redhat.com>
32476 PR tree-optimization/80216
32477 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
32478 function name. Limit recursion depth.
32479 (record_temporary_equivalences): Corresponding changes.
32481 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
32483 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
32486 2017-03-27 Jakub Jelinek <jakub@redhat.com>
32489 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
32491 * cfgcleanup.c (reg_note_cfa_p): New array.
32492 (insns_have_identical_cfa_notes): New function.
32493 (old_insns_match_p): Don't cross-jump in between /f
32494 and non-/f instructions. If both i1 and i2 are frame related,
32495 verify all CFA notes, their order and content.
32497 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
32500 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
32501 HImode and SImode with zero extend to DImode to one insn.
32502 (bswap<mode>2_extenddi): Likewise.
32503 (bswapsi2_extenddi): Likewise.
32504 (bswaphi2_extendsi): Likewise.
32505 (bswaphi2): Combine bswap HImode and SImode into one insn.
32506 Separate memory insns from swapping register.
32507 (bswapsi2): Likewise.
32508 (bswap<mode>2): Likewise.
32509 (bswaphi2_internal): Delete, no longer used.
32510 (bswapsi2_internal): Likewise.
32511 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
32512 store, and gpr<-gpr swap insns.
32513 (bswap<mode>2_store): Likewise.
32514 (bswaphi2_reg): Register only splitter, combine with the splitter.
32515 (bswaphi2 splitter): Likewise.
32516 (bswapsi2_reg): Likewise.
32517 (bswapsi2 splitter): Likewise.
32518 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
32519 the insns into load, store, and register/register insns.
32520 (bswapdi2_ldbrx): Likewise.
32521 (bswapdi2_load): Likewise.
32522 (bswapdi2_store): Likewise.
32523 (bswapdi2_reg): Likewise.
32525 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
32527 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
32528 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
32530 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
32533 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
32535 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32536 special handling for target option conflicts between dform
32537 options (-mpower9-dform, -mpower9-dform-vector,
32538 -mpower9-dform-scalar) and -mno-direct-move.
32540 2017-03-27 Richard Biener <rguenther@suse.de>
32542 PR tree-optimization/80181
32543 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
32545 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32547 * config/arc/predicates.md (move_double_src_operand): Replace the
32548 call to move_double_src_operand with a call to address_operand.
32550 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32552 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
32553 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
32554 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
32556 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32558 * config/arc/predicates.md (long_immediate_loadstore_operand):
32559 Consider scaled addresses cases.
32561 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32563 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
32564 restored when in interrupt.
32565 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
32566 doesn't have delay slot.
32568 2017-03-27 Richard Biener <rguenther@suse.de>
32571 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
32572 inlined thunk clones.
32574 2017-03-27 Jakub Jelinek <jakub@redhat.com>
32577 * asan.c (instrument_derefs): Copy over last operand from
32578 original COMPONENT_REF to the new COMPONENT_REF with
32579 DECL_BIT_FIELD_REPRESENTATIVE.
32580 * ubsan.c (instrument_object_size): Likewise.
32582 2017-03-27 Richard Biener <rguenther@suse.de>
32584 PR tree-optimization/80170
32585 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
32586 sure DR/SCEV didnt fold in constants we do not see when looking
32587 at the reference base alignment.
32589 2017-03-27 Richard Biener <rguenther@suse.de>
32591 PR middle-end/80171
32592 * gimple-fold.c (fold_ctor_reference): Properly guard against
32593 NULL return value from canonicalize_constructor_val.
32595 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
32598 * config/i386/i386.c (ix86_expand_builtin)
32599 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
32600 flags reg setting and flags reg using instructions.
32601 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
32602 clobbering instructions to zero extend op2.
32604 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
32606 * doc/install.texi (Configuration) <--with-aix-soname>:
32607 Update link to AIX ld.
32609 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
32611 PR rtl-optimization/80160
32612 PR rtl-optimization/80159
32613 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
32614 reg_alternate_class into account.
32616 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
32619 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
32620 to consider in curr_insn_transform.
32622 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32624 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
32625 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
32626 and emit_mode_inner.
32628 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32630 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
32631 argument to the overloaded builtin variants. Use the new flag to
32632 deprecate certain builtin variants.
32633 * config/s390/s390-builtin-types.def: Add new builtin types.
32634 * config/s390/s390-builtins.h: Support new flags field for
32635 overloaded builtins.
32636 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
32637 (s390_macro_to_expand): Enable vector float data type.
32638 (s390_cpu_cpp_builtins_internal): Indicate support of the new
32639 builtins by incrementing the __VEC__ version number.
32640 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
32642 (s390_resolve_overloaded_builtin): Emit error messages depending
32643 on the builtin flags.
32644 * config/s390/s390.c (s390_expand_builtin): Support additional
32645 flags argument. Change error message to match the messages
32646 emitted in s390-c.c.
32647 * config/s390/s390.md: New UNSPEC_* constants.
32648 (op_type): Add new instruction types.
32649 * config/s390/vecintrin.h: Add new builtins and test data class
32651 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
32652 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
32653 (VEC_INEXACT, VEC_NOINEXACT): New constants.
32654 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
32655 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
32656 ("vec_mergel<mode>"): V_HW -> VEC_HW.
32658 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
32659 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
32660 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
32661 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
32663 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
32664 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
32665 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
32666 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
32668 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
32669 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
32670 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
32671 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
32672 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
32673 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
32674 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
32676 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
32677 ("vec_scatter_element<V_HW_4:mode>_DI")
32678 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
32679 ("vec_fpint<mode>", "vflls")
32680 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
32681 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
32682 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
32683 ("*vec_cmphe<mode>_cc"): ... these.
32685 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
32686 mode constant instead of magic value.
32688 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32690 * config/s390/s390.c (s390_expand_vec_compare): Support other
32691 vector floating point modes than just V2DF.
32692 (s390_expand_vcond): Likewise.
32693 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
32694 (s390_cannot_change_mode_class): Prevent mode changes between TF
32695 and V1TF in vector registers.
32696 * config/s390/s390.md (DF, SF): New mode attributes.
32697 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
32698 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
32699 SFmode support for VRs.
32700 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
32702 (VFT, VF_HW): New mode iterators.
32703 (vw, sdx): New mode attributes.
32704 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
32705 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
32706 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
32707 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
32708 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
32709 also the new vector floating point modes. Renaming to ...
32711 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
32712 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
32713 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
32714 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
32715 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
32716 ("vec_unordered<mode>"): ... these.
32718 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
32719 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
32720 ("*vec_extendv2df"): New insn definitions.
32722 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32724 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
32725 ("mulditi3_2", "*muldi3_sign"): New patterns.
32726 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
32727 rename the pattern definition.
32729 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32731 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
32733 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
32735 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32737 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
32738 instruction if possible.
32739 * config/s390/vector.md (vec_halfnumelts): New mode
32741 ("*vec_vllezlf<mode>"): New pattern.
32743 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32745 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
32746 ("popcountv4si2", "popcountv2di2"): Rename to ...
32747 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
32748 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
32750 ("popcount<mode>2_vxe"): New pattern.
32752 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32754 * common/config/s390/s390-common.c (processor_flags_table): Add
32756 * config.gcc: Add arch12.
32757 * config/s390/driver-native.c (s390_host_detect_local_cpu):
32758 Default to arch12 for unknown CPU model numbers.
32759 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
32760 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
32761 PROCESSOR_max sanity check.
32762 * config/s390/s390-opts.h (enum processor_type): Add
32764 * config/s390/s390.c (processor_table): Add arch12.
32765 (s390_expand_builtin): Add check for B_VXE flag.
32766 (s390_issue_rate): Add PROCESSOR_ARCH12.
32767 (s390_get_sched_attrmask): Likewise.
32768 (s390_get_unit_mask): Likewise.
32769 (s390_sched_score): Enable z13 scheduling for arch12.
32770 (s390_sched_reorder): Likewise.
32771 (s390_sched_variable_issue): Likewise.
32772 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
32774 (s390_tune_attr): Use z13 scheduling also for arch12.
32775 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
32776 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
32777 (TARGET_VXE_P): New macros.
32778 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
32779 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
32780 * config/s390/s390.opt: Add arch12 as processor_type.
32782 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32784 * config/s390/s390.md
32785 ("fixuns_truncdddi2", "fixuns_trunctddi2")
32786 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
32787 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
32789 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
32790 Rename expanders to ...
32792 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
32793 ("fixuns_truncdddi2_emu"): ... these.
32795 ("fixuns_trunc<mode>si2_emu"): New expander.
32797 ("*fixuns_truncdfdi2_z13"): Rename to ...
32798 ("*fixuns_truncdfdi2_vx"): ... this.
32800 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32802 * config/s390/2964.md: Remove the single element vector compare
32803 instructions which are no longer used.
32804 * config/s390/s390.c (s390_select_ccmode): Remove handling of
32806 (s390_canonicalize_comparison): Remove handling of DFmode
32808 (s390_expand_vec_compare_scalar): Remove function.
32809 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
32810 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
32812 ("*cmp<mode>_ccs"): Add wfcdb instruction.
32814 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32816 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
32818 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
32819 will anyway by matched by mov<mode>_64dfp.
32821 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32823 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
32824 vlef/vstef. Add missing operand to vleif.
32826 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32828 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
32829 pair for all vector types with 64 bit elements.
32830 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
32831 * config/s390/vector.md (V_HW_64): ... here.
32832 (V_128_NOSINGLE): New mode iterator.
32833 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
32834 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
32835 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
32836 ("*vec_load_pairv2di"): Change to ...
32837 ("*vec_load_pair<mode>"): ... this one.
32839 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32841 * config/s390/constraints.md: Add comments.
32842 (jKK): Reject element sizes > 8 bytes.
32843 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
32845 * config/s390/s390.md: Add the s_operand checks formerly in
32846 s390_split_ok_p to various splitters where they are still
32848 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
32849 for 128 bit vectors. Plus two splitters.
32851 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32853 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
32856 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32859 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
32860 error if the boundary argument is not constant.
32862 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32864 PR rtl-optimization/80112
32865 * loop-doloop.c (doloop_condition_get): Don't check condition
32866 if cmp isn't SET with IF_THEN_ELSE src.
32868 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32870 PR tree-optimization/80158
32871 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
32872 replacing a candidate statement, also replace it for the
32873 candidate's alternate interpretation.
32874 (replace_rhs_if_not_dup): Likewise.
32875 (replace_one_candidate): Likewise.
32877 2017-03-24 Richard Biener <rguenther@suse.de>
32879 PR tree-optimization/80167
32880 * graphite-isl-ast-to-gimple.c
32881 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
32883 (translate_isl_ast_to_gimple::get_rename): Likewise.
32885 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32887 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
32888 handling of certain combinations of target options, including the
32889 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
32890 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
32892 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32895 * config/arm/arm.md (*load_multiple): Add reload_completed to
32896 matching condition.
32898 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32899 Richard Biener <rguenth@suse.de>
32901 PR tree-optimization/79908
32902 PR tree-optimization/80136
32903 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32904 been cast away, gimplify_and_add suffices.
32906 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
32908 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
32910 2017-03-23 Richard Biener <rguenther@suse.de>
32912 PR tree-optimization/80032
32913 * gimplify.c (gimple_push_cleanup): Forced unconditional
32914 cleanups still have to go to the conditional_cleanups
32917 2017-03-22 Jakub Jelinek <jakub@redhat.com>
32919 PR tree-optimization/80072
32920 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
32922 (next_operand_entry_id): Change type to unsigned int.
32923 (sort_by_operand_rank): Make sure to return the right return value
32924 even if unsigned fields are bigger than INT_MAX.
32925 (struct oecount): Change cnt and id type to unsigned int.
32926 (oecount_hasher::equal): Formatting fix.
32927 (oecount_cmp): Make sure to return the right return value
32928 even if unsigned fields are bigger than INT_MAX.
32929 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
32932 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
32933 TREE_READONLY on result if writing it more than once.
32936 * doc/invoke.texi (-fsanitize=thread): Document that with
32937 -fnon-call-exceptions atomics are not able to throw
32941 * tsan.c: Include tree-eh.h.
32942 (instrument_builtin_call): Call maybe_clean_eh_stmt or
32943 maybe_clean_or_replace_eh_stmt where needed.
32944 (instrument_memory_accesses): Add cfg_changed argument.
32945 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
32946 if it returned true.
32947 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
32949 PR rtl-optimization/63191
32950 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
32951 wrapper function, moved the whole old content into ...
32952 (ix86_delegitimize_address_1): ... this. New inline function.
32953 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
32954 true as last argument instead of ix86_delegitimize_address.
32956 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32958 * config/aarch64/aarch64.c (generic_branch_cost): Copy
32959 cortexa57_branch_cost.
32961 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32963 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
32965 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32968 * doc/md.texi (Constraints): Document wA constraint.
32969 * config/rs6000/constraints.md (wA): New.
32970 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
32971 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
32972 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
32973 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
32975 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
32978 * gimplify.c (is_oacc_declared): New function.
32979 (oacc_default_clause): Use it to set default flags for acc declared
32980 variables inside parallel regions.
32981 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
32982 declared variables.
32983 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
32984 declare attribute to any decl as necessary.
32986 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
32989 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
32990 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
32991 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
32992 (arm_arch_lpae): This.
32993 * config/arm/arm.c (arm_arch7ve): Rename into ...
32994 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
32995 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
32998 2017-03-22 Martin Liska <mliska@suse.cz>
33001 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
33002 error message instead of an ICE.
33004 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33006 * doc/extend.texi (6.11 Additional Floating Types): Revise.
33008 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
33010 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
33012 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33015 2017-03-21 Martin Sebor <msebor@redhat.com>
33017 * doc/extend.texi: Use "cannot" instead of "can't."
33018 * doc/hostconfig.texi: Same.
33019 * doc/install.texi: Same.
33020 * doc/invoke.texi: Same.
33021 * doc/loop.texi: Same.
33022 * doc/md.texi: Same.
33023 * doc/objc.texi: Same.
33024 * doc/rtl.texi: Same.
33025 * doc/tm.texi: Same.
33026 * doc/tm.texi.in: Same.
33027 * doc/trouble.texi: Same.
33029 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
33032 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
33033 (collect_checksum_attributes): Set it.
33034 (die_checksum_ordered): Use it.
33036 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33038 PR tree-optimization/79908
33039 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
33040 change: For a VA_ARG whose LHS has been cast away, use
33041 force_gimple_operand to construct the side effects.
33043 2017-03-21 David Malcolm <dmalcolm@redhat.com>
33045 PR translation/80001
33046 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
33047 more amenable to translation.
33048 (oacc_loop_auto_partitions): Likewise.
33050 2017-03-21 Marek Polacek <polacek@redhat.com>
33051 Martin Sebor <msebor@redhat.com>
33053 PR tree-optimization/80109
33054 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
33055 on INTEGRAL_TYPE_P.
33057 2017-03-21 Jakub Jelinek <jakub@redhat.com>
33058 Segher Boessenkool <segher@kernel.crashing.org>
33061 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
33062 check reg_used_between_p between insn and one of succ or succ2
33063 depending on if succ is artificial insn not inserted into insn
33066 2017-03-21 Martin Liska <mliska@suse.cz>
33068 PR gcov-profile/80081
33069 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
33070 * doc/gcc.texi: Include gcov-dump stuff.
33071 * doc/gcov-dump.texi: New file.
33073 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
33075 PR rtl-optimization/79150
33076 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
33077 conditional jump, if the jump is the last insn of the loop.
33079 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33080 Richard Biener <rguenth@suse.de>
33082 PR tree-optimization/79908
33083 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
33084 been cast away, use force_gimple_operand to construct the side
33087 2017-03-21 Martin Liska <mliska@suse.cz>
33089 PR libfortran/79956
33090 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
33093 2017-03-21 Brad Spengler <spender@grsecurity.net>
33096 * plugin.c (htab_hash_plugin): New function.
33097 (add_new_plugin): Use it and adjust.
33098 (parse_plugin_arg_opt): Adjust.
33099 (init_one_plugin): Likewise.
33101 2017-03-21 Richard Biener <rguenther@suse.de>
33103 PR tree-optimization/80032
33104 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
33105 if set force the cleanup to happen unconditionally.
33106 (gimplify_target_expr): Push inserted clobbers with force_uncond
33107 to avoid them being removed by control-dependent DCE.
33109 2017-03-21 Richard Biener <rguenther@suse.de>
33111 PR tree-optimization/80122
33112 * tree-inline.c (copy_bb): Do not expans va-arg packs or
33113 va_arg_pack_len when the inlined call stmt requires pack
33115 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
33117 2017-03-21 Jakub Jelinek <jakub@redhat.com>
33120 * tsan.c (instrument_builtin_call): If the memory model argument
33121 is not a constant, assume it is valid.
33124 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
33127 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
33129 PR rtl-optimization/79910
33130 * combine.c (can_combine_p): Do not allow combining an I0 or I1
33131 if its dest is used by an insn before I2 (other than the combined
33132 insns themselves, which are properly handled already).
33134 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
33137 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
33139 * combine.c (record_used_regs): New static function.
33140 (try_combine): Handle situations where there is an additional
33141 instruction between I2 and I3 which needs to have a LOG_LINK
33145 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
33147 * combine.c (try_combine): Delete redundant i1 test. Call
33148 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
33150 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33153 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
33154 alternatives 13/14.
33156 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33158 PR tree-optimization/80054
33159 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
33160 the optimization if a PHI or any of its arguments is not dominated
33161 by the candidate's basis. Use gphi* rather than gimple* as
33163 (replace_profitable_candidates): Clean up a gimple* variable that
33164 should be a gphi* variable.
33166 2017-03-20 Martin Sebor <msebor@redhat.com>
33169 * doc/extend.texi (attribute constructor): Document present limitation.
33171 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
33174 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
33175 __POWER9_VECTOR__ #ifdef control, change template definition to
33176 use Power9-specific built-in function.
33177 (vec_any_eq): Likewise.
33178 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
33179 to control outcomes from this test.
33180 (vector_ae_<mode>p): For VEC_F modes, likewise.
33182 2017-03-20 Ian Lance Taylor <iant@google.com>
33184 * config/i386/i386.c (ix86_function_regparm): Save an extra
33185 register for -fsplit-stack with DECL_STATIC_CHAIN.
33187 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
33190 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
33191 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
33193 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
33195 * config/riscv/riscv.c (riscv_print_operand): Use "fence
33197 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
33200 2017-03-20 Marek Polacek <polacek@redhat.com>
33203 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
33205 2017-03-20 Richard Biener <rguenther@suse.de>
33207 PR tree-optimization/80113
33208 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
33209 allocate extra SSA name for PHI def.
33210 (add_close_phis_to_outer_loops): Likewise.
33211 (add_close_phis_to_merge_points): Likewise.
33212 (copy_loop_close_phi_args): Likewise.
33213 (copy_cond_phi_nodes): Likewise.
33215 2017-03-20 Martin Liska <mliska@suse.cz>
33217 PR middle-end/79753
33218 * tree-chkp.c (chkp_build_returned_bound): Do not build
33219 returned bounds for a LHS that's not a BOUNDED_P type.
33221 2017-03-20 Martin Liska <mliska@suse.cz>
33225 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
33226 COMPLEX_CST and VECTOR_CST.
33228 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33231 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
33232 target operand. A new splitter adds the clobber statement in case
33233 the target operand is dead anyway.
33235 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
33237 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
33238 to age-old versions of binutils and glibc.
33240 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
33242 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
33244 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
33246 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
33248 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
33250 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
33251 requirement for binutils 2.13.
33253 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
33255 * combine.c (try_combine): Delete redundant i1 test. Call
33256 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
33258 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
33260 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
33261 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
33263 <riscv64-*-elf>: Re-arrange section
33264 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
33265 <riscv32-*-linux>: Likewise.
33266 <riscv64-*-elf>: Likewise
33267 <riscv64-*-linux>: Likewise.
33269 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
33272 * aarch64.opt(verbose-cost-dump): Fix typo.
33274 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
33277 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
33278 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
33280 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
33282 * reload.c (find_reloads): When reloading a nonoffsettable address,
33283 use RELOAD_OTHER for it and its address reloads.
33285 PR rtl-optimization/79910
33286 * combine.c (record_used_regs): New static function.
33287 (try_combine): Handle situations where there is an additional
33288 instruction between I2 and I3 which needs to have a LOG_LINK
33291 2017-03-17 Jeff Law <law@redhat.com>
33293 PR tree-optimization/71437
33294 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
33295 conditional in the hash table first.
33296 (vrp_dom_walker::before_dom_children): Extract condition from
33297 ASSERT_EXPR. Record condition, its inverion and any implied
33298 conditions as well.
33300 2017-03-17 Marek Polacek <polacek@redhat.com>
33301 Markus Trippelsdorf <markus@trippelsdorf.de>
33303 PR tree-optimization/80079
33304 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
33307 2017-03-17 Richard Biener <rguenther@suse.de>
33309 PR middle-end/80075
33310 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
33311 Properly verify the LHS before the RHS possibly claims to be
33313 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
33316 2017-03-17 Martin Jambor <mjambor@suse.cz>
33318 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
33319 (List of -O2 options): Likewise.
33320 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
33323 2017-03-17 Tom de Vries <tom@codesourcery.com>
33325 * gcov-dump.c (print_usage): Print bug_report_url.
33327 2017-03-17 Richard Biener <rguenther@suse.de>
33329 PR middle-end/80050
33330 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
33331 (parser::peek): Likewise.
33333 2017-03-17 Richard Biener <rguenther@suse.de>
33335 PR tree-optimization/80048
33336 * sese.c (free_sese_info): Properly release rename_map and
33337 copied_bb_map elements.
33339 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
33341 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
33342 Add linked-list forward and backlinks. Insert on
33343 construction, remove on destruction.
33344 (class pass_store_merging): Add m_stores_head field.
33345 (pass_store_merging::terminate_and_process_all_chains):
33346 Iterate over m_stores_head list.
33347 (pass_store_merging::terminate_all_aliasing_chains):
33349 (pass_store_merging::execute): Check for debug stmts first.
33350 Push new chains onto the m_stores_head stack.
33352 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
33355 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
33356 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
33357 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
33359 2017-03-16 Jeff Law <law@redhat.com>
33361 PR tree-optimization/71437
33362 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
33363 member function. Implementation moved into after_dom_children
33364 member function and into the threader's thread_outgoing_edges
33366 (dom_opt_dom_walker::after_dom_children): Simplify by moving
33367 some code into new thread_outgoing_edges.
33368 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
33369 definition. Simplify marker handling (do it here). Assume we always
33370 have the available expression and the const/copies tables.
33371 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
33373 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
33374 * tree-vrp.c (equiv_stack): No longer file scoped.
33375 (vrp_dom_walker): New class.
33376 (vrp_dom_walker::before_dom_children): New member function.
33377 (vrp_dom_walker::after_dom_children): Likewise.
33378 (identify_jump_threads): Setup domwalker. Use it rather than
33379 walking edges in a random order by hand. Simplify setup/finalization.
33380 (finalize_jump_threads): Remove.
33381 (vrp_finalize): Do not call identify_jump_threads here.
33382 (execute_vrp): Do it here instead and call thread_through_all_blocks
33385 PR tree-optimization/71437
33386 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
33388 (simplify_stmt_for_jump_threading): Add basic_block argument. All
33390 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
33391 (dom_opt_dom_walker::thread_across_edge): Remove
33392 handle_dominating_asserts argument. All callers changed.
33393 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
33394 changes. Remove calls to lhs_of_dominating_assert. Other
33395 uses of handle_dominating_asserts turn into unconditional code
33396 (simplify_control_stmt_condition_1): Likewise.
33397 (simplify_control_stmt_condition): Likewise.
33398 (thread_through_normal_block, thread_across_edge): Likewise.
33399 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
33400 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
33401 object if it is not an SSA_NAME.
33402 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
33403 before calling into the VRP specific simplifiers.
33404 (identify_jump_threads): Remove handle_dominating_asserts
33407 2017-03-16 Jakub Jelinek <jakub@redhat.com>
33410 * tree-diagnostic.c (default_tree_printer): No longer static.
33411 * tree-diagnostic.h (default_tree_printer): New prototype.
33413 2017-03-16 Tamar Christina <tamar.christina@arm.com>
33415 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
33416 Change ins into fmov.
33418 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33420 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
33421 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
33422 Use h_con constraint for operand 1.
33423 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
33424 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
33426 2017-03-15 Jeff Law <law@redhat.com>
33428 PR tree-optimization/71437
33429 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
33430 (record_temporary_equivalences): Use it.
33432 PR tree-optimization/71437
33433 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
33434 tree-ssa-scopedtables.
33435 (lookup_avail_expr, build_and_record_new_cond): Likewise.
33436 (record_conditions, record_cond, vuse_eq): Likewise.
33437 (record_edge_info): Adjust to API tweak of record_conditions.
33438 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
33439 (record_temporary_equivalences, optimize_stmt): Likewise.
33440 (eliminate_redundant_computations): Likewise.
33441 (record_equivalences_from_stmt): Likewise.
33442 * tree-ssa-scopedtables.c: Include options.h and params.h.
33443 (vuse_eq): New function, moved from tree-ssa-dom.c
33444 (build_and_record_new_cond): Likewise.
33445 (record_conditions): Likewise. Accept vector of conditions rather
33446 than edge_equivalence structure for first argument.
33447 for the first argument.
33448 (avail_exprs_stack::lookup_avail_expr): New member function, moved
33449 from tree-ssa-dom.c.
33450 (avail_exprs_stack::record_cond): Likewise.
33451 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
33452 from tree-ssa-dom.c.
33453 (avail_exprs_stack): Add new member functions lookup_avail_expr
33455 (record_conditions): Declare.
33457 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
33460 * lra-constraints.c (process_alt_operands): Increase reject for
33461 reloading an input/output operand.
33463 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
33466 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
33467 insns to convert from signed/unsigned char/short to IEEE 128-bit
33469 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
33471 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
33474 * config/i386/i386.c (ix86_vector_duplicate_value): Create
33475 subreg of inner mode for values already in registers.
33477 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
33479 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
33480 iteration reg is used after the loop.
33482 2017-03-14 Martin Sebor <msebor@redhat.com>
33484 PR tree-optimization/79800
33485 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
33486 precision in negative-positive range.
33487 (format_floating): Call non-const overload with adjusted precision.
33489 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33492 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
33495 2017-03-14 Martin Sebor <msebor@redhat.com>
33497 PR middle-end/80020
33498 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
33499 * builtins.def (aligned_alloc): Use it.
33502 * Makefile.in (GTFILES): Add calls.c.
33503 * calls.c: Include "gt-calls.h".
33505 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
33507 PR rtl-optimization/79728
33508 * regs.h (struct target_regs): New field
33509 x_contains_allocatable_regs_of_mode.
33510 (contains_allocatable_regs_of_mode): New macro.
33511 * reginfo.c (init_reg_sets_1): Initialize it, and change
33512 contains_reg_of_mode so it includes global regs as well.
33513 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
33514 rather than contains_regs_of_mode.
33516 2017-03-14 Martin Liska <mliska@suse.cz>
33518 * doc/invoke.texi: Document options that can't be combined with
33519 -fcheck-pointer-bounds.
33521 2017-03-14 Martin Liska <mliska@suse.cz>
33523 PR middle-end/79831
33524 * doc/invoke.texi (-Wchkp): Document the option.
33526 2017-03-14 Martin Liska <mliska@suse.cz>
33528 * Makefile.in: Install gcov-dump.
33530 2017-03-14 Martin Liska <mliska@suse.cz>
33532 * multiple_target.c (expand_target_clones): Bail out for
33533 an invalid attribute.
33535 2017-03-14 Richard Biener <rguenther@suse.de>
33537 * alias.c (struct alias_set_entry): Pack properly.
33538 * cfgloop.h (struct loop): Likewise.
33539 * cse.c (struct set): Likewise.
33540 * ipa-utils.c (struct searchc_env): Likewise.
33541 * loop-invariant.c (struct invariant): Likewise.
33542 * lra-remat.c (struct cand): Likewise.
33543 * recog.c (struct change_t): Likewise.
33544 * rtl.h (struct address_info): Likewise.
33545 * symbol-summary.h (function_summary): Likewise.
33546 * tree-loop-distribution.c (struct partition): Likewise.
33547 * tree-object-size.c (struct object_size_info): Likewise.
33548 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
33549 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
33550 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
33551 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
33552 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
33553 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
33554 (struct _stmt_vec_info): Likewise.
33556 2017-03-14 Martin Liska <mliska@suse.cz>
33559 * multiple_target.c (create_dispatcher_calls): Check that
33560 a target can create a function dispatcher.
33562 2017-03-14 Martin Liska <mliska@suse.cz>
33565 * multiple_target.c (expand_target_clones): Drop local.local
33566 flag for default implementation.
33568 2017-03-14 Richard Biener <rguenther@suse.de>
33570 PR tree-optimization/80030
33571 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
33573 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
33575 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
33576 gcc_fallthrough() instead of __attribute__((fallthrough));
33578 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
33580 * doc/gcc.texi: Remove "up" link to (DIR).
33581 * doc/gccint.texi: Ditto.
33583 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
33585 * doc/install.texi (Specific) <avr>: Remove reference to
33588 2017-03-13 Jeff Law <law@redhat.com>
33590 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
33591 attribute rather than comments.
33593 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
33594 match_scratch operand is highest.
33596 2017-03-13 Martin Liska <mliska@suse.cz>
33598 PR middle-end/78339
33599 * ipa-pure-const.c (warn_function_noreturn): If the declarations
33600 is a CHKP clone, use original declaration.
33602 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33604 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
33605 (arc_conditional_register_usage): Use a different allocation order
33606 when optimizing for size.
33607 * common/config/arc/arc-common.c (arc_option_optimization_table):
33608 Section anchors default on when optimizing for size.
33610 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33612 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
33614 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33616 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
33617 * config/arc/arc.md (cpu_facility): Add cd variant.
33618 (*movqi_insn): Add code density variant.
33619 (*movhi_insn): Likewise.
33620 (*movqi_insn): Likewise.
33621 (*addsi3_mixed): Likewise.
33622 (subsi3_insn): Likewise.
33624 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33626 * config/arc/arc.md (movsi_cond_exec): Update constraint.
33628 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33630 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
33631 expressions with MINUS and UNARY ops.
33633 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33636 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
33638 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
33639 between vec_select and vector argument.
33640 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
33641 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
33642 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
33643 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
33644 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
33645 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
33647 2017-03-13 Richard Biener <rguenther@suse.de>
33650 * params.def (vect-max-peeling-for-alignment): Fix typo.
33652 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33654 * doc/install.texi (Specific) <mips-*-*>: Remove description of
33655 issue that only occurred with binutils below 2.18.
33657 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33659 * doc/install.texi (Specific) <cris-axis-elf>: No longer
33660 refer to binutils 2.11/2.12 minimum.
33662 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33664 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
33665 ftp.kernel.org and simplify binutils requirement.
33667 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
33669 * doc/invoke.texi (Warning Options): Fix spelling of link-time
33671 (Optimize Options): Ditto. Also remove redundancy.
33673 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33675 PR translation/79848
33676 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
33678 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
33679 to G_ to avoid double translation.
33681 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33683 PR translation/79923
33684 * auto-profile.c (get_combined_location): Convert leading
33685 character of diagnostics to lower case and remove trailing period.
33686 (read_profile): Likewise for various diagnostics.
33687 * config/arm/arm.c (arm_option_override): Remove trailing period
33688 from various diagnostics.
33689 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
33690 (msp430_expand_delay_cycles): Likewise.
33692 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33695 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
33696 full command-line argument, rather than just "str".
33697 (aarch64_validate_march): Likewise.
33698 (aarch64_validate_mtune): Likewise.
33700 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
33702 PR rtl-optimization/78911
33703 * lra-assigns.c (must_not_spill_p): New function.
33704 (spill_for): Use it.
33706 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33708 PR tree-optimization/79981
33709 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
33710 ATOMIC_COMPARE_EXCHANGE ifn result.
33711 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
33712 IFN_ATOMIC_COMPARE_EXCHANGE.
33714 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33717 * opts.c (parse_sanitizer_options): Add missing question mark to
33718 "did you mean" message.
33720 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33722 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
33724 (VMULEUH_UNS): Likewise.
33725 (VMULOUB_UNS): Likewise.
33726 (VMULOUH_UNS): Likewise.
33727 * config/rs6000/rs6000.c (builtin_function_type): Remove
33728 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
33730 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33733 * read-rtl-function.c (function_reader::read_rtx_operand): Update
33734 x with result of extra_parsing_for_operand_code_0.
33735 (function_reader::extra_parsing_for_operand_code_0): Convert
33736 return type from void to rtx, returning x. When reading
33737 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
33738 larger size containing struct block_symbol.
33740 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
33742 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
33743 -mfloat128-hardware without -m64.
33745 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
33748 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
33749 entries to the case statement that marks unsigned arguments to
33750 overloaded functions.
33752 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33754 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
33755 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
33757 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
33760 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
33761 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
33763 2017-03-10 Martin Liska <mliska@suse.cz>
33767 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
33768 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
33771 2017-03-10 Olivier Hainque <hainque@adacore.com>
33773 * tree-switch-conversion (array_value_type): Start by resetting
33774 candidate type to it's main variant.
33776 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33778 PR rtl-optimization/79909
33779 * combine.c (try_combine): Use simplify_replace_rtx on individual
33780 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
33781 of the whole CALL_INSN_FUNCTION_USAGE.
33783 PR tree-optimization/79972
33784 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
33785 get_range_info on SSA_NAMEs. Formatting fixes.
33787 2017-03-10 Richard Biener <rguenther@suse.de>
33788 Jakub Jelinek <jakub@redhat.com>
33790 PR tree-optimization/77975
33791 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
33792 edge to be constant.
33793 (get_val_for): For constant x return it. Formatting fix.
33794 (loop_niter_by_eval): Avoid pointless looping if the next iteration
33795 would use the same bases as the current one.
33797 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33799 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
33800 instead of vec_select for V1TImode.
33801 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
33803 (VSX_LE_128): Add V1TI to this mode iterator.
33804 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
33805 (*vsx_le_perm_store_<mode>): Likewise.
33806 (pre-reload splitter for VSX stores): Likewise.
33807 (post-reload splitter for VSX stores): Likewise.
33808 (*vsx_xxpermdi2_le_<mode>): Likewise.
33809 (*vsx_lxvd2x2_le_<mode>): Likewise.
33810 (*vsx_stxvd2x2_le_<mode>): Likewise.
33812 2017-03-09 Michael Eager <eager@eagercon.com>
33814 Correct failures with --enable-checking=yes,rtl.
33816 * config/microblaze/microblaze.c (microblaze_expand_shift):
33817 Replace GET_CODE test with CONST_INT_P and INTVAL test with
33818 test for const0_rtx.
33819 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
33820 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
33822 2017-03-09 Richard Biener <rguenther@suse.de>
33824 PR tree-optimization/79977
33825 * graphite-scop-detection.c (scop_detection::merge_sese):
33826 Handle the case of extra exits to blocks dominating the entry.
33828 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
33830 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
33833 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
33835 PR rtl-optimization/79949
33836 * lra-constraints.c (process_alt_operands): Check memory when
33837 trying to predict a cycle. Print about the overall increase.
33839 2017-03-09 Richard Biener <rguenther@suse.de>
33841 PR middle-end/79971
33842 * gimple-expr.c (useless_type_conversion_p): Preserve
33843 TYPE_SATURATING for fixed-point types.
33845 2017-03-09 Richard Biener <rguenther@suse.de>
33848 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
33849 alignment of BLKmode params.
33851 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33854 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
33855 (VALL_NO_V2Q): Likewise.
33857 * config/aarch64/aarch64-simd.md
33858 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
33860 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
33861 VALL_NO_V2Q mode iterator.
33862 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
33864 2017-03-09 Martin Liska <mliska@suse.cz>
33866 PR tree-optimization/79631
33867 * tree-chkp-opt.c (chkp_is_constant_addr): Call
33868 tree_int_cst_sign_bit just for INTEGER constants.
33870 2017-03-09 Martin Liska <mliska@suse.cz>
33874 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
33877 2017-03-09 Marek Polacek <polacek@redhat.com>
33880 * tree.c (inchash::add_expr): Handle TREE_VEC.
33882 2017-03-09 Martin Liska <mliska@suse.cz>
33885 (chkp_narrow_size_and_offset): New function.
33886 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
33887 (void chkp_parse_bit_field_ref): New function.
33888 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
33889 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
33891 2017-03-09 Martin Liska <mliska@suse.cz>
33894 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
33895 (chkp_find_bounds_1): Remove gcc_unreachable.
33897 2017-03-09 Jakub Jelinek <jakub@redhat.com>
33900 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
33901 BUILT_IN_SYNC*, determine the access type from the size suffix and
33902 always build a MEM_REF with that type. Handle forgotten
33903 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
33906 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
33907 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
33908 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
33909 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
33910 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
33911 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
33912 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
33913 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
33914 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
33915 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
33916 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
33917 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
33918 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
33919 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
33920 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
33921 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
33922 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
33923 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
33924 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
33925 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
33926 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
33927 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
33928 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
33929 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
33930 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
33931 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
33932 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
33933 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
33934 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
33935 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
33936 definitions outside of __OPTIMIZE__ guarded section.
33939 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
33940 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
33941 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
33942 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
33945 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33947 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
33948 ("vfenez<mode>"): Add missing constraints.
33950 2017-03-08 Martin Sebor <msebor@redhat.com>
33953 * config/nds32/nds32.c (nds32_option_override):
33954 Fix misspelled diagnostic.
33956 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33959 * gimplify.c (gimplify_omp_for): Replace index var in outer
33960 taskloop statement with an artificial variable and add
33961 OMP_CLAUSE_PRIVATE clause for it.
33963 2017-03-08 Richard Biener <rguenther@suse.de>
33965 PR tree-optimization/79955
33966 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
33967 for accesses that are completely outside of the variable.
33969 2017-03-08 Andrew Haley <aph@redhat.com>
33971 PR tree-optimization/79943
33972 * tree-ssa-loop-split.c (compute_new_first_bound): When
33973 calculating the new upper bound, (END-BEG) should be added, not
33976 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33978 * config/avr/avr.md (setmemhi): Make sure match_dup
33979 operand number comes before match_scratch.
33981 2017-03-08 Richard Biener <rguenther@suse.de>
33983 PR tree-optimization/79920
33984 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
33985 with ncopies == 1 to ...
33986 (vect_transform_slp_perm_load): ... here. Properly compute
33987 all element loads by iterating VF times over the group. Do
33988 not handle ncopies (computed in a broken way) in
33989 vect_create_mask_and_perm.
33991 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33994 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
33995 is a uniform vector, use uniform_vector_p return value instead of
33996 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
33998 2017-03-07 Marek Polacek <polacek@redhat.com>
34000 PR middle-end/79809
34001 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
34002 (alloca_call_type): Likewise.
34004 2017-03-07 Martin Liska <mliska@suse.cz>
34006 * gcov.c (process_args): Put comment to correct location.
34008 2017-03-07 Martin Liska <mliska@suse.cz>
34010 PR middle-end/68270
34011 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
34012 Use array_at_struct_end_p instead of DECL_CHAIN (field).
34013 (chkp_narrow_bounds_for_field): Likewise.
34014 (chkp_parse_array_and_component_ref): Pass one more argument to
34017 2017-03-07 Richard Biener <rguenther@suse.de>
34019 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
34022 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
34024 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
34025 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
34027 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34030 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
34031 to end of description.
34032 (PARAM_MAX_STORES_TO_MERGE): Likewise.
34034 2017-03-07 Jakub Jelinek <jakub@redhat.com>
34036 PR rtl-optimization/79901
34037 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
34039 (*avx512f_<code><mode>3<mask_name>): ... this.
34040 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
34041 iterator instead of VI8_AVX2_AVX512BW.
34043 PR rtl-optimization/79901
34044 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
34045 min/max expander, expand it using expand_vec_cond_expr.
34048 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
34051 2017-03-06 Jakub Jelinek <jakub@redhat.com>
34054 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
34055 to void * for PCH reasons.
34056 * dwarf2out.c (output_loc_operands, output_die): Cast
34057 v.val_vec.array to unsigned char *.
34059 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
34062 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
34065 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
34067 PR rtl-optimization/79571
34068 * lra-constraints.c (process_alt_operands): Calculate static
34069 reject and subtract it from overall when only addresses will be
34072 2017-03-06 Julia Koval <julia.koval@intel.com>
34075 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
34076 incoming stack boundary to 128 for 64-bit targets.
34078 2017-03-06 Richard Biener <rguenther@suse.de>
34080 PR tree-optimization/79894
34081 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
34082 to NULL after folding it.
34084 2017-03-06 Richard Biener <rguenther@suse.de>
34086 PR tree-optimization/79824
34087 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
34088 check disabling peeling for gaps.
34090 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
34092 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
34093 attributes): Document gettimeofday.
34095 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
34097 * config/s390/s390.c (s390_option_override_internal): Set
34098 PARAM_MIN_VECT_LOOP_BOUND
34100 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
34102 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
34103 * config/s390/s390.md: Likewise.
34105 2017-03-06 Jakub Jelinek <jakub@redhat.com>
34108 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
34109 (<avx2_avx512>_perm<mode>): Rename to ...
34110 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
34112 (<avx512>_perm<mode>_mask): Rename to ...
34113 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
34115 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
34116 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
34117 instead of VI8F_256_512.
34118 (avx512f_perm<mode>): New define_expand.
34119 (avx512f_perm<mode>_mask): Likewise.
34120 (avx512f_perm<mode>_1<mask_name>): New define_insn.
34121 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
34123 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34125 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
34126 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
34128 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
34130 2017-03-06 Martin Liska <mliska@suse.cz>
34133 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
34134 when having a SSA NAME w/o VAR_DECL assigned to it.
34136 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34138 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
34139 msa_dpsub_<su>_d): Fix MODE for vec_select.
34141 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34143 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
34145 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
34147 2017-03-06 Richard Biener <rguenther@suse.de>
34149 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
34150 * plugin.c (register_plugin_info): Likewise.
34151 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
34153 2017-03-05 Jakub Jelinek <jakub@redhat.com>
34155 * config/i386/sse.md (sse_storehps, sse_storelps,
34156 avx_<castmode><avxsizesuffix>_<castmode>,
34157 avx512f_<castmode><avxsizesuffix>_<castmode>,
34158 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
34159 in condition that at least one operand is not a MEM.
34161 2017-03-03 Jakub Jelinek <jakub@redhat.com>
34163 PR middle-end/79805
34164 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
34165 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
34167 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
34168 gimple_call_nothrow_p flag based on whether original builtin can throw.
34169 If it can, emit following stmts on the fallthrough edge.
34170 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
34171 don't create new bb if inserting just debug stmts on the edge, try to
34172 insert them on the fallthru bb or just reset debug stmts.
34174 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
34177 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
34178 restore recog_data (including the operand rtxes inside it) around
34179 the call to get_insn_template.
34181 2017-03-03 Martin Sebor <msebor@redhat.com>
34183 PR tree-optimization/79699
34184 * context.c (context::~context): Free MPFR caches to avoid
34185 a memory leak on program exit.
34187 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34189 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
34190 Use wide_int::ulow () instead of .elt (0).
34192 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
34194 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
34195 (*pushxf): Limit oF constraint to 32bit targets and add oC
34196 constraint for 64bit targets.
34197 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
34198 (*pushdf): Change rmF constraint to rmC.
34200 2017-03-03 Martin Liska <mliska@suse.cz>
34202 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
34203 Remove unused variable.
34205 2017-03-03 Jakub Jelinek <jakub@redhat.com>
34208 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
34209 is a memory operand, increase num_memory.
34210 (ix86_expand_args_builtin): Likewise.
34212 2017-03-03 Jan Hubicka <jh@suse.cz>
34215 * ipa-devirt.c (maybe_record_node): Properly handle
34216 __cxa_pure_virtual visibility.
34218 2017-03-03 Martin Liska <mliska@suse.cz>
34220 PR tree-optimization/79803
34221 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
34223 (pass_loop_prefetch::execute): Disabled optimization if an
34224 assumption about L1 cache size is not met.
34226 2017-03-03 Martin Liska <mliska@suse.cz>
34228 PR rtl-optimization/79574
34229 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
34230 (hash_scan_set): Likewise.
34231 (dump_hash_table): Likewise.
34232 (hoist_code): Likewise.
34234 2017-03-03 Richard Biener <rguenther@suse.de>
34236 * fixed-value.c (fixed_from_string): Restore use of elt (1)
34237 in place of uhigh ().
34238 (fixed_convert_from_real): Likewise.
34240 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
34243 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
34245 2017-03-03 Richard Biener <rguenther@suse.de>
34247 PR middle-end/79818
34248 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
34249 TYPE_OVERFLOW_UNDEFINED check.
34251 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34253 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
34255 (vector_ae_<mode>_p): Likewise.
34256 (vector_nez_<mode>_p): Likewise.
34257 (vector_ne_v2di_p): Likewise.
34258 (vector_ae_v2di_p): Likewise.
34259 (vector_ne_<mode>_p): Likewise.
34260 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
34262 (vsx_tsqrt<mode>2_fe): Likewise.
34264 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
34267 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
34269 2017-03-02 Jakub Jelinek <jakub@redhat.com>
34271 PR rtl-optimization/79780
34272 * cprop.c (one_cprop_pass): When second and further conditional trap
34273 in a single basic block is turned into an unconditional trap, turn it
34274 into a deleted note to avoid RTL verification failures.
34276 2017-03-02 Richard Biener <rguenther@suse.de>
34278 * fold-const.c (const_binop): Use ulow () instead of elt (0).
34280 2017-03-02 Richard Biener <rguenther@suse.de>
34282 PR tree-optimization/79345
34284 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
34285 param and abort the walk, returning -1 if it is hit.
34286 (walk_aliased_vdefs): Take a limit param and pass it on.
34287 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
34288 defaulting to 0 and return a signed int.
34289 * tree-ssa-uninit.c (struct check_defs_data): New struct.
34290 (check_defs): New helper.
34291 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
34292 about uninitialized memory.
34293 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
34294 bogus uninitialized warning.
34295 (fixed_convert_from_real): Likewise.
34297 2017-03-02 Bin Cheng <bin.cheng@arm.com>
34299 PR tree-optimization/66768
34300 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
34301 iv_use if base object can't be determined.
34303 2017-03-02 Jakub Jelinek <jakub@redhat.com>
34305 PR tree-optimization/79345
34306 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
34307 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
34308 (get_pattern_stats): Initialize it.
34309 * genemit.c (gen_expand): Verify match_scratch numbers come after
34310 match_operand/match_dup numbers.
34311 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
34312 match_scratch numbers.
34313 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
34315 * config/s390/s390.md (trunctdsd2): Likewise.
34317 2017-03-02 Richard Biener <rguenther@suse.de>
34319 * wide-int.h (wide_int_storage::operator=): Implement in terms
34322 2017-03-02 Richard Biener <rguenther@suse.de>
34324 PR tree-optimization/79777
34325 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
34326 the to insert expression to sth existing.
34328 2017-03-01 Martin Sebor <msebor@redhat.com>
34330 PR middle-end/79692
34331 * gimple-ssa-sprintf.c
34332 (directive::known_width_and_precision): New function.
34333 (format_integer): Use it.
34334 (get_mpfr_format_length): Consider the full range of precision
34335 when computing %g output with the # flag. Set the likely byte
34336 count to 3 rather than 1 when precision is indeterminate.
34337 (format_floating): Correct the lower bound of precision.
34339 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34341 * doc/invoke.texi: Document default code model for 64-bit Linux.
34343 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34346 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
34347 udiv rather than div since input pattern is unsigned.
34349 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
34351 * config/i386/i386.c (print_reg): Warn for values of
34352 unsupported size in integer register.
34354 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
34357 * config/rs6000/predicates.md (current_file_function_operand): Do
34358 not allow self calls to be local if the function is replaceable.
34360 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
34363 * config/rs6000/altivec.h (vec_ctz and others): Change the
34364 preprocessor macro that controls conditional compilation from
34365 _ARCH_PWR9 to __POWER9_VECTOR__.
34366 (vec_all_ne): Change parameterization of __altivec_scalar_pred
34367 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
34368 control (instead of _ARCH_PWR9 control) so that template
34369 definition uses power9-specific function.
34370 (vec_any_eq): Likewise.
34371 (vec_all_ne): Change macro definition to use a power9-specific
34372 expansion under #ifdef __POWER9_VECTOR__ control (instead of
34373 _ARCH_PWR9 control).
34374 (vec_any_eq) Likewise.
34375 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
34376 expansion for CMPNEF to remove support for xvcmpnesp instruction.
34377 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
34378 support for xvcmpnedp instruction.
34379 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
34380 macro expansion so that Power9 implementation of vec_all_ne does
34381 not use the AltiVec predicate framework.
34382 (VCMPNEH_P): Likewise.
34383 (VCMPNEW_P): Likewise.
34384 (VCMPNED_P): Likewise.
34385 (VCMPNEFP_P): Likewise.
34386 (VCMPNEDP_P): Likewise.
34387 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
34388 implementation of vec_any_eq to not use AltiVec predicate
34390 (VCMPAEH_P): Likewise.
34391 (VCMPAEW_P): Likewise.
34392 (VCMPAED_P): Likewise.
34393 (VCMPAEFP_P): Likewise.
34394 (VCMPAEDP_P): Likewise.
34395 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
34396 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
34397 not use the AltiVec predicate framework.
34398 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
34399 of vec_any_eq to not use AltiVec predicate framework.
34400 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34401 support for predefined __POWER9_VECTOR__ macro to indicate that
34402 Power9 instruction selection is enabled.
34403 (altivec_overloaded_builtins): Remove extraneous
34404 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
34405 function argument types RS6000_BTI_bool_V16QI and
34406 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
34407 entry for overloaded function argument types RS6000_BTI_bool_V4SI
34408 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
34409 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
34410 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
34411 Power9 for implementations of vec_cmpne. Change the signature for
34412 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
34413 (representing vec_all_ne) to remove the previously described first
34414 argument of type RS6000_BTI_INTSI, as this was an artifact of
34415 reliance on the AltiVec predicate framework, which is no longer
34416 used in the implementation of these functions. Add
34417 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
34418 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
34419 since, unlike the AltiVec predicate framework implementation, we
34420 do not share function descriptors between vec_alle and vec_anyeq.
34421 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
34422 set of modes that receive special treatment even when
34423 TARGET_P9_VECTOR is true. The special treatment emits code that
34424 does not depend on Power9 instructions.
34425 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
34426 define_expand to not rely on AltiVec predicate framework.
34427 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
34429 (vector_ne_v2di_p): Change this define_expand to not rely on
34430 AltiVec predicate framework.
34431 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
34433 (vector_ne_<mode>_p): Change this define_expand to not rely on
34434 AltiVec predicate framework.
34435 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
34437 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
34438 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
34439 define_insn pattern.
34440 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
34441 define_insn pattern because the xvcmpne<VSs>. instruction is not
34443 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
34444 instruction is not supported.
34446 2017-03-01 Jakub Jelinek <jakub@redhat.com>
34448 * config/nvptx/nvptx.c: Include intl.h.
34450 2017-03-01 Martin Jambor <mjambor@suse.cz>
34453 * ipa-prop.h (ipa_bits): Removed field known.
34454 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
34455 to pointers. Adjusted their comments to warn about their sharing.
34456 (ipcp_transformation_summary): Change bits to a vector of pointers.
34457 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
34458 (ipa_get_ipa_bits_for_value): Declare.
34459 * tree-vrp.h (value_range): Mark as GTY((for_user)).
34460 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
34461 (ipa_bits_hash_table): Likewise.
34462 (ipa_vr_ggc_hash_traits): Likewise.
34463 (ipa_vr_hash_table): Likewise.
34464 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
34465 being pointers and vr_known being removed.
34466 (ipa_set_jf_unknown): Likewise.
34467 (ipa_get_ipa_bits_for_value): New function.
34468 (ipa_set_jfunc_bits): Likewise.
34469 (ipa_get_value_range): New overloaded functions.
34470 (ipa_set_jfunc_vr): Likewise.
34471 (ipa_compute_jump_functions_for_edge): Use the above functions to
34472 construct bits and vr parts of jump functions.
34473 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
34474 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
34476 (ipcp_grow_transformations_if_necessary): Also allocate
34477 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
34479 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
34480 them. Fix too long lines.
34481 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
34482 vr_known being removed.
34483 (ipa_read_jump_function): Use new setter functions to construct bits
34484 and vr parts of jump functions or set them to NULL.
34485 (write_ipcp_transformation_info): Adjust for bits being pointers.
34486 (read_ipcp_transformation_info): Likewise.
34487 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
34489 Include gt-ipa-prop.h.
34490 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
34492 (ipcp_store_bits_results): Likewise.
34493 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
34494 Do not write to existing jump functions but use a temporary instead.
34496 2017-03-01 Jakub Jelinek <jakub@redhat.com>
34499 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
34500 attempt to use its first operand as BIT_FIELD_REF base.
34502 2017-03-01 Richard Biener <rguenther@suse.de>
34504 PR middle-end/79721
34505 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
34506 interpolating formula in wrapping arithmetic.
34507 (chrec_apply): Convert chrec_evaluate return value to wanted type.
34509 2017-03-01 Jakub Jelinek <jakub@redhat.com>
34511 PR tree-optimization/79734
34512 * tree-vect-generic.c (expand_vector_condition): Optimize
34513 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
34514 Handle VEC_COND_EXPR where comparison has different inner width from
34515 type's inner width.
34517 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
34519 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
34520 markup, and similar issues. Remove @opindex entries for things
34521 that aren't options. Add missing -mmpy-option entries.
34523 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34525 PR tree-optimization/79737
34526 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
34527 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
34528 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
34529 instead of byte_size. Formatting fix.
34530 (shift_bytes_in_array_right): Formatting fix.
34532 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
34535 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
34536 condition on optimize for the leaf function test.
34538 2017-02-28 Martin Liska <mliska@suse.cz>
34541 * read-rtl-function.c (function_reader::handle_unknown_directive):
34542 Bail out when one uses -flto.
34544 2017-02-28 Martin Liska <mliska@suse.cz>
34546 * common.opt: Replace space with tabular for options of <number>
34548 * config/i386/i386.opt: Show <number> value for
34549 -mlarge-data-threshold.
34550 * opts.c (print_filtered_help): Do not display number in hexadecimal
34553 2017-02-28 Martin Liska <mliska@suse.cz>
34555 * common.opt: Fix --help=option -Q for options which are of
34558 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
34560 * config/i386/i386.c (print_reg): Error out for values
34561 of 8-bit size in invalid integer register.
34563 2017-02-28 Martin Sebor <msebor@redhat.com>
34565 PR tree-optimization/79691
34566 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
34568 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34571 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
34572 gcc_unreachable with output_operand_lossage.
34574 2017-02-28 Richard Biener <rguenther@suse.de>
34576 PR tree-optimization/79740
34577 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
34579 (visit_nary_op): Insert the nary into the hashtable if we
34580 pattern-matched sth.
34581 * tree-ssa-pre.c (eliminate_insert): Robustify.
34583 2017-02-28 Richard Biener <rguenther@suse.de>
34585 PR middle-end/79731
34586 * fold-const.c (decode_field_reference): Reject out-of-bound
34589 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34591 * config/i386/i386.c: Include intl.h.
34592 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
34593 instead of just cond ? "..." : "...".
34594 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
34595 * coverage.c (read_counts_file): Likewise.
34596 * omp-offload.c: Include intl.h.
34597 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
34598 of just cond ? "..." : "...".
34599 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
34600 of just cond ? "..." : "...".
34602 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
34605 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
34607 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
34608 'tune for' CPU name.
34609 * config/arm/arm-cpu-data.h: Regenerated.
34611 2017-02-28 Richard Biener <rguenther@suse.de>
34613 PR tree-optimization/79732
34614 * tree-inline.c (expand_call_inline): Do not shadow var.
34616 2017-02-28 Richard Biener <rguenther@suse.de>
34618 PR tree-optimization/79723
34619 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
34620 address-space properly.
34622 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
34624 * doc/optinfo.texi (Optimization groups): Fix option used for
34626 * doc/invoke.texi (-fopt-info): Document "omp".
34627 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
34628 (OPTGROUP_ALL): Add OPTGROUP_OMP.
34629 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
34630 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
34631 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
34633 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
34635 * dumpfile.c (optgroup_options): Instead of "openmp", associate
34636 OPTGROUP_OMP with "omp".
34638 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
34641 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
34642 for arithmetic shift of unsigned V2DI.
34644 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
34646 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
34647 arc/linux.h headers.
34648 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
34649 (LINK_SPEC): Likewise.
34650 (ARC_TLS_EXTRA_START_SPEC): Likewise.
34651 (EXTRA_SPECS): Likewise.
34652 (STARTFILE_SPEC): Likewise.
34653 (ENDFILE_SPEC): Likewise.
34654 (LIB_SPEC): Likewise.
34655 (TARGET_SDATA_DEFAULT): Likewise.
34656 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
34657 (MULTILIB_DEFAULTS): Likewise.
34658 (DWARF2_UNWIND_INFO): Likewise.
34659 * config/arc/big.h: New file.
34660 * config/arc/elf.h: Likewise.
34661 * config/arc/linux.h: Likewise.
34662 * config/arc/t-uClibc: Remove.
34664 2017-02-27 Bin Cheng <bin.cheng@arm.com>
34666 PR tree-optimization/77536
34667 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
34668 (tree_transform_and_unroll_loop): Use above function to compute the
34669 estimated niter of unrolled loop and use it when scaling profile.
34670 Also use count info rather than frequency if it's non-zero.
34671 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
34672 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
34673 (vect_transform_loop): Call above function.
34675 2017-02-27 Richard Biener <rguenther@suse.de>
34677 PR tree-optimization/45397
34678 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
34679 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
34680 (visit_nary_op): Add pattern matching for CSEing sign-changed
34681 or truncated operations with wider ones.
34683 2017-02-27 Richard Biener <rguenther@suse.de>
34685 PR tree-optimization/79690
34686 * tree-vect-stmts.c (vectorizable_store): Use vector type
34687 built from the DR with address-space.
34689 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
34691 * doc/invoke.texi (Optimize Options): Refine the description
34692 of asan-use-after-return.
34694 2017-02-25 Alan Modra <amodra@gmail.com>
34696 PR rtl-optimization/79584
34697 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
34698 base, not ad->base_term, the reg within base. Remove assertion
34699 that ad->base == ad->base_term. Replace gen_int_mode using
34700 bogus mode with const0_rtx.
34702 2017-02-25 Jakub Jelinek <jakub@redhat.com>
34704 PR middle-end/79396
34705 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
34706 FMA_EXPR like tcc_binary or tcc_unary.
34708 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
34711 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
34713 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
34714 (output_loc_operands): Handle DW_OP_call_ref and
34715 DW_OP_GNU_variable_value.
34716 (struct variable_value_struct): New type.
34717 (struct variable_value_hasher): Likewise.
34718 (variable_value_hash): New variable.
34719 (string_types): Remove.
34720 (copy_loc_descr): New function.
34721 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
34722 (prepend_loc_descr_to_each): New function.
34723 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
34724 instead of add_loc_descr_to_each if the first argument is single
34725 location list and the second has multiple.
34726 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
34727 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
34728 when looking for variable value which doesn't have other location info.
34729 (loc_list_from_tree): Formatting fix.
34730 (gen_array_type_die): Simplify DW_AT_string_length handling.
34731 (adjust_string_types): Remove.
34732 (gen_subprogram_die): Don't call adjust_string_types nor test/set
34733 string_types. Call resolve_variable_values.
34734 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
34735 (resolve_addr_in_expr): Likewise. Add A argument.
34736 (copy_deref_exprloc): Remove deref argument. Adjust for the
34737 original expression being DW_OP_GNU_variable_value with optionally
34738 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
34739 optionally after it.
34740 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
34741 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
34742 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
34743 (variable_value_hasher::hash, variable_value_hasher::equal): New
34745 (resolve_variable_value_in_expr, resolve_variable_value,
34746 resolve_variable_values, note_variable_value_in_expr,
34747 note_variable_value): New functions.
34748 (dwarf2out_early_finish): Call note_variable_value on all toplevel
34751 2017-02-24 Jakub Jelinek <jakub@redhat.com>
34754 * opts.h (handle_generated_option): Add GENERATED_P argument.
34755 * opts-common.c (handle_option): Adjust function comment.
34756 (handle_generated_option): Add GENERATED_P argument, pass it to
34758 (control_warning_option): Pass false to handle_generated_option
34760 * opts.c (maybe_default_option): Pass true to handle_generated_option
34762 * optc-gen.awk: Likewise.
34764 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34766 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
34767 a REG, look at the REG it is a SUBREG of.
34768 (splitter for cmpeqsi_t): Ditto.
34770 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34772 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
34773 the special USEs with the pattern of the insn, not the insn itself.
34775 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
34778 * doc/invoke.texi: Document -mload-store-pairs.
34780 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34781 Sandra Loosemore <sandra@codesourcery.com>
34783 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
34784 argument isn't a CONST_INT.
34785 (nios2_alternate_compare_const): Assert op is a CONST_INT.
34786 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
34787 (nios2_validate_compare): Bypass alternate compare logic if *op2
34788 is not a CONST_INT.
34789 (ldstwm_operation_p): Return false if first_base is not a REG or
34790 if first_offset is not a CONST_INT.
34792 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34794 * config/cris/cris.md: Use correct operand in a define_peephole2.
34796 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34798 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
34800 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34802 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
34803 this_insn if it is an INSN or JUMP_INSN.
34804 (force_offsettable): Look at base, not at addr.
34805 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
34806 on things that aren't necessarily CONST_INTs.
34808 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
34810 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
34811 -mfpmath=sse is the default also for x86-32 targets with SSE2
34812 instruction set when @option{-ffast-math} is enabled
34814 2017-02-24 Jeff Law <law@redhat.com>
34816 PR rtl-optimizatoin/79286
34817 * ira.c (update_equiv_regs): Drop may_trap_p exception to
34820 2017-02-24 Richard Biener <rguenther@suse.de>
34822 PR tree-optimization/79389
34823 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
34826 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
34828 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
34829 function comment to reflect reality.
34830 (loop_exits_before_overflow): Fix typo in function description.
34832 2017-02-24 Richard Biener <rguenther@suse.de>
34834 PR tree-optimization/79389
34835 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
34836 properly that a threading opportunity exists. Detect conditional
34837 copy/constant propagation opportunities.
34839 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
34841 * config/visium/visium.md (type): Add trap.
34842 (b): New mode attribute.
34843 (*btst): Rename into...
34844 (*btst<mode>): ...this and adjust.
34845 (*cbranchsi4_btst_insn): Rename into...
34846 (*cbranch<mode>4_btst_insn): ...this and adjust.
34847 (trap): New define_insn.
34849 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34851 PR tree-optimization/79389
34852 * ifcvt.c (struct noce_if_info): Add rev_cond field.
34853 (noce_reversed_cond_code): New function.
34854 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
34855 reversed_comparison_code. Formatting fix.
34856 (noce_try_store_flag): Test rev_cond != NULL in addition to
34857 reversed_comparison_code.
34858 (noce_try_store_flag_constants): Likewise.
34859 (noce_try_store_flag_mask): Likewise.
34860 (noce_try_addcc): Use rev_cond if non-NULL instead of
34861 reversed_comparison_code.
34862 (noce_try_cmove_arith): Likewise. Formatting fixes.
34863 (noce_try_minmax, noce_try_abs): Clear rev_cond.
34864 (noce_find_if_block): Initialize rev_cond.
34865 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
34866 instead of false as last argument never attempt to reverse it
34869 2017-02-23 Bin Cheng <bin.cheng@arm.com>
34871 PR tree-optimization/79663
34872 * tree-predcom.c (combine_chains): Process refs in reverse order
34873 only for ZERO length chains, and add explaining comment.
34875 2017-02-23 Jeff Law <law@redhat.com>
34877 PR tree-optimization/79578
34878 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
34879 in call to operand_equal_p.
34881 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
34884 * config/i386/cpuid.h: Fix another undefined behavior.
34886 2017-02-23 Richard Biener <rguenther@suse.de>
34888 PR tree-optimization/79683
34889 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
34890 vector types for data-refs.
34892 2017-02-23 Martin Liska <mliska@suse.cz>
34894 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
34896 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34898 PR middle-end/79665
34899 * internal-fn.c (get_range_pos_neg): Moved to ...
34900 * tree.c (get_range_pos_neg): ... here. No longer static.
34901 * tree.h (get_range_pos_neg): New prototype.
34902 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
34903 are known to be in between 0 and signed maximum inclusive, try to
34904 expand both unsigned and signed divmod and use the cheaper one from
34907 2017-02-22 Jeff Law <law@redhat.com>
34909 PR tree-optimization/79578
34910 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
34911 to compare base operands.
34913 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
34916 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
34917 gpc_reg_operand instead of fpr_reg_operand.
34919 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
34921 * config/mips/mips.c (mips_return_in_memory): Force FP
34922 vector types to be returned in memory for o32 ABI.
34924 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34926 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
34927 instead of DW_TAG_member for static data member declarations and don't
34928 set no_linkage_name for static inline data members.
34929 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
34932 2017-02-22 Martin Liska <mliska@suse.cz>
34934 * doc/invoke.texi: Replace inequality signs with square brackets
34937 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
34940 * lra-constraints.c (simplify_operand_subreg): Handle
34941 WORD_REGISTER_OPERATIONS targets.
34943 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34946 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
34947 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
34948 elimination by swapping fld*.
34950 2017-02-22 Richard Biener <rguenther@suse.de>
34952 PR tree-optimization/79673
34953 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
34954 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
34955 irrelevant address-space qualifiers and avoiding a
34956 ADDR_SPACE_CONVERT_EXPR from fold_convert.
34958 2017-02-22 Richard Biener <rguenther@suse.de>
34960 PR tree-optimization/79666
34961 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
34962 to not symbolically negate if that may introduce undefined
34965 2017-02-22 Martin Liska <mliska@suse.cz>
34968 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
34969 * data-streamer-out.c (streamer_write_gcov_count_stream):
34971 * value-prof.c (stream_out_histogram_value): Make assert more
34972 precise based on type of counter.
34974 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
34977 * config/i386/i386.md (standard_x87sse_constant_load splitter):
34978 Use nonimmediate_operand instead of memory_operand for operand 1.
34979 (float-extend standard_x87sse_constant_load splitter): Ditto.
34981 2017-02-21 Jeff Law <law@redhat.com>
34983 PR tree-optimization/79621
34984 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
34985 blocks with edges to themselves.
34987 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34990 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
34991 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
34992 Use gimple_call_builtin_p.
34995 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
34996 on temporarily removed DEBUG_INSNs.
34998 PR tree-optimization/79649
34999 * tree-loop-distribution.c (classify_partition): Give up on
35000 non-generic address space loads/stores.
35002 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
35004 * doc/loop.texi (Loop manipulation): Remove nonexistent
35005 tree_ssa_loop_version from the documentation.
35006 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
35008 2017-02-21 Jakub Jelinek <jakub@redhat.com>
35011 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
35012 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
35013 * config/rs6000/rs6000.c: Include except.h.
35014 (rs6000_expand_split_stack_prologue): Call
35015 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
35017 2017-02-21 Martin Jambor <mjambor@suse.cz>
35020 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
35021 have been analyzed.
35023 2017-02-21 Martin Jambor <mjambor@suse.cz>
35025 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
35026 for backward compatibility only.
35027 * doc/invoke.texi (Option Summary): Remove all references to
35028 -fipa-cp-alignment.
35030 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
35034 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35036 * lra-constraints.c (curr_insn_transform): Handle
35037 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
35039 2017-02-21 Martin Liska <mliska@suse.cz>
35041 * config/i386/i386.opt: Replace -masm-dialect with -masm.
35043 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
35045 PR translation/79638
35046 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
35048 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
35051 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
35052 (arm_function_ok_for_sibcall): Return false for an indirect call by
35053 descriptor if all the argument registers are used.
35054 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
35055 alignment of the function.
35057 2017-02-21 Jakub Jelinek <jakub@redhat.com>
35059 PR tree-optimization/61441
35060 * simplify-rtx.c (simplify_const_unary_operation): For
35061 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
35062 the sNaN unmodified.
35064 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
35066 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
35067 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
35068 instead of SYSTEM_HEADER_DIR.
35070 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
35071 Martin Liška <mliska@suse.cz>
35073 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
35074 Fix typos and grammar, use active voice, and clarify.
35076 2017-02-20 Marek Polacek <polacek@redhat.com>
35078 PR middle-end/79537
35079 * gimplify.c (gimplify_expr): Handle unused *&&L;.
35082 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
35084 2017-02-20 Jakub Jelinek <jakub@redhat.com>
35087 * config/i386/i386.c (ix86_expand_builtin): Handle
35088 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
35089 ix86_builtins_isa[fcode].isa as a requirement of those
35090 flags and any other flag in the bitmask.
35091 (ix86_init_mmx_sse_builtins): Use 0 instead of
35092 ~OPTION_MASK_ISA_64BIT as mask.
35093 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
35094 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
35095 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
35096 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
35098 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35101 * lra-constraints.c (split_reg): Check requested split mode
35102 is supported by the register.
35104 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35106 * lra-constraints.c (simplify_operand_subreg): Remove early
35109 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35112 * lra-constraints.c (curr_insn_transform): Tighten condition
35113 for converting SUBREG reloads from OP_OUT to OP_INOUT.
35115 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35118 * lra-constraints.c (curr_insn_transform): Handle
35119 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
35121 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
35124 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
35126 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
35128 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
35131 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
35134 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
35136 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
35139 2017-02-18 Jakub Jelinek <jakub@redhat.com>
35141 * final.c (last_columnnum, override_columnnum): New variables.
35142 (final_start_function): Set last_columnnum, pass it to begin_prologue
35143 hook and pass 0 to dwarf2out_begin_prologue.
35144 (final_scan_insn): Update override_columnnum. Pass last_columnnum
35145 to source_line debug hook.
35146 (notice_source_line): Compute last_columnnum and for debug_column_info
35147 return true on column changes.
35148 * debug.h (struct gcc_debug_hooks): Add column argument to
35149 source_line and begin_prologue hooks.
35150 (debug_nothing_int_charstar_int_bool): Remove prototype.
35151 (debug_nothing_int_int_charstar,
35152 debug_nothing_int_int_charstar_int_bool): New prototypes.
35153 (dwarf2out_begin_prologue): Add column argument.
35154 * debug.c (do_nothing_debug_hooks): Adjust source_line and
35155 begin_prologue hooks.
35156 (debug_nothing_int_charstar_int_bool): Remove.
35157 (debug_nothing_int_int_charstar,
35158 debug_nothing_int_int_charstar_int_bool): New functions.
35159 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
35160 through to dwarf2out_source_line.
35161 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
35162 (dwarf2out_source_line): Add column argument, emit it if requested.
35163 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
35165 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
35166 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
35167 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
35168 through to dwarf2out_begin_prologue.
35169 (vmsdbgout_source_line): Add column argument, pass it through to
35170 dwarf2out_source_line.
35171 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
35172 dbxout_source_line caller.
35173 (dbxout_source_line): Add column argument.
35175 * common.opt (gno-column-info, gcolumn-info): New options.
35176 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
35177 (check_die): Also test for multiple DW_AT_decl_column attributes.
35178 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
35179 DW_AT_decl_column if requested.
35180 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
35182 (gen_variable_die): Likewise.
35183 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
35184 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
35187 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
35188 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
35189 (ix86_handle_option): Handle OPT_m3dnowa.
35190 * doc/invoke.texi (-m3dnowa): Document.
35191 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
35192 -m3dnowa instead of -m3dnow -march=athlon.
35195 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
35196 instead of gcc_assert for K, r and R code checks. Formatting fixes.
35198 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35201 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
35202 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
35203 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
35204 generator for vsx_xxpermdi_<mode>_be.
35205 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
35206 force big-endian semantics.
35207 (vsx_xxpermdi_<mode>_be): New define_expand with same
35208 implementation as previous version of vsx_xxpermdi_<mode>.
35210 2017-02-17 Jakub Jelinek <jakub@redhat.com>
35212 PR tree-optimization/79327
35213 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
35214 variable, its initialization and use.
35216 2017-02-17 Julia Koval <julia.koval@intel.com>
35218 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
35219 (OPTION_MASK_ISA_PKU_UNSET): New.
35220 (ix86_handle_option): Handle -mrdpid.
35221 * config/i386/cpuid.h (bit_RDPID): New.
35222 * config/i386/driver-i386.c (host_detect_local_cpu):
35223 Detect RDPID feature.
35224 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
35225 * config/i386/i386-c.c (ix86_target_macros_internal):
35227 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
35228 (ix86_valid_target_attribute_inner_p): Add "rdpid".
35229 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
35230 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
35231 * config/i386/i386.md (define_insn "rdpid"): New.
35232 * config/i386/i386.opt Add -mrdpid.
35233 * config/i386/immintrin.h (_rdpid_u32): New.
35235 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
35237 PR rtl-optimization/79541
35238 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
35239 instead of transforming it into USE.
35241 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
35243 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
35244 If HONOR_SNANS (SFmode) force the input to a register.
35245 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
35246 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
35247 an frsp or similar insn.
35249 2017-02-17 Martin Liska <mliska@suse.cz>
35251 PR rtl-optimization/79577
35252 * params.def (selsched-max-sched-times): Increase minimum to 1.
35254 2017-02-17 Martin Liska <mliska@suse.cz>
35256 PR rtl-optimization/79574
35257 * gcse.c (want_to_gcse_p): Prevent integer overflow.
35259 2017-02-17 Martin Liska <mliska@suse.cz>
35261 PR tree-optimization/79529
35262 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
35263 ssa_defined_default_def_p to handle cases which are implicitly
35265 * tree-ssa.c (ssa_defined_default_def_p): New function.
35266 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
35267 which are implicitly defined.
35268 * tree-ssa.h (ssa_defined_default_def_p): Declare.
35270 2017-02-17 Richard Biener <rguenther@suse.de>
35272 PR middle-end/79576
35273 * params.def (max-ssa-name-query-depth): Limit to 10.
35275 2017-02-17 Richard Biener <rguenther@suse.de>
35277 PR tree-optimization/79552
35278 * tree-ssa-structalias.c (visit_loadstore): Properly verify
35281 2017-02-17 Richard Biener <rguenther@suse.de>
35284 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
35286 2017-02-17 Marek Polacek <polacek@redhat.com>
35288 PR middle-end/79536
35289 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
35290 (fold_negate_expr): New wrapper.
35292 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
35294 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
35295 Correct terminology and de-emphasize pre-standard behavior.
35297 2017-02-16 Alan Modra <amodra@gmail.com>
35299 PR rtl-optimization/79286
35300 * ira.c (def_dominates_uses): New function.
35301 (update_equiv_regs): Don't create an equivalence for insns that
35302 may trap where the register def does not dominate the use.
35304 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
35306 PR rtl-optimization/78127
35307 * lra.c (lra): Call lra_eliminate before finish the loop after
35310 2017-02-16 Richard Biener <rguenther@suse.de>
35312 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
35314 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
35315 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
35316 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
35317 (isl_val_int_from_wi): New function.
35318 (extract_affine_gmp): Rename to ...
35319 (extract_affine_wi): ... this, take a widest_int.
35320 (extract_affine_int): Just wrap extract_affine_wi.
35321 (add_param_constraints): Use isl_val_int_from_wi.
35322 (add_loop_constraints): Likewise, and extract_affine_wi.
35324 2017-02-15 Jeff Law <law@redhat.com>
35326 PR middle-end/79521
35327 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
35328 ira_init_register_move_cost_if_necessary.
35330 2017-02-15 Martin Sebor <msebor@redhat.com>
35332 PR middle-end/32003
35333 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
35334 removed in a prior commit.
35336 2017-02-15 Bin Cheng <bin.cheng@arm.com>
35338 PR tree-optimization/79347
35339 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
35340 counters during peeling.
35342 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
35344 * Makefile.in (site.exp): Remove "set ISLVER".
35346 2017-02-15 Jakub Jelinek <jakub@redhat.com>
35349 * real.c (real_from_integer): Call real_convert even for decimal.
35351 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
35354 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
35356 2017-02-14 Andrew Pinski <apinski@cavium.com>
35358 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
35359 cores and change the partno/implementer to be correct.
35360 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
35361 the 'B" as the implementer.
35362 * config/aarch64/aarch64-tune.md: Regenerate.
35364 2017-02-14 Carl Love <cel@us.ibm.com>
35366 * config/rs6000/rs6000.c: Add case statement entry to make the
35367 xvcvuxdsp built-in argument unsigned.
35368 * config/rs6000/vsx.md: Fix the source and return operand types so they
35369 match the instruction definitions from the ISA document. Fix typo
35370 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
35373 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
35376 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
35377 member early_clobber_alts.
35378 * lra-lives.c (reg_early_clobber_p): New.
35379 (process_bb_lives): Use it.
35380 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
35381 (debug_operand_data): Initialize early_clobber_alts.
35382 (setup_operand_alternative): Set up early_clobber_alts.
35383 (collect_non_operand_hard_regs): Ditto. Pass early clobber
35384 alternatives to new_insn_reg.
35385 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
35387 (lra_update_insn_regno_info): Pass the new arg.
35389 2017-02-14 Jakub Jelinek <jakub@redhat.com>
35391 PR middle-end/79505
35392 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
35393 (new_oacc_loop_raw): Don't clear already cleared fields.
35396 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
35397 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
35398 _mm512_prefetch_i64gather_ps): New inline functions and macros.
35400 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
35403 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
35405 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
35408 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
35409 the extra instruction to the right place to store 128-bit constant
35412 2017-02-14 Martin Sebor <msebor@redhat.com>
35414 PR middle-end/79448
35415 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
35416 warning for strings of unknown length.
35418 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
35420 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
35422 2017-02-14 Jeff Law <law@redhat.com>
35425 * ira-costs.c (scan_one_insn): Initialize register move costs
35426 for pseudos seen in USE/CLOBBER insns.
35428 PR tree-optimization/79095
35429 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
35430 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
35431 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
35432 if the operands are known to be not equal, then the resulting range
35434 (intersect_ranges): If the new range is ~[0,0] and the old range is
35435 wide, then prefer ~[0,0].
35436 * tree-vrp.c (overflow_comparison_p_1): New function.
35437 (overflow_comparison_p): New function.
35438 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
35439 if NAME is used in an overflow test.
35440 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
35441 overflow check that can be expressed as an equality test, then adjust
35442 ops to be that equality test.
35444 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35446 * config/s390/s390-builtin-types.def: Remove flags argument.
35447 * config/s390/s390.c (s390_init_builtins): Likewise.
35449 2017-02-14 Martin Liska <mliska@suse.cz>
35451 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
35452 vector. Fix trailing white spaces.
35454 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
35456 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
35459 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35461 PR rtl-optimization/68664
35462 * config/arm/arm.c (arm_sched_can_speculate_insn):
35463 New function. Declare prototype.
35464 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
35466 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35468 PR rtl-optimization/68664
35469 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
35471 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
35473 2017-02-14 Amit Pawar <amit.pawar@amd.com>
35475 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
35476 max skip bytes for function, loop and jump.
35478 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
35480 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
35481 ABS_EXPR for gimple dump.
35483 2017-02-14 Jakub Jelinek <jakub@redhat.com>
35486 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
35488 PR tree-optimization/79408
35489 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
35490 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
35491 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
35492 also if rhs1 is INTEGER_CST.
35494 2017-02-14 Richard Biener <rguenther@suse.de>
35496 PR middle-end/79432
35497 * tree-into-ssa.c (insert_phi_nodes): When the function can
35498 have abnormal edges rewrite SSA names with broken use-def
35499 dominance out of SSA and register them for PHI insertion.
35501 2017-02-13 Martin Sebor <msebor@redhat.com>
35503 PR middle-end/79496
35504 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
35505 clearing info.nowrite flag when snprintf size argument is a range.
35507 2017-02-13 Jakub Jelinek <jakub@redhat.com>
35509 * cprop.c (cprop_jump): Add missing space in string literal.
35510 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
35511 (get_constraint_for_component_ref): Likewise.
35512 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
35513 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
35514 * lra-constraints.c (process_alt_operands): Likewise.
35515 * ipa-inline.c (inline_small_functions): Likewise.
35516 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
35517 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
35518 * trans-mem.c (diagnose_tm_1_op): Likewise.
35519 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
35520 (grid_parallel_clauses_gridifiable): Likewise.
35522 * config/nvptx/mkoffload.c (process): Add space in between
35525 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
35526 "MOD4_SSE_REGS" and "ALL_REGS".
35528 * spellcheck.c (test_data): Add , in between "foo" and "food".
35530 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35533 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
35534 boundary crossing check and subsequent code generation agree.
35536 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35538 * config/aarch64/aarch64.c (has_memory_op): Delete.
35539 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
35542 2017-02-13 Jakub Jelinek <jakub@redhat.com>
35544 PR rtl-optimization/79388
35545 PR rtl-optimization/79450
35546 * combine.c (distribute_notes): When removing TEM_INSN for which
35547 corresponding dest has last value recorded, invalidate that last
35550 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35552 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
35553 of explicit '@'. Add missing assembly comment marker on branch costs
35556 2017-02-13 Nathan Sidwell <nathan@acm.org>
35558 * gengtype-lex.l (<in_struct>): Add '/'.
35560 2017-02-13 Martin Liska <mliska@suse.cz>
35563 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
35565 2017-02-13 Richard Biener <rguenther@suse.de>
35567 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
35569 * configure: Re-generate.
35570 * config.in: Likewise.
35571 * graphite-dependences.c: Simplify as if
35572 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
35573 * graphite-isl-ast-to-gimple.c: Likewise.
35574 * graphite-optimize-isl.c: Likewise.
35575 * graphite-poly.c: Likewise.
35576 * graphite-sese-to-poly.c: Likewise.
35577 * graphite.h: Likewise.
35578 * toplev.c: Include isl/version.h and use isl_version () for
35579 printing the ISL version.
35580 * doc/install.texi: Update ISL requirement.
35582 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
35584 * doc/standards.texi (Standards): Update reference to
35587 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
35589 * doc/extend.texi (Named Address Spaces): sourceware.org now
35591 * doc/install.texi (Binaries): Ditto.
35594 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
35596 * doc/cpp.texi: Replace "stringify"/"stringification" with C
35597 standard terminology "stringize"/"stringizing" throughout.
35598 * doc/cppinternals.texi: Likewise.
35600 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
35602 * doc/extend.texi: Fix some spelling mistakes and typos.
35603 * doc/invoke.texi: Likewise.
35605 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
35608 * params.def (inline-min-speedup) Change from 10 to 8.
35610 2017-02-11 Jakub Jelinek <jakub@redhat.com>
35612 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
35615 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
35618 * ipa-inline-analysis.c (get_minimal_bb): New function.
35619 (record_modified): Use it.
35620 (remap_edge_change_prob): Handle also ancestor functions.
35622 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
35624 * doc/contrib.texi (Contributors): Remove broken link into
35625 the Mauve CVS repository.
35627 2017-02-11 Jakub Jelinek <jakub@redhat.com>
35629 PR middle-end/79454
35630 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
35631 result computation whenever lhs doesn't have vector mode, not
35632 just when it has BLKmode.
35634 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35636 * doc/makefile.texi (profiledbootstrap): Refer to the
35637 installation instructions only in textual form.
35639 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35642 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
35644 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35646 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
35647 (Specific): Update mingw-w64 reference.
35649 (Specific): Remove broken link to Renesas RX processor.
35651 2017-02-10 Richard Biener <rguenther@suse.de>
35653 * toplev.c (process_options): Do not mention obsolete graphite
35654 options when printing sorry message about missing graphite support.
35655 Mention -floop-nest-optimize.
35657 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
35659 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
35660 (vtst_p16): Likewise.
35661 (vtstq_p8): Likewise.
35662 (vtstq_p16): Likewise.
35664 (vtstq_p64): Likewise.
35665 * config/arm/arm_neon.h (vgetq_lane_p64): New.
35666 (vset_lane_p64): New.
35667 (vsetq_lane_p64): New.
35669 2017-02-10 Jakub Jelinek <jakub@redhat.com>
35671 PR tree-optimization/79411
35672 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
35673 stmt operands are SSA_NAMEs used in abnormal phis.
35674 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
35677 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35680 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
35683 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35685 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
35687 2017-02-09 Jakub Jelinek <jakub@redhat.com>
35689 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
35693 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
35694 not arbitrary TREE_CONSTANT.
35697 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
35698 "omp declare target link" attribute unless is_global_var.
35699 * omp-offload.c (find_link_var_op): Likewise.
35701 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
35702 Chung-Lin Tang <cltang@codesourcery.com>
35704 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
35706 (gimplify_adjust_omp_clauses): Don't delete TILE.
35707 (gimplify_omp_for): Deal with TILE.
35708 * internal-fn.c (expand_GOACC_TILE): New function.
35709 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
35711 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
35712 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
35714 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
35715 avoid DIV for outermost collapse var.
35716 (expand_oacc_for): Insert tile element loop as needed. Adjust.
35717 Remove out of date comments, fix whitespace.
35718 * omp-general.c (omp_extract_for_data): Deal with tiling.
35719 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
35720 adjust OLF_DIM_BASE value.
35721 (struct omp_for_data): Add tiling field.
35722 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
35723 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
35724 for auto loops. Remove default auto determining, moved to
35725 oacc_loop_fixed_partitions.
35726 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
35727 stmts, add e_mask field.
35728 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
35729 (oacc_thread_numbers): Use oacc_dim_call.
35730 (oacc_xform_tile): New.
35731 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
35732 (finish_oacc_loop): Adjust for ifns vector.
35733 (oacc_loop_discover_walk): Append loop abstraction sites to list,
35734 add case for GOACC_TILE fns.
35735 (oacc_loop_xform_loop): Delete.
35736 (oacc_loop_process): Iterate over call list directly, and add
35737 handling for GOACC_TILE fns.
35738 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
35740 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
35741 vector partitioning to outer loops. Assign 2 partitions to loops
35742 when available. Add TILE handling.
35743 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
35744 (execite_oacc_device_lower): Process GOACC_TILE fns,
35745 ignore unknown specs.
35746 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
35747 * tree.c (omp_clause_num_ops): Adjust TILE ops.
35748 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
35750 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
35752 * configure.ac (ACX_BUGURL): Update.
35753 * configure: Regenerate.
35755 2017-02-09 Richard Biener <rguenther@suse.de>
35757 PR tree-optimization/69823
35758 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
35759 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
35761 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35763 * config/arc/arc-c.def: Add __NPS400__ definition.
35764 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
35765 (TARGET_NPS400): Define.
35767 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35769 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
35771 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
35772 pointer, arch_info.
35773 (arc_cpu_types): Fill the arch_info field with a pointer into the
35774 arc_arch_types table.
35775 (arc_selected_cpu): Declare.
35776 * config/arc/arc.c (arc_selected_cpu): Make global.
35777 (arc_selected_arch): Delete.
35778 (arc_base_cpu): Delete.
35779 (arc_override_options): Remove references to deleted variables,
35780 update access to arch information.
35781 (ARC_OPT): Update access to arch information.
35782 (ARC_OPTX): Likewise.
35783 * config/arc/arc.h (arc_base_cpu): Remove declaration.
35784 (TARGET_ARC600): Update access to arch information.
35785 (TARGET_ARC601): Likewise.
35786 (TARGET_ARC700): Likewise.
35787 (TARGET_EM): Likewise.
35788 (TARGET_HS): Likewise.
35789 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
35792 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
35795 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
35796 condition/operands for integer GE/LE/GEU/LEU operations.
35798 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
35800 PR translation/79397
35801 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
35804 2017-02-08 Martin Jambor <mjambor@suse.cz>
35807 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
35808 whether allocation happened.
35809 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
35810 nothing was allocated.
35812 2017-02-08 Jakub Jelinek <jakub@redhat.com>
35814 PR tree-optimization/79408
35815 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
35816 constant, but SSA_NAME with a known integer range, use the minimum
35817 of that range instead of op1 to determine if modulo can be replaced
35818 with its first operand.
35820 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35822 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
35824 2017-02-08 Richard Biener <rguenther@suse.de>
35826 PR tree-optimization/71824
35827 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35828 Check all loops contained in the merged region.
35830 2017-02-07 Andrew Pinski <apinski@cavium.com>
35832 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
35834 2017-02-07 Andrew Pinski <apinski@cavium.com>
35836 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
35837 (thunderxt88): Likewise.
35838 (thunderxt81): Disable LSE and change v8.1 to v8.
35839 (thunderxt83): Likewise.
35841 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35842 Richard Biener <rguenther@suse.de>
35844 PR middle-end/79399
35845 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
35846 type from int to size_t.
35847 * ira-costs.c (struct_costs_size): Change type from int to size_t.
35849 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35851 PR rtl-optimization/79386
35852 * cprop.c (bypass_conditional_jumps): Initialize
35853 bypass_last_basic_block already before splitting bbs after
35854 unconditional traps...
35855 (bypass_conditional_jumps): ... rather than here.
35858 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
35859 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
35860 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
35861 fix -masm=intel patterns.
35863 2017-02-07 Richard Biener <rguenther@suse.de>
35865 PR tree-optimization/79256
35866 PR middle-end/79278
35867 * builtins.c (get_object_alignment_2): Use min_align_of_type
35868 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
35869 and ADJUST_FIELD_ALIGN.
35871 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
35873 * doc/tm.texi: Regenerate.
35874 * stor-layout.c (layout_decl): Adjust.
35875 (update_alignment_for_field): Likewise.
35876 (place_field): Likewise.
35877 (min_align_of_type): Likewise.
35878 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
35879 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
35880 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
35881 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
35882 * config/frv/frv.c (frv_adjust_field_align): Likewise.
35883 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
35884 * config/i386/i386.c (x86_field_alignment): Likewise.
35885 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
35886 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
35887 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
35888 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
35889 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
35890 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
35894 2017-01-30 Richard Biener <rguenther@suse.de>
35896 PR tree-optimization/79256
35897 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35898 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35901 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
35903 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
35904 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
35905 builtins to SImode and emit a zero-extend, if necessary.
35907 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35909 * docs/invoke.texi (RISC-V Options): Alphabetize.
35911 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35913 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
35916 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35918 * config/riscv/riscv.c: New file.
35919 * common/config/riscv/riscv-common.c: Likewise.
35920 * config.gcc: Likewise.
35921 * config/riscv/constraints.md: Likewise.
35922 * config/riscv/elf.h: Likewise.
35923 * config/riscv/generic.md: Likewise.
35924 * config/riscv/linux.h: Likewise.
35925 * config/riscv/multilib-generator: Likewise.
35926 * config/riscv/peephole.md: Likewise.
35927 * config/riscv/pic.md: Likewise.
35928 * config/riscv/predicates.md: Likewise.
35929 * config/riscv/riscv-builtins.c: Likewise.
35930 * config/riscv/riscv-c.c: Likewise.
35931 * config/riscv/riscv-ftypes.def: Likewise.
35932 * config/riscv/riscv-modes.def: Likewise.
35933 * config/riscv/riscv-opts.h: Likewise.
35934 * config/riscv/riscv-protos.h: Likewise.
35935 * config/riscv/riscv.h: Likewise.
35936 * config/riscv/riscv.md: Likewise.
35937 * config/riscv/riscv.opt: Likewise.
35938 * config/riscv/sync.md: Likewise.
35939 * config/riscv/t-elf-multilib: Likewise.
35940 * config/riscv/t-linux: Likewise.
35941 * config/riscv/t-linux-multilib: Likewise.
35942 * config/riscv/t-riscv: Likewise.
35943 * configure.ac: Likewise.
35944 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
35945 Waterman as RISC-V maintainers.
35946 * doc/install.texi: Add RISC-V entries.
35947 * doc/invoke.texi: Add RISC-V options section.
35948 * doc/md.texi: Add RISC-V constraints section.
35949 * configure: Regenerated.
35951 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
35954 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
35955 false values to be constant vectors with all 0 or all 1 bits set.
35956 (vcondu<mode><mode>): Likewise.
35957 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
35959 (fpmask_comparison_operator): Update comment.
35960 (vecint_comparison_operator): New predicate.
35961 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
35962 vector conditionals when the true and false values are constant
35963 vectors with all 0 bits or all 1 bits set.
35965 2017-02-06 Martin Sebor <msebor@redhat.com>
35967 PR tree-optimization/79376
35968 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
35970 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
35972 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
35973 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
35974 to simplify split condition.
35976 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35978 * omp-expand.c (oxpand_omp_atomic_fetch_op,
35979 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
35982 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
35984 PR rtl-optimization/68664
35985 * target.def (can_speculate_insn): New hook.
35986 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
35987 * doc/tm.texi: Regenerate.
35988 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
35989 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
35990 (rs6000_sched_can_speculate_insn): New function.
35992 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35994 PR tree-optimization/79284
35995 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
35996 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
35997 vectorizable_mask_load_store, vectorizable_operation,
35998 vect_is_simple_cond, get_same_sized_vectype): Use it instead
35999 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
36000 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
36001 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
36002 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
36003 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
36004 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
36005 is_gimple_assign (stmt). Replace another such test with
36006 is_gimple_assign (stmt).
36008 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
36011 * config/avr/avr.c (rtl-iter.h): Include it.
36012 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
36013 (avr_legitimate_combined_insn): ...and implementation.
36015 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
36017 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
36018 * config/s390/s390.c (s390_const_operand_ok)
36019 (s390_canonicalize_comparison, s390_extract_part)
36020 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
36021 (s390_contiguous_bitmask_p, s390_rtx_costs)
36022 (legitimize_pic_address): Likewise.
36023 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
36024 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
36025 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
36026 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
36027 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
36029 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
36031 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
36032 REGNO($0) == REGNO($1).
36034 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36036 * config/s390/linux.h(SIZE_TYPE): Add comment.
36038 2017-02-06 Julian Brown <julian@codesourcery.com>
36039 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
36040 Virendra Pathak <virendra.pathak@broadcom.com>
36042 * config/aarch64/aarch64-cores.def: Change the scheduler
36044 * config/aarch64/aarch64.md: Include thunderx2t99.md.
36045 * config/aarch64/thunderx2t99.md: New file.
36047 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36049 * doc/standards.texi (Go Language): Update link to language
36052 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
36054 * tree-eh.c (lower_resx): Sanitize profile.
36055 (cleanup_empty_eh_move_lp): Likewise.
36057 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
36060 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
36062 * cfgloopmanip.h (loop_version): Update prototype.
36063 * modulo-sched.c (sms_schedule): Update call of loop_version.
36064 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
36065 * tree-parloops.c (gen_parallel_loop): Likewise.
36066 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
36067 * tree-ssa-loop-split.c (split_loop): Likewise.
36068 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
36069 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
36071 2017-02-05 Martin Liska <mliska@suse.cz>
36074 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
36076 (print_operand_address): Initialize a struct to zero.
36078 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36080 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
36081 garbage collector only in textual form.
36083 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36085 * doc/extend.texi (x86 specific memory model extensions for
36086 transactional memory): Simplify a phrase.
36088 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
36091 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
36092 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
36093 (atomic_storedi_1): Likewise.
36095 2017-02-04 Jakub Jelinek <jakub@redhat.com>
36097 PR tree-optimization/79338
36098 * tree-parloops.c (gather_scalar_reductions): Don't call
36099 vect_analyze_loop_form for loop->inner before destroying loop's
36102 2017-02-03 Martin Sebor <msebor@redhat.com>
36104 PR tree-optimization/79327
36105 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
36106 when precision has resulted in leading zeros.
36107 (format_integer): Adjust the likely counter to assume an unknown
36108 argument that may be zero is non-zero.
36110 2017-02-03 Jason Merrill <jason@redhat.com>
36113 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
36114 avoid copying non-taken branch.
36116 2017-02-03 Jakub Jelinek <jakub@redhat.com>
36118 PR tree-optimization/79340
36119 * tree-vect-loop.c (vectorizable_reduction): Release
36120 vec_defs elements after safe_splicing them into other vectors.
36123 PR tree-optimization/79327
36124 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
36125 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
36127 (format_integer): Use wide_int_to_tree instead of build_int_cst
36128 + to_?hwi. If argmin is NULL, just set argmin and argmax to
36129 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
36130 of shortest and longest sequence.
36132 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
36134 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
36135 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
36137 2017-02-03 Walter Lee <walt@tilera.com>
36140 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
36141 after initial stackframe link reg save.
36142 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
36144 2017-02-03 Jakub Jelinek <jakub@redhat.com>
36147 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
36148 wu for stxssp alternative.
36150 2017-02-03 Martin Sebor <msebor@redhat.com>
36152 PR tree-optimization/79352
36153 * gimple-fold.c (get_range_strlen): Add argument.
36154 (get_range_strlen): Change return type to bool.
36155 (get_maxval_strlen): Pass in a dummy argument.
36156 * gimple-fold.h (get_range_strlen): Change return type to bool.
36157 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
36158 * tree.h (array_at_struct_end_p): Add argument.
36159 * tree.c (array_at_struct_end_p): Handle it.
36161 2017-02-03 Martin Liska <mliska@suse.cz>
36164 * multiple_target.c (create_dispatcher_calls): Redirect edge
36165 from a caller of a dispatcher.
36166 (expand_target_clones): Make the clones local.
36167 (ipa_target_clone): Do both target clones and resolvers.
36168 (ipa_dispatcher_calls): Remove the pass.
36169 (pass_dispatcher_calls::gate): Likewise.
36170 (make_pass_dispatcher_calls): Likewise.
36171 * passes.def (pass_target_clone): Put as very first IPA early
36174 2017-02-03 Martin Liska <mliska@suse.cz>
36176 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
36177 in case of a function with ifunc attribute.
36179 2017-02-03 Martin Liska <mliska@suse.cz>
36181 * cgraph.c (cgraph_node::dump): Dump function version info.
36182 * symtab.c (symtab_node::dump_base): Add missing new line.
36184 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
36186 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
36187 (ifcombine_ifandif): Use it.
36189 2017-02-03 Martin Liska <mliska@suse.cz>
36191 * doc/invoke.texi: Document default value for
36192 use-after-scope-direct-emission-threshold.
36194 2017-02-03 Martin Liska <mliska@suse.cz>
36196 PR tree-optimization/79339
36197 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
36198 (format_floating): Likewise.
36200 2017-02-03 Martin Liska <mliska@suse.cz>
36203 * ipa-prop.c (ipa_node_params_t::insert): Remove current
36205 (ipa_node_params_t::remove): Likewise.
36206 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
36207 initialization from removed ipa_node_params_t::insert.
36208 (ipa_node_params::~ipa_node_params): Move from removed
36209 ipa_node_params_t::release.
36210 * symbol-summary.h (symbol_summary::m_released): New member.
36211 Do not release a summary twice. Do not allow to call finalizer
36212 for types of a summary that live in GGC memory.
36214 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
36216 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
36219 2017-02-02 Martin Sebor <msebor@redhat.com>
36221 PR middle-end/79275
36222 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
36223 (format_string): Tighten up the range of output for non-constant
36224 strings and correct the expected range for wide non-constant strings.
36226 2017-02-02 Martin Sebor <msebor@redhat.com>
36228 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
36230 PR middle-end/32003
36231 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
36233 (-fdump-tree-@var): Add to index and document how to come up
36234 with pass-specific option and dump file names.
36235 (-fdump-passes): Clarify where to look for output.
36237 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
36239 PR middle-end/77445
36240 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
36241 statistics of the analyzed path; allow threading for speed when
36242 any of BBs along the path are optimized for speed.
36244 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
36246 PR middle-end/78468
36247 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
36248 settings of the virtual registers.
36251 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
36253 * explow.c (get_dynamic_stack_size): Take known alignment of stack
36254 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
36257 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36259 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
36260 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
36262 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36264 * config/s390/s390.md: Add missing comments with the expanded
36266 * config/s390/vector.md: Likewise.
36267 * config/s390/vx-builtins.md: Likewise.
36269 2017-02-02 Jakub Jelinek <jakub@redhat.com>
36272 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
36273 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
36274 conditions on a single line.
36276 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36278 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
36279 __S390_VX__ to __VX__.
36281 2017-02-01 Andrew Pinski <apinski@cavium.com>
36283 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
36284 stmt_info to record_stmt_cost.
36285 (vect_get_known_peeling_cost): Pass stmt_info if known to
36287 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
36288 cpu_vector_cost field into
36289 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
36290 field into vec_int_stmt_cost and vec_fp_stmt_cost.
36291 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
36292 splitting of scalar_stmt_cost and vec_stmt_cost.
36293 (thunderx_vector_cost): Likewise.
36294 (cortexa57_vector_cost): LIkewise.
36295 (exynosm1_vector_cost): Likewise.
36296 (xgene1_vector_cost): Likewise.
36297 (thunderx2t99_vector_cost): Improve after the splitting of the two
36299 (aarch64_builtin_vectorization_cost): Update for the splitting of
36300 scalar_stmt_cost and vec_stmt_cost.
36302 2017-02-01 Torvald Riegel <triegel@redhat.com>
36303 Richard Henderson <rth@redhat.com>
36305 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
36306 conditional on existance of a fast atomic load.
36307 * optabs-query.c (can_atomic_load_p): New function.
36308 * optabs-query.h (can_atomic_load_p): Declare it.
36309 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
36310 no fast atomic load is available for the particular size of access.
36311 (expand_atomic_compare_and_swap): Likewise.
36312 (expand_atomic_load): Likewise.
36313 (expand_atomic_store): Likewise.
36314 (expand_atomic_fetch_op): Likewise.
36315 * testsuite/lib/target-supports.exp
36316 (check_effective_target_sync_int_128): Remove x86 because it provides
36317 no fast atomic load.
36318 (check_effective_target_sync_int_128_runtime): Likewise.
36320 2017-02-01 Richard Biener <rguenther@suse.de>
36322 * graphite.c: Include tree-vectorizer.h for find_loop_location.
36323 (graphite_transform_loops): Provide opt-info for optimized nests.
36324 * tree-parloop.c (parallelize_loops): Provide opt-info for
36325 parallelized loops.
36327 2017-02-01 Richard Biener <rguenther@suse.de>
36329 PR middle-end/79315
36330 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
36331 was not set before.
36333 2017-02-01 Richard Biener <rguenther@suse.de>
36335 PR tree-optimization/71824
36336 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
36337 Verify the loops are valid in the merged SESE region.
36338 (scop_detection::can_represent_loop_1): Check analyzing the
36339 evolution of the number of iterations in the region succeeds.
36341 2017-01-31 Ian Lance Taylor <iant@golang.org>
36343 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
36344 REG_ARGS_SIZE note to 32-bit push insns and call insn.
36346 2017-01-31 David Malcolm <dmalcolm@redhat.com>
36348 PR preprocessor/79210
36349 * input.c (get_substring_ranges_for_loc): Replace line_width
36350 assertion with error-handling.
36352 2017-01-31 Richard Biener <rguenther@suse.de>
36354 PR tree-optimization/77318
36355 * graphite-sese-to-poly.c (extract_affine): Fix assert.
36356 (create_pw_aff_from_tree): Take loop parameter.
36357 (add_condition_to_pbb): Pass loop of the condition to
36358 create_pw_aff_from_tree.
36360 2017-01-31 Jakub Jelinek <jakub@redhat.com>
36362 * config/s390/s390.c (s390_asan_shadow_offset): New function.
36363 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
36365 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
36369 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
36371 (convert_int_to_float128): Likewise.
36372 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
36373 (convert_int_to_float128): Likewise.
36374 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
36375 (UNSPEC_IEEE128_CONVERT): Likewise.
36376 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
36377 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
36378 Use local variables for IBM extended format.
36379 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
36380 (fix_trunc<mode>si2_fprs): Likewise.
36381 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
36382 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
36383 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
36384 to know that we can now have integers of all sizes in vector
36386 (fix<uns>_<mode>di2_hw): Likewise.
36387 (float<uns>_<mode>si2_hw): Likewise.
36388 (fix_<mode>si2_hw): Likewise.
36389 (fixuns_<mode>si2_hw): Likewise.
36390 (float<uns>_<mode>di2_hw): Likewise.
36391 (float_<mode>di2_hw): Likewise.
36392 (float_<mode>si2_hw): Likewise.
36393 (floatuns_<mode>di2_hw): Likewise.
36394 (floatuns_<mode>si2_hw): Likewise.
36395 (xscvqp<su>wz_<mode>): Delete, no longer used.
36396 (xscvqp<su>dz_<mode>): Likewise.
36397 (xscv<su>dqp_<mode>): Likewise.
36398 (ieee128_mfvsrd_64bit): Likewise.
36399 (ieee128_mfvsrd_32bit): Likewise.
36400 (ieee128_mfvsrwz): Likewise.
36401 (ieee128_mtvsrw): Likewise.
36402 (ieee128_mtvsrd_64bit): Likewise.
36403 (ieee128_mtvsrd_32bit): Likewise.
36405 2017-01-31 Martin Liska <mliska@suse.cz>
36408 * ipa-prop.c (ipa_free_all_node_params): Call release method
36409 instead of ~sumbol_summary to not to trigger double times
36412 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
36414 PR tree-optimization/71691
36415 * bitmap.h (class auto_bitmap): New.
36416 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
36417 is_maybe_undefined instead of ssa_undefined_value_p.
36419 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36421 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
36422 __S390_ARCH_LEVEL__ to __ARCH__.
36424 2017-01-31 Jakub Jelinek <jakub@redhat.com>
36426 PR tree-optimization/79267
36427 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
36428 if should_remove_lhs_p is true.
36430 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
36433 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
36434 (add_alignment_attribute): New.
36435 (base_type_die): Add alignment attribute.
36436 (subrange_type_die): Likewise.
36437 (modified_type_die): Likewise.
36438 (gen_array_type_die): Likewise.
36439 (gen_descr_array_type_die: Likewise.
36440 (gen_enumeration_type_die): Likewise.
36441 (gen_subprogram_die): Likewise.
36442 (gen_variable_die): Likewise.
36443 (gen_field_die): Likewise.
36444 (gen_ptr_to_mbr_type_die): Likewise.
36445 (gen_struct_or_union_type_die): Likewise.
36446 (gen_subroutine_type_die): Likewise.
36447 (gen_typedef_die): Likewise.
36448 (base_type_cmp): Compare alignment attribute.
36450 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36453 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
36454 (setb_unsigned) New pattern for setb with CCUNS.
36455 * config/rs6000/rs6000.c (expand_block_compare): Use a different
36456 subfc./subfe sequence to avoid overflow problems. Generate a
36457 shorter sequence with cmpld/setb for power9.
36458 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
36459 for generating subfc. instruction.
36460 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
36461 now uses this instruction.
36463 2017-01-30 Ian Lance Taylor <iant@google.com>
36466 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
36467 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
36469 2017-01-30 Martin Sebor <msebor@redhat.com>
36471 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
36472 Move constant to the right of a relational operator.
36473 (get_mpfr_format_length, format_character, format_string): Ditto.
36474 (should_warn_p, maybe_warn): Same.
36476 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
36478 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
36481 * asan.c (get_translation_unit_decl): Remove function.
36482 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
36484 2017-01-30 Martin Liska <mliska@suse.cz>
36486 PR gcov-profile/79259
36487 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
36488 -fprofile-generate.
36490 2017-01-30 Martin Liska <mliska@suse.cz>
36493 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
36494 Initialize variables with NULL value.
36496 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
36499 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
36501 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
36503 2017-01-30 Richard Biener <rguenther@suse.de>
36505 PR tree-optimization/79276
36506 * tree-vrp.c (process_assert_insertions): Properly adjust common
36507 when removing a duplicate.
36509 2017-01-30 Richard Biener <rguenther@suse.de>
36511 PR tree-optimization/79256
36512 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
36513 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
36515 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
36517 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
36520 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
36521 ("*r<noxa>sbg_<mode>_sll_bitmask")
36522 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
36523 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
36524 Use contiguous_bitmask_nowrap_operand.
36526 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36529 * config/rs6000/altivec.h (vec_xl): Revise #define.
36530 (vec_xst): Likewise.
36532 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
36534 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
36536 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
36538 PR rtl-optimization/79194
36539 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
36540 traps before call to bypass_conditional_jumps.
36542 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
36544 PR tree-optimization/71374
36545 * lra-constraints.c (check_conflict_input_operands): New.
36546 (match_reload): Use it.
36548 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
36551 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
36552 account to calculate conflict_set.
36554 2017-01-27 Bin Cheng <bin.cheng@arm.com>
36556 PR rtl-optimization/78559
36557 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
36558 other_insn in combine.
36560 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
36562 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
36563 uint16_type_node for BT_UINT16.
36565 2017-01-27 David Malcolm <dmalcolm@redhat.com>
36567 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
36568 "RTL Tests" to menu.
36569 (GIMPLE Tests): New node.
36570 (RTL Tests): New node.
36572 2017-01-27 Richard Biener <rguenther@suse.de>
36574 PR tree-optimization/79245
36575 * tree-loop-distribution.c (distribute_loop): Apply cost
36576 modeling also to detected patterns.
36578 2017-01-27 Richard Biener <rguenther@suse.de>
36580 PR tree-optimization/71433
36581 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
36582 (compare_assert_loc): New function.
36583 (process_assert_insertions): Sort and optimize assert locations
36584 to remove duplicates and push down identical assertions on
36585 edges to their destination block.
36587 2017-01-27 Richard Biener <rguenther@suse.de>
36589 PR tree-optimization/79244
36590 * tree-vrp.c (remove_range_assertions): Forcefully propagate
36591 out SSA names even if abnormal.
36593 2017-01-27 Jakub Jelinek <jakub@redhat.com>
36595 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
36596 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
36597 instead of MPFR_RNDN.
36599 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
36602 * arm.c (arm_option_override): Don't call build_target_option_node
36603 until after doing all option overrides.
36604 (arm_valid_target_attribute_tree): Likewise.
36606 2017-01-27 Martin Liska <mliska@suse.cz>
36608 * doc/invoke.texi (-fprofile-arcs): Document profiling support
36609 for {cd}tors and C++ {cd}tors.
36611 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
36613 * config/s390/s390.md ("*setmem_long_and")
36614 ("*setmem_long_and_31z"): Use zero_extend instead of and.
36616 2017-01-26 Martin Sebor <msebor@redhat.com>
36618 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
36621 2017-01-26 Martin Sebor <msebor@redhat.com>
36623 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
36624 HAVE_DFmode before using XFmode or DFmode.
36625 (parse_directive): Avoid using the z length modifier to avoid
36626 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
36628 PR middle-end/78703
36629 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
36630 to accept adjustment as an array.
36631 (get_int_range): New function.
36632 (struct directive): Make width and prec arrays.
36633 (directive::set_width, directive::set_precision): Call get_int_range.
36634 (format_integer, format_floating): Handle width and precision ranges.
36635 (format_string, parse_directive): Same.
36637 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36640 * dwarf2out.c (generate_skeleton_bottom_up): For children with
36641 comdat_type_p set, just clone them, but keep the children in the
36645 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
36646 which have direct callers with -fvar-tracking-assignments enabled
36648 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
36649 inside of type units.
36651 2017-01-26 Martin Sebor <msebor@redhat.com>
36653 PR middle-end/78703
36654 * gimple-ssa-sprintf.c (struct result_range): Add likely and
36656 (struct format_result): Replace number_chars, number_chars_min,
36657 and number_chars_max with a single member of struct result_range.
36659 (format_result::operator+=): Adjust.
36660 (struct fmtresult): Remove bounded. Handle likely and unlikely
36662 (fmtresult::adjust_for_width_or_precision): New function.
36663 (fmtresult:type_max_digits): New function.
36664 (bytes_remaining): Handle likely and unlikely counters.
36665 (min_bytes_remaining): Remove.
36666 (format_percent): Simplify.
36667 (format_integer, format_floating): Set likely and unlikely counters.
36668 (get_string_length, format_character, format_string): Same.
36669 (format_plain, should_warn_p): New function.
36670 (maybe_warn): Call should_warn_p. Update diagnostic messages
36671 and handle those for all directives, including plain strings.
36672 (format_directive): Handle likely and unlikely counters.
36673 Remove unnecessary quoting from diagnostics. Add an informational
36675 (add_bytes): Remove.
36676 (pass_sprintf_length::compute_format_length): Simplify.
36677 (try_substitute_return_value): Handle likely and unlikely counters.
36679 2017-01-26 Carl Love <cel@us.ibm.com>
36681 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
36682 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
36684 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
36687 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
36688 endianess for subregs into account.
36689 * lra-constraints.c (lra_constraints): Do risky transformations
36690 always on the first iteration.
36691 * lra-lives.c (check_pseudos_live_through_calls): Add arg
36692 last_call_used_reg_set.
36693 (process_bb_lives): Define and use last_call_used_reg_set.
36694 * lra.c (lra): Always continue after lra_constraints on the first
36697 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
36699 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
36701 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
36703 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36705 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
36706 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
36707 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
36708 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
36709 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
36710 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
36711 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
36712 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
36713 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
36715 2017-01-26 Marek Polacek <polacek@redhat.com>
36718 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
36719 for the third operand.
36721 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36723 PR middle-end/79236
36724 * omp-low.c (struct omp_context): Add simt_stmt field.
36725 (scan_omp_for): Return omp_context *.
36726 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
36727 context to the _simt_ SIMD stmt.
36728 (lower_omp_for): For combined SIMD with sibling _simt_
36729 SIMD, make sure to use the same decls in _looptemp_
36730 clauses as in the sibling.
36732 2017-01-26 David Sherwood <david.sherwood@arm.com>
36734 PR middle-end/79212
36735 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
36738 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36741 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
36742 emit fld b; fld a; if possible.
36744 * brig-builtins.def: Update copyright years.
36745 * config/arm/arm_acle_builtins.def: Update copyright years.
36747 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
36750 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
36751 constraint instead of o for the stxsd instruction.
36753 2017-01-25 Carl Love <cel@us.ibm.com>
36755 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
36756 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
36758 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
36760 * doc/invoke.texi (C++ Dialect Options): Fix typo.
36762 2017-01-25 Richard Biener <rguenther@suse.de>
36764 PR tree-optimization/69264
36765 * target.def (vector_alignment_reachable): Improve documentation.
36766 * doc/tm.texi: Regenerate.
36767 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
36769 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
36770 earlier changes with respect to TYPE_USER_ALIGN.
36771 (vector_alignment_reachable_p): Likewise. Improve dumping.
36773 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36776 * config/arm/arm.md (xordi3): Force constant operand into a register
36779 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36781 * doc/invoke.texi (-fstore-merging): Correct default optimization
36782 levels at which it is enabled.
36783 (-O): Move -fstore-merging from list to...
36786 2017-01-25 Richard Biener <rguenther@suse.de>
36789 * omp-expand.c: Include debug.h.
36790 (expand_omp_taskreg): Make sure to generate early debug before
36791 outlining anything from a function.
36792 (expand_omp_target): Likewise.
36793 (grid_expand_target_grid_body): Likewise.
36795 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
36798 * asan.c (get_translation_unit_decl): New function.
36799 (asan_add_global): Extract modules file name from globals
36800 TRANSLATION_UNIT_DECL name.
36802 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
36805 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
36806 for long calls with APCS frame and VFP.
36808 2017-01-24 David Malcolm <dmalcolm@redhat.com>
36810 * cfg.c (original_copy_tables_initialized_p): New function.
36811 * cfg.h (original_copy_tables_initialized_p): New decl.
36812 * cfgrtl.c (relink_block_chain): Guard the call to
36813 free_original_copy_tables with a call to
36814 original_copy_tables_initialized_p.
36815 * cgraph.h (symtab_node::native_rtl_p): New decl.
36816 * cgraphunit.c (symtab_node::native_rtl_p): New function.
36817 (symtab_node::needed_p): Don't assert for early assembly output
36818 for __RTL functions.
36819 (cgraph_node::finalize_function): Set "force_output" for __RTL
36821 (cgraph_node::analyze): Bail out early for __RTL functions.
36822 (analyze_functions): Update assertion to support __RTL functions.
36823 (cgraph_node::expand): Bail out early for __RTL functions.
36824 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
36826 * function.h (struct function): Update comment for field
36828 * gimple-expr.c: Include "tree-pass.h".
36829 (gimple_has_body_p): Return false for __RTL functions.
36830 * Makefile.in (OBJS): Add run-rtl-passes.o.
36831 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
36833 (gcc::pass_manager::get_clean_slate): New accessor.
36834 * passes.c: Include "insn-addr.h".
36835 (should_skip_pass_p): Add logging. Update logic for running
36836 "expand" to be compatible with both __GIMPLE and __RTL. Guard
36837 property-provider override so it is only done for gimple passes.
36839 (skip_pass): New function.
36840 (execute_one_pass): Call skip_pass when skipping passes.
36841 * read-md.c (md_reader::read_char): Support filtering
36842 the input to a subset of line numbers.
36843 (md_reader::md_reader): Initialize fields
36844 m_first_line and m_last_line.
36845 (md_reader::read_file_fragment): New function.
36846 * read-md.h (md_reader::read_file_fragment): New decl.
36847 (md_reader::m_first_line): New field.
36848 (md_reader::m_last_line): New field.
36849 * read-rtl-function.c (function_reader::create_function): Only
36850 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
36851 curr_properties. Set DECL_INITIAL to a dummy block.
36852 (read_rtl_function_body_from_file_range): New function.
36853 * read-rtl-function.h (read_rtl_function_body_from_file_range):
36855 * run-rtl-passes.c: New file.
36856 * run-rtl-passes.h: New file.
36858 2017-01-24 Jeff Law <law@redhat.com>
36860 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
36863 2017-01-24 Bin Cheng <bin.cheng@arm.com>
36865 PR tree-optimization/79159
36866 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
36867 (record_nonwrapping_iv): Improve boundary using above function if no
36868 value range information.
36870 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
36871 Martin Jambor <mjambor@suse.cz>
36873 * brig-builtins.def: New file.
36874 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
36875 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
36876 (DEF_HSAIL_SAT_BUILTIN): Likewise.
36877 (DEF_HSAIL_INTR_BUILTIN): Likewise.
36878 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
36879 * builtin-types.def (BT_INT8): New.
36880 (BT_INT16): Likewise.
36881 (BT_UINT8): Likewise.
36882 (BT_UINT16): Likewise.
36883 (BT_FN_ULONG): Likewise.
36884 (BT_FN_UINT_INT): Likewise.
36885 (BT_FN_UINT_ULONG): Likewise.
36886 (BT_FN_UINT_LONG): Likewise.
36887 (BT_FN_UINT_PTR): Likewise.
36888 (BT_FN_ULONG_PTR): Likewise.
36889 (BT_FN_INT8_FLOAT): Likewise.
36890 (BT_FN_INT16_FLOAT): Likewise.
36891 (BT_FN_UINT32_FLOAT): Likewise.
36892 (BT_FN_UINT16_FLOAT): Likewise.
36893 (BT_FN_UINT8_FLOAT): Likewise.
36894 (BT_FN_UINT64_FLOAT): Likewise.
36895 (BT_FN_UINT16_UINT32): Likewise.
36896 (BT_FN_UINT32_UINT16): Likewise.
36897 (BT_FN_UINT16_UINT16_UINT16): Likewise.
36898 (BT_FN_INT_PTR_INT): Likewise.
36899 (BT_FN_UINT_PTR_UINT): Likewise.
36900 (BT_FN_LONG_PTR_LONG): Likewise.
36901 (BT_FN_ULONG_PTR_ULONG): Likewise.
36902 (BT_FN_VOID_UINT64_UINT64): Likewise.
36903 (BT_FN_UINT8_UINT8_UINT8): Likewise.
36904 (BT_FN_INT8_INT8_INT8): Likewise.
36905 (BT_FN_INT16_INT16_INT16): Likewise.
36906 (BT_FN_INT_INT_INT): Likewise.
36907 (BT_FN_UINT_FLOAT_UINT): Likewise.
36908 (BT_FN_FLOAT_UINT_UINT): Likewise.
36909 (BT_FN_ULONG_UINT_UINT): Likewise.
36910 (BT_FN_ULONG_UINT_PTR): Likewise.
36911 (BT_FN_ULONG_ULONG_ULONG): Likewise.
36912 (BT_FN_UINT_UINT_UINT): Likewise.
36913 (BT_FN_VOID_UINT_PTR): Likewise.
36914 (BT_FN_UINT_UINT_PTR: Likewise.
36915 (BT_FN_UINT32_UINT64_PTR): Likewise.
36916 (BT_FN_INT_INT_UINT_UINT): Likewise.
36917 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
36918 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
36919 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
36920 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
36921 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
36922 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
36923 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
36924 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
36925 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
36926 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
36927 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
36928 * doc/frontends.texi: List BRIG FE.
36929 * doc/install.texi (Testing): Add BRIG tesring requirements.
36930 * doc/invoke.texi (Overall Options): Mention BRIG.
36931 * doc/standards.texi (Standards): Doucment BRIG HSA version.
36933 2017-01-24 Richard Biener <rguenther@suse.de>
36935 PR translation/79208
36936 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
36938 2017-01-24 Martin Jambor <mjambor@suse.cz>
36941 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
36942 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
36943 and known_contexts.
36945 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
36947 PR middle-end/79123
36948 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
36949 casts from signed to unsigned really don't have a range.
36951 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
36953 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
36954 GMP_RNDx for compatiblity.
36956 2017-01-24 Martin Liska <mliska@suse.cz>
36959 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
36960 that would prevent us to call alloca with -1 as argument.
36962 2017-01-24 Jakub Jelinek <jakub@redhat.com>
36964 * dwarf2out.c (output_compilation_unit_header, output_file_names):
36965 Avoid -Wformat-security warning.
36967 2017-01-23 Andrew Pinski <apinski@cavium.com>
36969 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
36972 2017-01-23 Martin Sebor <msebor@redhat.com>
36974 PR middle-end/78703
36975 * gimple-ssa-sprintf.c (warn_level): New global.
36976 (format_integer): Use it here and throughout the rest of the file.
36977 Use the same switch to compute sign as base.
36978 (maybe_warn): New function.
36979 (format_directive): Factor out warnings into maybe_warn.
36980 Add debugging output. Use warn_level.
36981 (add_bytes): Use warn_level.
36982 (pass_sprintf_length::compute_format_length): Add debugging output.
36983 (try_substitute_return_value): Same.
36984 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
36986 PR middle-end/78703
36987 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
36988 (struct fmtresult, format_integer, format_floating): Adjust.
36989 (fmtresult::fmtresult): Set max correctly in two argument ctor.
36990 (get_string_length, format_string,format_directive): Same.
36991 (pass_sprintf_length::compute_format_length): Same.
36992 (try_substitute_return_value): Simplify slightly.
36994 PR middle-end/78703
36995 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
36996 (fmtresult::operator+=): Outlined.
36997 (struct fmtresult): Add ctors.
36998 (struct conversion_spec): Rename...
36999 (struct directive): ...to this. Add and remove data members.
37000 (directive::set_width, directive::set_precision): New functions.
37001 (format_percent): Use fmtresult ctor.
37002 (get_width_and_precision): Remove.
37003 (format_integer): Make naming changes. Avoid computing width and
37005 (format_floating): Same. Adjust indentation.
37006 (format_character, format_none): New functions.
37007 (format_string): Moved character handling to format_character.
37008 (format_directive): Remove arguments, change return type.
37009 (parse_directive): New function.
37010 (pass_sprintf_length::compute_format_length): Move directive
37011 parsing to parse_directive.
37013 2017-01-23 Jakub Jelinek <jakub@redhat.com>
37015 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
37016 (assign_assembler_name_if_needed): ... this.
37017 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
37018 (assign_assembler_name_if_needed): ... this.
37019 (free_lang_data_in_cgraph): Adjust callers.
37020 * cgraphunit.c (cgraph_node::analyze): Likewise.
37021 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
37024 2017-01-23 Richard Biener <rguenther@suse.de>
37026 PR tree-optimization/79088
37027 PR tree-optimization/79188
37028 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
37029 resetting loop bounds after last path deletion. Reset loop
37030 bounds of the target loop, make code match the comments.
37031 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
37032 Make sure loops need no fixups.
37034 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
37036 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
37037 exponent support with double type for first argument.
37038 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
37039 type returned by __builtin_vec_extract_sig,
37040 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
37041 functions from "vector int" to "vector unsigned int" or from
37042 "vector long long int" to "vector unsigned long long int".
37043 Changed type returned by __builtin_vec_extract_exp,
37044 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
37045 functions from "vector int" to "vector unsigned int" or from
37046 "vector long long int" to "vector unsigned long long int".
37047 Changed return type of __builtin_vec_test_data_class,
37048 __builtin_vec_test_data_class_sp, and
37049 __builtin_vec_test_data_class_dp from "vector int" to
37050 "vector bool int" or from "vector long long int" to "vector bool
37051 long long int" and changed second argument type from "unsigned
37052 int" to "int". Added new overloaded function forms "vector float
37053 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
37054 "vector float __builtin_vec_insert_exp_sp (vector float, vector
37055 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
37056 double, vector unsigned long long int)" and "vector double
37057 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
37058 long int)". Changed return type of
37059 __builtin_scalar_test_data_class and
37060 __builtin_scalar_test_data_class_sp and
37061 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
37062 int" and changed second argument from "unsigned int" to "int".
37063 Changed type returned by __builtin_scalar_test_neg,
37064 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
37065 from "int" to "bool int". Added new overloaded function form
37066 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
37067 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
37068 exponent double-precision with floating point first argument.
37069 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
37070 documentation of scalar_test_data_class, scalar_test_neg,
37071 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
37072 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
37073 vec_test_data_class built-in functions to reflect refinements in
37074 their type signatures.
37076 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
37078 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
37080 (aarch64_elf_asm_destructor): Likewise.
37082 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
37084 PR rtl-optimization/78634
37085 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
37086 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
37087 * ifcvt.c (noce_try_cmove): Add missing cost check.
37089 PR rtl-optimization/71724
37090 * combine.c (if_then_else_cond): Look for situations where it is
37091 beneficial to undo the work of one of the recursive calls.
37093 2017-01-23 Bin Cheng <bin.cheng@arm.com>
37095 PR tree-optimization/70754
37096 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
37097 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
37098 combined stmt before it if not NULL.
37099 (combine_chains): Process refs reversely and compute dominance point
37102 2017-01-23 Martin Liska <mliska@suse.cz>
37104 PR tree-optimization/79196
37105 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
37106 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
37108 (strlen_optimize_stmt): Call the renamed function.
37110 2017-01-23 Michael Matz <matz@suse.de>
37112 PR tree-optimization/78384
37113 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
37115 2017-01-23 Richard Biener <rguenther@suse.de>
37117 PR tree-optimization/79186
37118 * tree-vrp.c (register_new_assert_for): Make sure we've seen
37119 both incoming edges before moving an assert.
37121 2017-01-23 Martin Jambor <mjambor@suse.cz>
37123 * ipa-prop.c (load_from_param_1): Removed.
37124 (load_from_unmodified_param): Bits from load_from_param_1 put back
37126 (load_from_param): Removed.
37127 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
37128 with stmt. Reverted back to use of load_from_unmodified_param.
37130 2017-01-23 Martin Jambor <mjambor@suse.cz>
37133 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
37134 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
37135 field a pointer to garbage collected vector, mark lattices and
37136 ipcp_orig_node with GTY((skip)).
37137 (ipa_get_param_count): Adjust to descriptors being a pointer.
37138 (ipa_get_param): Likewise.
37139 (ipa_get_type): Likewise.
37140 (ipa_get_param_move_cost): Likewise.
37141 (ipa_set_param_used): Likewise.
37142 (ipa_get_controlled_uses): Likewise.
37143 (ipa_set_controlled_uses): Likewise.
37144 (ipa_is_param_used): Likewise.
37145 (ipa_node_params_t): Move into garbage collector. New methods insert
37147 (ipa_node_params_sum): Annotate wth GTY(()).
37148 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
37150 (ipa_load_from_parm_agg): Adjust declaration.
37151 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
37152 * ipa-profile.c (ipa_profile): Likewise.
37153 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
37154 (ipa_populate_param_decls): Make descriptors parameter garbage
37156 (ipa_dump_param): Adjust to descriptors being a pointer.
37157 (ipa_alloc_node_params): Likewise.
37158 (ipa_initialize_node_params): Likewise.
37159 (load_from_param_1): Make descriptors parameter garbage collected.
37160 (load_from_unmodified_param): Likewise.
37161 (load_from_param): Likewise.
37162 (ipa_load_from_parm_agg): Likewise.
37163 (ipa_node_params::~ipa_node_params): Removed.
37164 (ipa_free_all_node_params): Remove call to delete operator.
37165 (ipa_node_params_t::insert): New.
37166 (ipa_node_params_t::remove): Likewise.
37167 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
37168 copy known_csts and known_contexts vectors.
37169 (ipa_read_node_info): Adjust to descriptors being a pointer.
37170 (ipcp_modif_dom_walker): Make m_descriptors field garbage
37172 (ipcp_transform_function): Make descriptors variable garbage
37175 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
37177 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
37178 * config/i386/avx512dqintrin.h: Ditto.
37179 * config/i386/avx512fintrin.h: Ditto.
37180 * config/i386/i386.c: Handle new builtins.
37181 * config/i386/i386-builtin.def: Add new builtins.
37182 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
37183 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
37185 2017-01-23 Jakub Jelinek <jakub@redhat.com>
37186 Martin Liska <mliska@suse.cz>
37188 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
37189 * asan.c (asan_expand_poison_ifn): Support stores and use
37190 appropriate ASAN report function.
37191 * internal-fn.c (expand_ASAN_POISON_USE): New function.
37192 * internal-fn.def (ASAN_POISON_USE): Declare.
37193 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
37194 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
37195 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
37196 ASAN_POISON calls w/o LHS.
37197 * tree-ssa.c (execute_update_addresses_taken): Create clobber
37198 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
37199 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
37200 * gimplify.c (asan_poison_variables): Add attribute
37201 use_after_scope_memory to variables that really needs to live
37203 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
37204 having the attribute.
37206 2017-01-23 Martin Liska <mliska@suse.cz>
37208 * asan.c (create_asan_shadow_var): New function.
37209 (asan_expand_poison_ifn): Likewise.
37210 * asan.h (asan_expand_poison_ifn): New declaration.
37211 * internal-fn.c (expand_ASAN_POISON): Likewise.
37212 * internal-fn.def (ASAN_POISON): New builtin.
37213 * sanopt.c (pass_sanopt::execute): Expand
37214 asan_expand_poison_ifn.
37215 * tree-inline.c (copy_decl_for_dup_finish): Make function
37217 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
37218 * tree-ssa.c (is_asan_mark_p): New function.
37219 (execute_update_addresses_taken): Rewrite local variables
37220 (identified just by use-after-scope as addressable) into SSA.
37222 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
37224 * doc/install.texi (Specific): opensource.apple.com uses https
37225 now. Remove trailing slash.
37227 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
37229 * README.Portability: Remove note on an Irix compatibility issue.
37231 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
37233 * gcov.c (INCLUDE_ALGORITHM): Define.
37234 (INCLUDE_VECTOR): Define.
37235 No longer include <vector> and <algorithm> directly.
37237 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
37239 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
37241 * doc/invoke.texi (Code Gen Options): Ditto.
37243 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
37246 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
37248 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
37250 rtl-optimization/79125
37251 * cprop.c (local_cprop_pass): Handle cases where we make an
37252 unconditional trap.
37254 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
37258 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
37259 read from, for big endian.
37261 2017-01-20 Jiong Wang <jiong.wang@arm.com>
37263 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
37264 register pauth builtins for LP64 only.
37266 2017-01-20 Marek Polacek <polacek@redhat.com>
37269 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
37272 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
37274 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
37276 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
37277 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
37278 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
37280 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37283 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
37284 in big-endian mode when they are not a single duplicated value.
37286 2017-01-20 Richard Biener <rguenther@suse.de>
37288 * BASE-VER: Bump to 7.0.1.
37290 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
37292 * omp-low.c (omplow_simd_context): New struct. Use it...
37293 (lower_rec_simd_input_clauses): ...here and...
37294 (lower_rec_input_clauses): ...here to hold common data. Adjust all
37295 references to idx, lane, max_vf, is_simt.
37297 2017-01-20 Graham Markall <graham.markall@embecosm.com>
37299 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
37302 2017-01-20 Martin Jambor <mjambor@suse.cz>
37304 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
37305 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
37307 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
37308 (GTFILES): Rename hsa.c to hsa-common.c.
37309 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
37310 * hsa-dump.c: Likewise.
37311 * hsa-gen.c: Likewise.
37312 * hsa-regalloc.c: Likewise.
37313 * ipa-hsa.c: Likewise.
37314 * omp-expand.c: Likewise.
37315 * omp-low.c: Likewise.
37316 * toplev.c: Likewise.
37318 2017-01-20 Marek Polacek <polacek@redhat.com>
37321 * doc/invoke.texi: Document -Wduplicated-branches.
37322 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
37323 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
37324 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
37325 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
37326 return 0 only when not OEP_LEXICOGRAPHIC.
37327 (fold_build_cleanup_point_expr): Use the expression
37328 location when building CLEANUP_POINT_EXPR.
37329 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
37330 * tree.c (add_expr): Handle error_mark_node.
37332 2017-01-20 Martin Liska <mliska@suse.cz>
37335 * tree-profile.c (init_ic_make_global_vars): Do not call
37337 (gimple_init_gcov_profiler): Likewise.
37339 2017-01-20 Martin Liska <mliska@suse.cz>
37342 * cgraph.h (maybe_create_reference): Remove argument and
37344 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
37346 * ipa-cp.c (create_specialized_node): Likewise.
37347 * symtab.c (symtab_node::maybe_create_reference): Handle
37348 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
37350 2017-01-20 Martin Liska <mliska@suse.cz>
37352 * read-rtl-function.c (function_reader::create_function): Use
37353 build_decl instread of build_decl_stat.
37355 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
37357 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
37358 * config/i386/avx512dqintrin.h: Ditto.
37359 * config/i386/avx512fintrin.h: Ditto.
37360 * config/i386/i386-builtin-types.def: Add new types.
37361 * config/i386/i386.c: Handle new types.
37362 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
37363 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
37364 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
37365 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
37366 (__builtin_ia32_kshiftridi): New.
37367 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
37369 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
37373 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
37374 define to rs6000_init_stack_protect_guard.
37375 (rs6000_init_stack_protect_guard): New function.
37377 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
37378 Yunqiang Su <yunqiang.su@imgtec.com>
37380 * config.gcc (supported_defaults): Add madd4.
37381 (with_madd4): Add validation.
37382 (all_defaults): Add madd4.
37383 * config/mips/mips.opt (mmadd4): New option.
37384 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
37386 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
37388 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
37389 (ISA_HAS_FUSED_MADD4): Likewise.
37390 * doc/invoke.texi (-mmadd4): Document the new option.
37391 * doc/install.texi (--with-madd4): Document the new option.
37393 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37395 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
37396 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
37397 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
37398 (aarch64_init_pauth_hint_builtins): New.
37399 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
37400 (aarch64_expand_builtin): Expand new builtins.
37402 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37404 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
37405 * combine-stack-adj.c (no_unhandled_cfa): Handle
37406 REG_CFA_TOGGLE_RA_MANGLE.
37407 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
37408 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
37409 info for return address signing.
37410 (aarch64_expand_epilogue): Likewise.
37412 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37414 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
37415 * config/aarch64/aarch64-protos.h
37416 (aarch64_return_address_signing_enabled): New declaration.
37417 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
37419 (aarch64_expand_prologue): Sign return address before it's pushed onto
37421 (aarch64_expand_epilogue): Authenticate return address fetched from
37423 (aarch64_override_options): Sanity check for ILP32 and ISA level.
37424 (aarch64_attributes): New function attributes for "sign-return-address".
37425 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
37426 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
37427 ("*do_return"): Generate combined instructions according to key index.
37428 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
37429 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
37431 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
37432 * config/aarch64/aarch64.opt (msign-return-address=): New.
37433 * doc/extend.texi (AArch64 Function Attributes): Documents
37434 "sign-return-address=".
37435 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
37437 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
37439 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
37440 overall option summary.
37442 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37444 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
37445 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
37446 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
37447 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
37449 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
37451 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
37452 -mpower9-minmax by default for -mcpu=power9.
37453 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
37454 128-bit floating point.
37456 2017-01-20 Alan Modra <amodra@gmail.com>
37458 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
37459 optimizing for size.
37461 2017-01-20 Alan Modra <amodra@gmail.com>
37464 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
37465 for strcmp and strncmp from corresponding builtin decl.
37467 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
37469 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
37470 instead of i386/rtems-64.h.
37471 * config/i386/rtems-64.h: Remove.
37473 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
37477 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
37479 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
37481 2017-01-19 Tamar Christina <tamar.christina@arm.com>
37483 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
37484 Change int to HOST_WIDE_INT.
37485 * config/aarch64/aarch64-protos.h
37486 (aarch64_simd_gen_const_vector_dup): Likewise.
37487 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
37489 2017-01-19 David Malcolm <dmalcolm@redhat.com>
37491 * langhooks-def.h (lhd_type_for_size): New decl.
37492 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
37493 * langhooks.c (lhd_type_for_size): New function, taken from
37496 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
37498 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
37499 define_bypass for CR latency.
37500 (power9-cracked-alu): Update bypass latency and remove power9-branch.
37501 (power9-alu2): Add define_bypass for CR latency.
37503 (power9-mul): Update insn latency.
37504 (power9-mul-compare): Update insn latency, bypass latency and remove
37507 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37509 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
37511 * config/aarch64/aarch64.md
37512 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
37513 aarch64_nopcrelative_literal_loads.
37514 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
37516 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
37518 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
37519 TARGET_LOONGSON_3A.
37520 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
37522 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
37525 * config.gcc (supported_defaults): Add lxc1-sxc1.
37526 (with_lxc1_sxc1): Add validation.
37527 (all_defaults): Add lxc1-sxc1.
37528 * config/mips/mips.opt (mlxc1-sxc1): New option.
37529 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
37531 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
37532 __mips_no_lxc1_sxc1.
37533 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
37534 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
37535 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
37537 2017-01-19 Richard Biener <rguenther@suse.de>
37539 PR tree-optimization/72488
37540 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
37541 sure to restore SSA info.
37542 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
37544 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
37546 PR rtl-optimization/79121
37547 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
37548 of the inner type when shifting an extended value.
37550 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
37553 * symtab.c (symtab_node::equal_address_to): Fix comparing of
37554 interposable aliases.
37556 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
37559 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
37560 Use the evmergelohi instruction.
37561 (mov_si<mode>_e500_subreg4_2_le): Likewise.
37562 (mov_sitf_e500_subreg8_2_be): Likewise.
37563 (mov_sitf_e500_subreg12_2_le): Likewise.
37564 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
37565 (mov_si<mode>_e500_subreg4_2_be): Likewise.
37566 (mov_sitf_e500_subreg8_2_le): Likewise.
37567 (mov_sitf_e500_subreg12_2_be): Likewise.
37569 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37571 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
37572 attribute from vecsimple to vecperm.
37573 (altivec_vbpermq2): Likewise.
37575 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37578 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
37580 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37581 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
37582 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
37583 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
37584 case where N arg is SIZE_MAX.
37585 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
37586 (cmpstrsi): Add pattern.
37588 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
37590 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
37591 __builtin_vec_revb builtins.
37592 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
37593 built-in functions to support generation of the ISA 3.0 XXBR<x>
37594 vector byte reverse instructions.
37595 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
37596 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
37597 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
37598 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
37599 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
37600 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
37601 (P9V_BUILTIN_VEC_REVB): Likewise.
37602 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
37603 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
37604 (p9_xxbrq_v16qi): Likewise.
37605 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
37606 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
37607 (p9_xxbrh_v8hi): Likewise.
37608 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
37609 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
37610 vec_revb built-in functions.
37612 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
37614 PR rtl-optimization/78952
37615 * config/i386/i386.md (any_extract): New code iterator.
37616 (*insvqi_2): Use any_extract for source operand.
37617 (*insvqi_3): Use any_shiftrt for source operand.
37619 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
37621 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
37623 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
37625 2017-01-18 Matthias Klose <doko@ubuntu.com>
37627 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
37629 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37631 * config/rs6000/altivec.h (vec_bperm): Change #define.
37632 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
37633 (altivec_vbpermq2): New define_insn.
37634 (altivec_vbpermd): Likewise.
37635 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
37636 function interface.
37637 (VBPERMD): Likewise.
37638 (VBPERM): New polymorphic function interface.
37639 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
37640 Add entries for P9V_BUILTIN_VEC_VBPERM.
37641 * doc/extend.texi: Add interfaces for vec_bperm.
37643 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37645 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
37646 first letter of error messages.
37647 (s390_resolve_overloaded_builtin): Likewise.
37648 * config/s390/s390.c (s390_expand_builtin): Likewise.
37649 (s390_invalid_arg_for_unprototyped_fn): Likewise.
37650 (s390_valid_target_attribute_inner_p): Likewise.
37651 * config/s390/s390.md ("tabort"): Likewise.
37653 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
37655 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
37656 (ISA_AVOID_DIV_HILO): New macro.
37657 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
37658 (ISA_HAS_DDIV): Likewise.
37660 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37662 * doc/invoke.texi (fabi-version): Correct number of occurrences.
37664 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37666 * doc/invoke.texi (fabi-version): Spelling fix.
37668 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37671 * doc/invoke.texi (fabi-version): Mention mangling fix for
37674 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37677 * doc/invoke.texi (fabi-version): Document discriminator mangling.
37679 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
37682 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
37683 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
37685 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
37687 (stack_protect_test): Ditto.
37688 * config/rs6000/rs6000.opt (mstack-protector-guard=,
37689 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
37691 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
37692 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
37693 -mstack-protector-guard-offset=.
37694 (RS/6000 and PowerPC Options): Ditto.
37696 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37698 * config/i386/i386.h (MASK_CLASS_P): New define.
37699 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
37700 there are no registers from different register sets also when
37701 mask registers are used. Update function comment.
37702 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
37703 to (*k/*r) and (*k/*km) alternatives.
37705 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
37707 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
37708 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
37709 (EH_RETURN_HANDLER_RTX): New define.
37710 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
37711 Force frame pointer in EH return functions.
37712 (aarch64_expand_epilogue): Add barrier for eh_return.
37713 (aarch64_final_eh_return_addr): Remove.
37714 (aarch64_eh_return_handler_rtx): New function.
37715 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
37717 (aarch64_eh_return_handler_rtx): New prototype.
37719 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37721 * config/rs6000/altivec.h (vec_rlmi): New #define.
37722 (vec_vrlnm): Likewise.
37723 (vec_rlnm): Likewise.
37724 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
37725 (UNSPEC_VRLNM): Likewise.
37726 (VIlong): New mode iterator.
37727 (altivec_vrl<VI_char>mi): New define_insn.
37728 (altivec_vrl<VI_char>nm): Likewise.
37729 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
37731 (VRLDNM): Likewise.
37732 (RLNM): New polymorphic function entry.
37733 (VRLWMI): New monomorphic function entry.
37734 (VRLDMI): Likewise.
37735 (RLMI): New polymorphic function entry.
37736 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
37737 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
37738 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
37741 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37744 * dwarf2out.c (field_byte_offset): Restore the
37745 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
37746 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
37747 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
37750 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37753 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
37755 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37758 * dwarf2out.c (add_data_member_location_attribute): For constant
37759 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
37760 instead of DW_AT_data_member_location, DW_AT_bit_offset and
37761 DW_AT_byte_size attributes.
37763 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37765 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
37766 after forcing to constant memory when the code model is medium.
37768 2017-01-17 Julia Koval <julia.koval@intel.com>
37771 * config/i386/avx512fintrin.h
37772 (_mm512_i32gather_ps): Change __addr type to void const*.
37773 (_mm512_mask_i32gather_ps): Ditto.
37774 (_mm512_i32gather_pd): Ditto.
37775 (_mm512_mask_i32gather_pd): Ditto.
37776 (_mm512_i64gather_ps): Ditto.
37777 (_mm512_mask_i64gather_ps): Ditto.
37778 (_mm512_i64gather_pd): Ditto.
37779 (_mm512_mask_i64gather_pd): Ditto.
37780 (_mm512_i32gather_epi32): Ditto.
37781 (_mm512_mask_i32gather_epi32): Ditto.
37782 (_mm512_i32gather_epi64): Ditto.
37783 (_mm512_mask_i32gather_epi64): Ditto.
37784 (_mm512_i64gather_epi32): Ditto.
37785 (_mm512_mask_i64gather_epi32): Ditto.
37786 (_mm512_i64gather_epi64): Ditto.
37787 (_mm512_mask_i64gather_epi64): Ditto.
37788 (_mm512_i32scatter_ps): Change __addr type to void*.
37789 (_mm512_mask_i32scatter_ps): Ditto.
37790 (_mm512_i32scatter_pd): Ditto.
37791 (_mm512_mask_i32scatter_pd): Ditto.
37792 (_mm512_i64scatter_ps): Ditto.
37793 (_mm512_mask_i64scatter_ps): Ditto.
37794 (_mm512_i64scatter_pd): Ditto.
37795 (_mm512_mask_i64scatter_pd): Ditto.
37796 (_mm512_i32scatter_epi32): Ditto.
37797 (_mm512_mask_i32scatter_epi32): Ditto.
37798 (_mm512_i32scatter_epi64): Ditto.
37799 (_mm512_mask_i32scatter_epi64): Ditto.
37800 (_mm512_i64scatter_epi32): Ditto.
37801 (_mm512_mask_i64scatter_epi32): Ditto.
37802 (_mm512_i64scatter_epi64): Ditto.
37803 (_mm512_mask_i64scatter_epi64): Ditto.
37804 * config/i386/avx512pfintrin.h
37805 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
37806 (_mm512_mask_prefetch_i32gather_ps): Ditto.
37807 (_mm512_mask_prefetch_i64gather_pd): Ditto.
37808 (_mm512_mask_prefetch_i64gather_ps): Ditto.
37809 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
37810 (_mm512_prefetch_i32scatter_ps): Ditto.
37811 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
37812 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
37813 (_mm512_prefetch_i64scatter_pd): Ditto.
37814 (_mm512_prefetch_i64scatter_ps): Ditto.
37815 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
37816 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
37817 * config/i386/avx512vlintrin.h
37818 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
37819 (_mm_mmask_i32gather_ps): Ditto.
37820 (_mm256_mmask_i32gather_pd): Ditto.
37821 (_mm_mmask_i32gather_pd): Ditto.
37822 (_mm256_mmask_i64gather_ps): Ditto.
37823 (_mm_mmask_i64gather_ps): Ditto.
37824 (_mm256_mmask_i64gather_pd): Ditto.
37825 (_mm_mmask_i64gather_pd): Ditto.
37826 (_mm256_mmask_i32gather_epi32): Ditto.
37827 (_mm_mmask_i32gather_epi32): Ditto.
37828 (_mm256_mmask_i32gather_epi64): Ditto.
37829 (_mm_mmask_i32gather_epi64): Ditto.
37830 (_mm256_mmask_i64gather_epi32): Ditto.
37831 (_mm_mmask_i64gather_epi32): Ditto.
37832 (_mm256_mmask_i64gather_epi64): Ditto.
37833 (_mm_mmask_i64gather_epi64): Ditto.
37834 (_mm256_i32scatter_ps): Change __addr type to void*.
37835 (_mm256_mask_i32scatter_ps): Ditto.
37836 (_mm_i32scatter_ps): Ditto.
37837 (_mm_mask_i32scatter_ps): Ditto.
37838 (_mm256_i32scatter_pd): Ditto.
37839 (_mm256_mask_i32scatter_pd): Ditto.
37840 (_mm_i32scatter_pd): Ditto.
37841 (_mm_mask_i32scatter_pd): Ditto.
37842 (_mm256_i64scatter_ps): Ditto.
37843 (_mm256_mask_i64scatter_ps): Ditto.
37844 (_mm_i64scatter_ps): Ditto.
37845 (_mm_mask_i64scatter_ps): Ditto.
37846 (_mm256_i64scatter_pd): Ditto.
37847 (_mm256_mask_i64scatter_pd): Ditto.
37848 (_mm_i64scatter_pd): Ditto.
37849 (_mm_mask_i64scatter_pd): Ditto.
37850 (_mm256_i32scatter_epi32): Ditto.
37851 (_mm256_mask_i32scatter_epi32): Ditto.
37852 (_mm_i32scatter_epi32): Ditto.
37853 (_mm_mask_i32scatter_epi32): Ditto.
37854 (_mm256_i32scatter_epi64): Ditto.
37855 (_mm256_mask_i32scatter_epi64): Ditto.
37856 (_mm_i32scatter_epi64): Ditto.
37857 (_mm_mask_i32scatter_epi64): Ditto.
37858 (_mm256_i64scatter_epi32): Ditto.
37859 (_mm256_mask_i64scatter_epi32): Ditto.
37860 (_mm_i64scatter_epi32): Ditto.
37861 (_mm_mask_i64scatter_epi32): Ditto.
37862 (_mm256_i64scatter_epi64): Ditto.
37863 (_mm256_mask_i64scatter_epi64): Ditto.
37864 (_mm_i64scatter_epi64): Ditto.
37865 (_mm_mask_i64scatter_epi64): Ditto.
37866 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
37867 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
37868 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
37869 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
37870 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
37871 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
37872 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
37873 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
37874 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
37875 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
37876 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
37877 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
37878 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
37879 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
37880 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
37881 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
37882 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
37883 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
37884 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
37885 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
37886 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
37887 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
37888 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
37889 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
37890 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
37891 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
37892 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
37893 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
37894 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
37895 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
37896 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
37897 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
37898 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
37899 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
37900 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
37901 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
37902 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
37903 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
37904 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
37905 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
37906 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
37907 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
37908 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
37909 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
37910 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
37911 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
37912 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
37913 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
37914 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
37915 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
37916 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
37917 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
37918 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
37919 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
37920 definitions accordingly.
37922 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
37923 Kuan-Lin Chen <kuanlinchentw@gmail.com>
37926 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
37929 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
37932 * ira-conflicts.c (ira_build_conflicts): Update total conflict
37933 hard regs for inner regno.
37935 2017-01-17 Martin Liska <mliska@suse.cz>
37938 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
37939 assumption and add comment.
37941 2017-01-17 Nathan Sidwell <nathan@acm.org>
37943 * ipa-visibility.c (localize_node): New function, broken out of ...
37944 (function_and_variable_visibility): ... here. Call it.
37946 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
37948 PR middle-end/77445
37949 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
37950 correctly set frequency of oudgoing edge.
37951 (duplicate_thread_path): Fix profile updating.
37953 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37956 * configure.ac: Add GCC_BASE_VER.
37957 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
37958 version from BASE-VER file.
37959 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
37960 (gcc.o): Depend on $(BASEVER).
37961 * common.opt (dumpfullversion): New option.
37962 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
37963 * doc/invoke.texi: Document -dumpfullversion.
37964 * doc/install.texi: Document --with-gcc-major-version-only.
37965 * configure: Regenerated.
37967 2017-01-17 Richard Biener <rguenther@suse.de>
37969 PR tree-optimization/71433
37970 * tree-vrp.c (register_new_assert_for): Merge same asserts
37971 on all incoming edges.
37972 (process_assert_insertions_for): Handle insertions at the
37975 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
37977 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
37978 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
37980 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
37983 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
37986 2017-01-17 Alan Modra <amodra@gmail.com>
37989 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
37990 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
37991 symbolic stack limit when pic.
37993 2017-01-16 Martin Sebor <msebor@redhat.com>
37995 PR tree-optimization/78608
37996 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
37998 2017-01-16 Jeff Law <law@redhat.com>
38001 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
38002 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
38003 for several include directories that may be relative to sysroot.
38004 * config/i386/x-mingw32 (gplus_includedir): Define.
38005 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
38006 (native_system_includedir): Likewise.
38007 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
38008 override if TARGET_SYSTEM_ROOT is defined.
38009 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
38011 PR tree-optimization/79090
38012 PR tree-optimization/33562
38013 PR tree-optimization/61912
38014 PR tree-optimization/77485
38015 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
38016 and computed trims into the dump file.
38018 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
38020 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
38022 2017-01-16 Jakub Jelinek <jakub@redhat.com>
38025 * gimplify.c (gimplify_init_constructor): If want_value and
38026 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
38030 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
38031 sequence. Formatting fixes.
38032 (doloop_optimize): Formatting fixes.
38035 * gcc.c (debug_level_greater_than_spec_func): New function.
38036 (static_spec_functions): Add debug-level-gt spec function.
38037 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
38039 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
38040 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
38041 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
38042 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
38043 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
38044 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
38046 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
38048 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
38049 QImode fixups to general and mask registers only.
38051 2017-01-16 Carl Love <cel@us.ibm.com>
38053 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
38054 for built-in functions
38055 vector signed char vec_nabs (vector signed char)
38056 vector signed short vec_nabs (vector signed short)
38057 vector signed int vec_nabs (vector signed int)
38058 vector signed long long vec_nabs (vector signed long long)
38059 vector float vec_nabs (vector float)
38060 vector double vec_nabs (vector double)
38061 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
38063 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
38064 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
38065 * doc/extend.texi: Update the documentation file for the new built-in
38068 2017-01-16 Martin Sebor <msebor@redhat.com>
38070 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
38073 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38075 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
38076 UNSPEC_VSX__XXSPLTD to require special splat handling.
38078 2017-01-16 David Malcolm <dmalcolm@redhat.com>
38081 * system.h: Poison strndup.
38083 2017-01-16 Alan Modra <amodra@gmail.com>
38086 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
38089 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
38091 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
38093 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
38095 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
38096 call recog here. Assert that INSN_CODE (insn) is non-negative.
38098 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
38101 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
38103 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
38104 in the currently scheduled RTL fragment.
38106 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
38108 PR rtl-optimization/78751
38109 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
38112 2017-01-14 Jeff Law <law@redhat.com>
38114 PR tree-optimization/79090
38115 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
38116 variable length stores.
38117 (compute_trims): Delete dead assignment to *trim_tail.
38118 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
38121 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
38123 PR rtl-optimization/78626
38124 PR rtl-optimization/78727
38125 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
38126 of a block, and split such blocks after everything else is finished.
38128 2017-01-14 Alan Modra <amodra@gmail.com>
38131 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
38132 target legitimate_combined_insn.
38133 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
38134 (rs6000_legitimate_combined_insn): New function.
38135 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
38137 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
38138 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
38139 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
38141 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
38143 * doc/frontends.texi (G++ and GCC): Remove references to Java.
38145 2017-01-13 Jeff Law <law@redhat.com>
38147 PR tree-optimization/33562
38148 PR tree-optimization/61912
38149 PR tree-optimization/77485
38150 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
38152 (delete_dead_assignment): Likewise.
38153 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
38154 statement to delete_dead_call and delete_dead_assignment.
38156 2017-01-13 David Malcolm <dmalcolm@redhat.com>
38159 * substring-locations.c (format_warning_va): Strengthen case 1 so
38160 that both endpoints of the substring must be within the format
38161 range for just the substring to be printed.
38163 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
38165 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
38166 * config/i386/i386.c (ix86_target_string): Add missing options
38167 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
38168 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
38169 flags_other and ix86_target_other to flags2_other. Display unknown
38171 (ix86_valid_target_attribute_inner_p): Add missing options and
38172 reorder options by implied ISAs, as in ix86_target_string.
38174 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
38176 * hash-table.h (hash_table::too_empty_p): New function.
38177 (hash_table::expand): Use it.
38178 (hash_table::traverse): Likewise.
38179 (hash_table::empty_slot): Use sizeof (value_type) instead of
38180 sizeof (PTR) to convert bytes to elements. Shrink the table
38181 if the current size is excessive for the current number of
38184 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
38186 * ira-costs.c (record_reg_classes): Break from the inner loop
38187 early once alt_fail is known to be true. Update outer loop
38188 handling accordingly.
38190 2017-01-13 Jeff Law <law@redhat.com>
38192 * tree-ssa-dse.c (decrement_count): New function.
38193 (increment_start_addr, maybe_trim_memstar_call): Likewise.
38194 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
38195 when we know the partially dead statement is a mem* function.
38197 PR tree-optimization/61912
38198 PR tree-optimization/77485
38199 * tree-ssa-dse.c: Include expr.h.
38200 (maybe_trim_constructor_store): New function.
38201 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
38203 PR tree-optimization/33562
38204 PR tree-optimization/61912
38205 PR tree-optimization/77485
38206 * doc/invoke.texi: Document new dse-max-object-size param.
38207 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
38208 * tree-ssa-dse.c: Include params.h.
38209 (dse_store_status): New enum.
38210 (initialize_ao_ref_for_dse): New, partially extracted from
38212 (valid_ao_ref_for_dse, normalize_ref): New.
38213 (setup_live_bytes_from_ref, compute_trims): Likewise.
38214 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
38215 (maybe_trim_partially_dead_store): Likewise.
38216 (maybe_trim_complex_store): Likewise.
38217 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
38218 Track what bytes live from the original store. Return tri-state
38219 for dead, partially dead or live.
38220 (dse_dom_walker): Add constructor, destructor and new private members.
38221 (delete_dead_call, delete_dead_assignment): New extracted from
38223 (dse_optimize_stmt): Make a member of dse_dom_walker.
38224 Use initialize_ao_ref_for_dse.
38226 PR tree-optimization/33562
38227 PR tree-optimization/61912
38228 PR tree-optimization/77485
38229 * sbitmap.h (bitmap_count_bits): Prototype.
38230 (bitmap_clear_range, bitmap_set_range): Likewise.
38231 * sbitmap.c (bitmap_clear_range): New function.
38232 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
38234 2017-01-13 Martin Liska <mliska@suse.cz>
38237 * function.c (set_cfun): Add new argument force.
38238 * function.h (set_cfun): Likewise.
38239 * ipa-inline-transform.c (inline_call): Use the function when
38240 strict alising from is dropped for function we inline to.
38242 2017-01-13 Richard Biener <rguenther@suse.de>
38244 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
38245 for dumping GIMPLE INTEGER_CSTs.
38247 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
38249 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
38250 to 201112L since C++17.
38252 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
38255 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
38256 if -fsanitize=kernel-address is present.
38258 2017-01-13 Richard Biener <rguenther@suse.de>
38260 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
38261 as _Literal ( type ) number in case usual suffixes do not
38262 preserve all information.
38264 2017-01-13 Richard Biener <rguenther@suse.de>
38266 PR tree-optimization/77283
38267 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
38268 and ssa-iterators.h.
38269 (is_feasible_trace): Implement a cost model based on joiner
38272 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
38275 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
38276 char or short to __float128/_Float128 directly.
38278 2017-01-12 Martin Sebor <msebor@redhat.com>
38280 to -Wformat-overflow.
38281 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
38282 (min_bytes_remaining): Same.
38283 (get_string_length): Same.
38284 (format_string): Same.
38285 (format_directive): Same.
38287 (pass_sprintf_length::handle_gimple_call): Same.
38289 2017-01-12 Jakub Jelinek <jakub@redhat.com>
38291 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
38292 info.nowrite calls with no lhs that can't throw. Return bool
38293 whether gsi_remove has been called or not.
38294 (pass_sprintf_length::handle_gimple_call): Return bool whether
38295 try_substitute_return_value called gsi_remove. Formatting fix.
38296 (pass_sprintf_length::execute): Don't use gsi_remove if
38297 handle_gimple_call returned true.
38300 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
38301 be removed due to side-effects, don't remove following barrier nor
38302 turn the successor edge into fallthru edge.
38304 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38307 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
38308 element-reversing loads and stores as not swappable.
38310 2017-01-12 Nathan Sidwell <nathan@acm.org>
38311 Nicolai Stange <nicstange@gmail.com>
38313 * combine.c (try_combine): Don't ignore result of overlap checking
38314 loop. Combine overlap & asm check into single loop.
38316 2017-01-12 Richard Biener <rguenther@suse.de>
38318 * tree-pretty-print.c (dump_generic_node): Provide -gimple
38319 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
38321 2017-01-12 Richard Biener <rguenther@suse.de>
38323 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
38324 and TS_TARGET_OPTION directly derive from TS_BASE.
38325 * tree-core.h (tree_optimization_option): Derive from tree_base.
38326 (tree_target_option): Likewise.
38328 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
38330 * config/i386/i386.c (memory_address_length): Increase len
38331 only when rip_relative_addr_p returns false.
38333 2017-01-11 Julia Koval <julia.koval@intel.com>
38335 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
38336 (OPTION_MASK_ISA_SGX_SET): New.
38337 (ix86_handle_option): Handle OPT_msgx.
38338 * config.gcc: Added sgxintrin.h.
38339 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
38340 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
38341 * config/i386/i386.c (ix86_target_string): Add -msgx.
38343 (ix86_option_override_internal): Handle new options.
38344 (ix86_valid_target_attribute_inner_p): Add sgx.
38345 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
38346 * config/i386/i386.opt: Add msgx.
38347 * config/i386/sgxintrin.h: New file.
38348 * config/i386/x86intrin.h: Add sgxintrin.h.
38350 2017-01-11 Jakub Jelinek <jakub@redhat.com>
38353 * fold-const.c (maybe_nonzero_address): Return 1 for function
38355 (tree_single_nonzero_warnv_p): Don't handle function local objects
38359 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
38362 2017-01-11 David Malcolm <dmalcolm@redhat.com>
38365 * opts.c: Include "spellcheck.h"
38366 (struct string_fragment): New struct.
38367 (struct edit_distance_traits<const string_fragment &>): New
38369 (get_closest_sanitizer_option): New function.
38370 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
38372 2017-01-11 Jakub Jelinek <jakub@redhat.com>
38374 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
38376 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
38377 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
38378 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
38379 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
38380 for initial die_offset if dwarf_split_debug_info.
38381 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
38382 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
38384 (output_skeleton_debug_sections): Formatting fix. Use
38385 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
38386 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
38388 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
38390 * config/arm/cortex-a53.md: Add bypasses for
38391 cortex_a53_r2f_cvt.
38392 (cortex_a53_r2f): Only use for transfers.
38393 (cortex_a53_f2r): Likewise.
38394 (cortex_a53_r2f_cvt): Add reservation for conversions.
38395 (cortex_a53_f2r_cvt): Likewise.
38397 2017-01-11 Tamar Christina <tamar.christina@arm.com>
38399 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
38400 to all inlined functions, change static to extern.
38402 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
38405 * config/arm/arm.c (legitimize_pic_address): Handle reference to
38407 (arm_assemble_integer): Likewise.
38409 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
38411 * config.gcc: Use new awk script to check CPU, FPU and architecture
38412 parameters for --with-... options.
38413 * config/arm/parsecpu.awk: New file
38414 * config/arm/arm-cpus.in: New file.
38415 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
38417 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
38419 * config/arm/t-arm: Update dependency rules.
38420 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
38421 of processing .def files.
38422 * config/arm/genopt.sh: Deleted.
38423 * config/arm/gentune.sh: Deleted.
38424 * config/arm/arm-cores.def: Deleted.
38425 * config/arm/arm-arches.def: Deleted.
38426 * config/arm/arm-fpus.def: Deleted.
38427 * config/arm/arm-tune.md: Regenerated.
38428 * config/arm/arm-tables.opt: Regenerated.
38429 * config/arm/arm-cpu.h: New generated file.
38430 * config/arm/arm-cpu-data.h: New generated file.
38431 * config/arm/arm-cpu-cdata.h: New generated file.
38433 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
38436 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
38438 (input_varpool_node): Unpack dynamically_initialized bit.
38440 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
38442 PR rtl-optimization/79032
38443 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
38444 the alignment of the adjusted memory reference against that of MODE,
38445 instead of the alignment of the original memory reference.
38447 2017-01-11 Martin Jambor <mjambor@suse.cz>
38449 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
38451 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
38452 decorated functions.
38454 2017-01-11 Richard Biener <rguenther@suse.de>
38456 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
38457 set range/nonnull info for PHI results. Do not set it on
38458 stmts marked for removal.
38460 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
38462 * expr.c (store_field): In the bitfield case, fetch the return value
38463 from the registers before applying a single big-endian adjustment.
38464 Always do a final load for a BLKmode value not larger than a word.
38466 2017-01-10 David Malcolm <dmalcolm@redhat.com>
38469 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
38470 that we correctly handle column numbers greater than
38471 LINE_MAP_MAX_COLUMN_NUMBER.
38473 2017-01-10 Martin Sebor <msebor@redhat.com>
38475 PR middle-end/78245
38476 * gimple-ssa-sprintf.c (get_destination_size): Call
38477 {init,fini}object_sizes.
38478 * tree-object-size.c (addr_object_size): Adjust.
38479 (pass_through_call): Adjust.
38480 (pass_object_sizes::execute): Adjust.
38481 * tree-object-size.h (fini_object_sizes): Declare.
38483 2017-01-10 Martin Sebor <msebor@redhat.com>
38485 PR tree-optimization/78775
38486 * builtins.c (get_size_range): Move...
38487 * calls.c: ...to here.
38488 (alloc_max_size): Accept zero argument.
38489 (operand_signed_p): Remove.
38490 (maybe_warn_alloc_args_overflow): Call get_size_range.
38491 * calls.h (get_size_range): Declare.
38493 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
38495 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
38496 from TI's devices.csv file as of September 2016.
38497 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
38499 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
38501 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
38502 * doc/invoke.texi: Likewise.
38503 * doc/md.texi: Likewise.
38504 * doc/objc.texi: Likewise.
38506 2017-01-10 Joshua Conner <joshconner@google.com>
38508 * config/arm/fuchsia-elf.h: New file.
38509 * config/fuchsia.h: New file.
38510 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
38511 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
38513 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
38515 2016-01-10 Richard Biener <rguenther@suse.de>
38517 PR tree-optimization/79034
38518 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
38519 Propagate out degenerate PHIs in the joiner.
38521 2017-01-10 Martin Liska <mliska@suse.cz>
38523 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
38524 (sort_congruence_classes_by_decl_uid): Likewise.
38525 (sort_congruence_class_groups_by_decl_uid): Likewise.
38526 (sem_item_optimizer::merge_classes): Sort class, groups in these
38527 classes and members in the groups by DECL_UID of declarations.
38528 This would make merge operations stable.
38530 2017-01-10 Martin Liska <mliska@suse.cz>
38532 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
38533 usage of m_classes_vec.
38534 (sem_item_optimizer::~sem_item_optimizer): Likewise.
38535 (sem_item_optimizer::get_group_by_hash): Likewise.
38536 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
38537 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
38538 (sem_item_optimizer::verify_classes): Likewise.
38539 (sem_item_optimizer::process_cong_reduction): Likewise.
38540 (sem_item_optimizer::dump_cong_classes): Likewise.
38541 (sem_item_optimizer::merge_classes): Likewise.
38542 * ipa-icf.h (congruence_class_hash): Rename from
38543 congruence_class_group_hash. Remove declaration of m_classes_vec.
38545 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
38547 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
38548 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
38549 * config.gcc: Add avx512vpopcntdqintrin.h.
38550 * config/i386/avx512vpopcntdqintrin.h: New.
38551 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
38552 * config/i386/i386-builtin-types.def: Add new types.
38553 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
38554 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
38555 __builtin_ia32_vpopcountq_v8di_mask): New.
38556 * config/i386/i386-c.c (ix86_target_macros_internal): Define
38557 __AVX512VPOPCNTDQ__.
38558 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
38559 (PTA_AVX512VPOPCNTDQ): Define.
38560 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
38561 TARGET_AVX512VPOPCNTDQ_P): Define.
38562 * config/i386/i386.opt: Add mavx512vpopcntdq.
38563 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
38564 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
38566 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38568 PR middle-end/77484
38569 * predict.def (PRED_CALL): Set to 67.
38571 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
38573 * expr.c (store_field): In the bitfield case, if the value comes from
38574 a function call and is of an aggregate type returned in registers, do
38575 not modify the field mode; extract the value in all cases if the mode
38576 is BLKmode and the size is not larger than a word.
38578 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
38581 * config/i386/cpuid.h: Fix undefined behavior.
38583 2017-01-04 Jeff Law <law@redhat.com>
38585 PR tree-optimization/79007
38586 PR tree-optimization/67955
38587 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
38588 conservative for pt.null when flag_non_call_exceptions is on.
38590 2017-01-09 Jakub Jelinek <jakub@redhat.com>
38592 PR translation/79019
38593 PR translation/79020
38594 * params.def (PARAM_INLINE_MIN_SPEEDUP,
38595 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
38596 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
38598 * config/avr/avr.opt (maccumulate-args): Likewise.
38599 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
38600 * common.opt (freport-bug): Likewise.
38601 * cif-code.def (CIF_FINAL_ERROR): Likewise.
38602 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
38603 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
38604 translatable string.
38605 * config/i386/i386.c (function_value_32): Likewise.
38606 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
38607 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
38609 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
38610 * common/config/msp430/msp430-common.c (msp430_handle_option):
38612 * symtab.c (symtab_node::verify_base): Likewise.
38613 * opts.c (set_debug_level): Likewise.
38614 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
38615 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
38616 missing whitespace to translatable strings.
38617 * config/avr/avr.md (bswapsi2): Fix typo in comment.
38618 * config/sh/superh.h: Likewise.
38619 * config/i386/xopintrin.h: Likewise.
38620 * config/i386/znver1.md: Likewise.
38621 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
38622 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
38623 * double-int.h (struct double_int): Likewise.
38624 * double-int.c (div_and_round_double): Likewise.
38625 * wide-int.cc: Likewise.
38626 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
38627 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
38628 * cfgcleanup.c (crossjumps_occured): Renamed to ...
38629 (crossjumps_occurred): ... this.
38630 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
38633 PR tree-optimization/78899
38634 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
38635 returning bool return struct loop *, NULL for failure and the new
38637 (versionable_outer_loop_p): Don't version outer loop if it has
38638 dont_vectorized bit set.
38639 (tree_if_conversion): When versioning outer loop, ensure
38640 tree_if_conversion is performed also on the inner loop of the
38641 non-vectorizable outer loop copy.
38642 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
38643 LOOP_VECTORIZED in inner loop of the scalar outer loop and
38644 prevent vectorization of it.
38645 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
38646 the outer loop vectorization of the non-scalar version is attempted
38647 before vectorization of the inner loop in scalar version. If
38648 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
38649 vectorization of its inner loop.
38650 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
38651 has 2 inner loops, rename also on edges from bb whose single pred
38652 is outer_loop->header. Fix typo in function comment.
38654 2017-01-09 Martin Sebor <msebor@redhat.com>
38657 * asan.c (asan_emit_stack_protection): Increase local buffer size
38658 to avoid snprintf truncation warning.
38660 2017-01-09 Andrew Pinski <apinski@cavium.com>
38662 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
38663 to reference thunderx2t99 for the tuning structure
38664 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
38666 (thunderx2t99_extra_costs): This.
38667 * config/aarch64/aarch64-tune.md: Regenerate.
38668 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
38669 (vulcan_addrcost_table): This.
38670 (vulcan_regmove_cost): Rename to ...
38671 (thunderx2t99_regmove_cost): This.
38672 (vulcan_vector_cost): Rename to ...
38673 (thunderx2t99_vector_cost): this.
38674 (vulcan_branch_cost): Rename to ...
38675 (thunderx2t99_branch_cost): This.
38676 (vulcan_tunings): Rename to ...
38677 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
38678 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
38680 2017-01-09 Martin Jambor <mjambor@suse.cz>
38684 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
38685 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
38686 (propagate_vr_accross_jump_function): Use the above function for all
38687 value range computations for pass-through jump functions and type
38688 converasion from explicit value range values.
38689 (ipcp_propagate_stage): Do not attempt to deduce types of formal
38690 parameters from TYPE_ARG_TYPES.
38691 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
38692 (ipa_write_node_info): Stream type of the actual argument.
38693 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
38695 2017-01-09 Martin Liska <mliska@suse.cz>
38698 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
38699 (lookup_compiler): Do not show error message with have_E.
38701 2017-01-09 Jakub Jelinek <jakub@redhat.com>
38703 PR tree-optimization/78938
38704 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
38705 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
38706 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
38707 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
38710 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38712 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
38715 2017-01-09 Richard Biener <rguenther@suse.de>
38717 PR tree-optimization/78997
38718 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
38719 name condition properly.
38721 2017-01-09 Richard Biener <rguenther@suse.de>
38724 * dwarf2out.c (is_cxx): New overload with context.
38725 (is_naming_typedef_decl): Use it.
38727 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38729 * invoke.texi (Option Summary): Correct spacing in option lists
38730 and add line breaks to fix over-long lines.
38732 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38734 PR middle-end/17660
38736 * extend.texi (Common Variable Attributes): Add xref to GCC
38737 Internals manual to explain mode attribute keywords.
38739 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38742 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
38743 and Preprocessor Options.
38744 (Options for Linking): Document -pthread here....
38745 (RS/6000 and PowerPC Options): ...not here.
38746 (Solaris 2 Options): ...or here.
38747 * doc/cppopts.texi: Document -pthread.
38749 2017-01-08 Martin Sebor <msebor@redhat.com>
38751 PR middle-end/77708
38752 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
38753 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
38754 New member functions.
38755 (format_directive): Used them.
38757 (pass_sprintf_length::handle_gimple_call): Same.
38758 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
38759 to avoid truncation for any argument.
38760 (extract_affine_mul): Same.
38761 * tree.c (get_file_function_name): Same.
38763 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38765 PR middle-end/77484
38766 * predict.def (PRED_INDIR_CALL): Set to 86.
38768 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38770 PR preprocessor/54124
38771 * doc/cppopts.texi: Reformat -d subtable to list the full name
38772 of the options. Add cross-reference to the docs for the general
38773 compiler -d options.
38774 * doc/invoke.texi (Developer Options): Add cross-reference to the
38775 preprocessor-specific -d option documentation.
38777 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38779 PR preprocessor/13498
38780 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
38781 redudant material, and reflect new command-line options.
38782 (System Headers): Likewise.
38784 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38786 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
38787 -isystem, and -idirafter. Copy-edit.
38788 * doc/cppopts.texi: Copy-edit. Remove contradiction about
38789 default for -ftrack-macro-expansion. Delete obsolete and
38790 badly-formatted implementation details about -fdebug-cpp output.
38791 * doc/cppwarnopts.texi: Copy-edit.
38793 2017-01-07 David Malcolm <dmalcolm@redhat.com>
38796 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
38797 that the transition from a max line width >= 1<<10 to narrower
38798 lines works correctly.
38800 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
38802 * doc/options.texi (PerFunction): New.
38803 * opt-functions.awk (switch_flags): Map both Optimization and
38804 PerFunction to CL_OPTIMIZATION.
38805 * opth-gen.awk: Test for PerFunction flag along with
38807 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
38808 it only when the latter is present. Skip those that don't in
38809 the hash function generator.
38810 * common.opt (fvar-tracking): Mark as PerFunction instead of
38812 (fvar-tracking-assignments): Likewise.
38813 (fvar-tracking-assignments-toggle): Likewise.
38814 (fvar-tracking-uninit): Likewise.
38816 2017-01-07 Jakub Jelinek <jakub@redhat.com>
38818 PR translation/79018
38819 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
38822 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
38825 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
38826 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
38827 TARGET_LONG_JUMP_TABLE_OFFSETS.
38828 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
38829 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
38830 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
38831 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
38832 * config/m68k/m68k.md (tablejump expander): Likewise.
38833 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
38834 TARGET_LONG_JUMP_TABLE_OFFSETS.
38835 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
38836 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
38838 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
38839 David Holsgrove <david.holsgrove@xilinx.com>
38841 * common/config/microblaze/microblaze-common.c
38842 (TARGET_EXCEPT_UNWIND_INFO): Remove.
38843 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
38845 * config/microblaze/microblaze.c (microblaze_must_save_register)
38846 (microblaze_expand_epilogue, microblaze_return_addr): Handle
38848 (microblaze_eh_return): New function.
38849 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
38850 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
38851 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
38852 * config/microblaze/microblaze.md (eh_return): New pattern.
38854 2017-01-06 Jakub Jelinek <jakub@redhat.com>
38856 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
38857 GCC_DIAGNOSTIC_STRINGIFY): Define.
38859 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
38861 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38863 * config/arm/arm.md (<mcrr>): New.
38865 * config/arm/arm.c (arm_arch5te): New.
38866 (arm_option_override): Set arm_arch5te.
38867 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
38869 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
38870 (arm_mcrr_qualifiers): ... this. New.
38871 (MRRC_QUALIFIERS): Define to...
38872 (arm_mrrc_qualifiers): ... this. New.
38873 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
38875 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
38876 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
38877 (MRRCI, mrrc, MRRC): New.
38878 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
38879 VUNSPEC_MRRC2): New.
38881 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38883 * config/arm/arm.md (<mcr>): New.
38885 * config/arm/arm.c (arm_coproc_builtin_available): Add
38886 support for mcr, mrc, mcr2 and mrc2.
38887 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
38888 (arm_mcr_qualifiers): ... this. New.
38889 (MRC_QUALIFIERS): Define to ...
38890 (arm_mrc_qualifiers): ... this. New.
38891 (MCR_QUALIFIERS): Define to ...
38892 (arm_mcr_qualifiers): ... this. New.
38893 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
38895 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
38896 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
38897 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
38898 VUNSPEC_MRC2): New.
38900 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38902 * config/arm/arm.md (*ldc): New.
38906 * config/arm/arm.c (arm_coproc_builtin_available): Add
38907 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
38908 (arm_coproc_ldc_stc_legitimate_address): New.
38909 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
38910 'qualifier_const_pointer'.
38911 (LDC_QUALIFIERS): Define to...
38912 (arm_ldc_qualifiers): ... this. New.
38913 (STC_QUALIFIERS): Define to...
38914 (arm_stc_qualifiers): ... this. New.
38915 * config/arm/arm-protos.h
38916 (arm_coproc_ldc_stc_legitimate_address): New.
38917 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
38918 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
38919 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
38920 stc2, stcl, stc2l): New.
38921 * config/arm/constraints.md (Uz): New.
38922 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
38923 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
38924 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
38925 VUNSPEC_STC2L): New.
38927 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38929 * config/arm/arm.md (<cdp>): New.
38930 * config/arm/arm.c (neon_const_bounds): Rename this ...
38931 (arm_const_bounds): ... this.
38932 (arm_coproc_builtin_available): New.
38933 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
38934 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
38935 (CDP_QUALIFIERS): Define to...
38936 (arm_cdp_qualifiers): ... this. New.
38938 (arm_expand_builtin_args): Add case for 6 arguments.
38939 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
38940 (arm_const_bounds): ... this.
38941 (arm_coproc_builtin_available): New.
38942 * config/arm/arm_acle.h (__arm_cdp): New.
38944 * config/arm/arm_acle_builtins.def (cdp): New.
38946 * config/arm/iterators.md (CDPI,CDP,cdp): New.
38947 * config/arm/neon.md: Rename all 'neon_const_bounds' to
38948 'arm_const_bounds'.
38949 * config/arm/types.md (coproc): New.
38950 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
38951 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
38952 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
38953 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
38955 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38957 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
38958 (UBINOP_QUALIFIERS): New.
38960 (acle_builtin_data): New. Change comment.
38961 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
38962 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
38963 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
38964 arm_acle_builtins.def.
38965 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
38966 (arm_init_acle_builtins): New.
38967 (CRC32_BUILTIN): Remove.
38968 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
38969 crc32cb, crc32ch and crc32cw.
38970 (arm_init_crc32_builtins): Remove.
38971 (arm_init_builtins): Use arm_init_acle_builtins rather
38972 than arm_init_crc32_builtins.
38973 (arm_expand_acle_builtin): New.
38974 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
38975 * config/arm/arm_acle_builtins.def: New.
38977 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38979 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
38980 (arm_builtin_datum): ... this.
38981 (arm_init_neon_builtin): Rename to ...
38982 (arm_init_builtin): ... this. Add a new parameters PREFIX
38983 and USE_SIG_IN_NAME.
38984 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
38985 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
38986 'arm_builtin_datum'.
38987 (arm_init_vfp_builtins): Likewise.
38988 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
38989 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
38990 (arm_expand_neon_args): Rename to ...
38991 (arm_expand_builtin_args): ... this. Rename builtin_arg
38992 enum values and differentiate between ARG_BUILTIN_MEMORY
38993 and ARG_BUILTIN_NEON_MEMORY.
38994 (arm_expand_neon_builtin_1): Rename to ...
38995 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
38996 values, arm_expand_builtin_args and add bool parameter NEON.
38997 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
38998 (arm_expand_vfp_builtin): Likewise.
38999 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
39001 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39003 PR middle-end/77484
39004 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
39005 * predict.c (tree_estimate_probability_bb): Reverse direction of
39006 polymorphic call predictor.
39008 2017-01-06 David Malcolm <dmalcolm@redhat.com>
39010 * passes.c (execute_one_pass): Split out pass-skipping logic into...
39011 (determine_pass_name_match): ...this new function and...
39012 (should_skip_pass_p): ...this new function.
39014 2017-01-06 Nathan Sidwell <nathan@acm.org>
39016 * ipa-visibility.c (function_and_variable_visibility): Reformat
39017 comments and long lines. Remove extrneous if.
39018 * symtab.c (symtab_node::make_decl_local): Fix code format.
39019 (symtab_node::set_section_for_node): Fix comment typo.
39021 2017-01-06 Martin Liska <mliska@suse.cz>
39024 * lra-constraints.c: Rename invariant to lra_invariant.
39025 * predict.c (set_even_probabilities): Initialize e to NULL.
39027 2017-01-05 Martin Sebor <msebor@redhat.com>
39029 PR tree-optimization/78910
39030 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
39031 (format_integer): Correct off-by-one error in the handling
39032 of precision with negative numbers in signed conversions..
39034 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
39036 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
39038 2017-01-05 Jakub Jelinek <jakub@redhat.com>
39040 PR tree-optimization/71016
39041 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
39042 factor_out_conditional_conversion. Formatting fix.
39043 (factor_out_conditional_conversion): Add cond_stmt argument.
39044 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
39045 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
39048 2017-01-05 David Malcolm <dmalcolm@redhat.com>
39050 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
39051 read-rtl-function.o, and selftest-rtl.o.
39052 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
39053 (selftest::aarch64_test_loading_full_dump): New function.
39054 (selftest::aarch64_run_selftests): New function.
39055 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
39056 selftest::aarch64_run_selftests.
39057 * config/i386/i386.c
39058 (selftest::ix86_test_loading_dump_fragment_1): New function.
39059 (selftest::ix86_test_loading_call_insn): New function.
39060 (selftest::ix86_test_loading_full_dump): New function.
39061 (selftest::ix86_test_loading_unspec): New function.
39062 (selftest::ix86_run_selftests): Call the new functions.
39063 * emit-rtl.c (maybe_set_max_label_num): New function.
39064 * emit-rtl.h (maybe_set_max_label_num): New decl.
39065 * function.c (instantiate_decls): Guard call to
39066 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
39067 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
39069 * gensupport.c (gen_reader::gen_reader): Pass "false"
39070 for new "compact" param of rtx_reader.
39071 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
39072 rather than an empty string for NULL strings.
39073 * read-md.c: Potentially include config.h rather than bconfig.h.
39074 Wrap include of errors.h with #ifdef GENERATOR_FILE.
39075 (have_error): New global, copied from errors.c.
39076 (md_reader::read_name): Rename to...
39077 (md_reader::read_name_1): ...this, adding "out_loc" param,
39078 and converting "missing name or number" to returning false, rather
39080 (md_reader::read_name): Reimplement in terms of read_name_1.
39081 (md_reader::read_name_or_nil): New function.
39082 (md_reader::read_string): Handle "(nil)" by returning NULL.
39083 (md_reader::md_reader): Add new param "compact".
39084 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
39085 (md_reader::read_file): New method.
39086 * read-md.h (md_reader::md_reader): Add new param "compact".
39087 (md_reader::read_file): New method.
39088 (md_reader::is_compact): New accessor.
39089 (md_reader::read_name): Convert return type from void to file_location.
39090 (md_reader::read_name_or_nil): New decl.
39091 (md_reader::read_name_1): New decl.
39092 (md_reader::m_compact): New field.
39093 (noop_reader::noop_reader): Pass "false" for new "compact" param
39095 (rtx_reader::rtx_reader): Add new "compact" param.
39096 (rtx_reader::read_rtx_operand): Make virtual and convert return
39097 type from void to rtx.
39098 (rtx_reader::read_until): New decl.
39099 (rtx_reader::handle_any_trailing_information): New virtual function.
39100 (rtx_reader::postprocess): New virtual function.
39101 (rtx_reader::finalize_string): New virtual function.
39102 (rtx_reader::m_in_call_function_usage): New field.
39103 (rtx_reader::m_reuse_rtx_by_id): New field.
39104 * read-rtl-function.c: New file.
39105 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
39106 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
39107 (selftest::verify_three_block_rtl_cfg): New decl.
39108 * read-rtl-function.h: New file.
39109 * read-rtl.c: Potentially include config.h rather than bconfig.h.
39110 For host, include function.h, memmodel.h, and emit-rtl.h.
39111 (one_time_initialization): New function.
39112 (struct compact_insn_name): New struct.
39113 (compact_insn_names): New array.
39114 (find_code): Handle insn codes in compact dumps.
39115 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
39116 (bind_subst_iter_and_attr): Likewise.
39117 (add_condition_to_string): Likewise.
39118 (add_condition_to_rtx): Likewise.
39119 (apply_attribute_uses): Likewise.
39120 (add_current_iterators): Likewise.
39121 (apply_iterators): Likewise.
39122 (initialize_iterators): Guard usage of apply_subst_iterator with
39123 #ifdef GENERATOR_FILE.
39124 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
39125 (md_reader::read_mapping): Likewise.
39126 (add_define_attr_for_define_subst): Likewise.
39127 (add_define_subst_attr): Likewise.
39128 (read_subst_mapping): Likewise.
39129 (check_code_iterator): Likewise.
39130 (rtx_reader::read_rtx): Likewise. Move one-time initialization
39132 (one_time_initialization): New function.
39133 (rtx_reader::read_until): New method.
39134 (read_flags): New function.
39135 (parse_reg_note_name): New function.
39136 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
39137 Handle reuse_rtx ids.
39138 Wrap iterator lookup within #ifdef GENERATOR_FILE.
39139 Add parsing support for RTL dumps, mirroring the special-cases in
39140 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
39141 values, and calling handle_any_trailing_information.
39142 (rtx_reader::read_rtx_operand): Convert return type from void
39143 to rtx, returning return_rtx. Handle case 'e'. Call
39144 finalize_string on XSTR and XTMPL fields.
39145 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
39146 "(nil)" values were omitted. Call the postprocess vfunc on the
39148 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
39149 class ctor. Initialize m_in_call_function_usage. Call
39150 one_time_initialization.
39151 * rtl-tests.c (selftest::test_uncond_jump): Call
39152 set_new_first_and_last_insn.
39153 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
39154 * selftest-rtl.c: New file.
39155 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
39156 (selftest::get_insn_by_uid): New decl.
39157 * selftest-run-tests.c (selftest::run_tests): Call
39158 read_rtl_function_c_tests.
39159 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
39160 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
39163 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
39165 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
39166 operands in a special way. Assert that pos+len <= mode precision.
39168 2017-01-05 Jakub Jelinek <jakub@redhat.com>
39170 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
39171 3 argument Alias with unlimited for the negative form.
39172 (fno-vect-cost-model): Removed.
39174 2017-01-05 Martin Liska <mliska@suse.cz>
39176 * hsa-gen.c (gen_hsa_divmod): New function.
39177 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
39179 2017-01-05 Martin Liska <mliska@suse.cz>
39182 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
39185 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39187 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
39188 small constant length operands.
39190 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39192 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
39193 between loop iterations.
39195 2017-01-05 Martin Liska <mliska@suse.cz>
39198 * gimplify.c (gimplify_decl_expr): Compare to
39199 asan_poisoned_variables instread of checking flags.
39200 (gimplify_target_expr): Likewise.
39201 (gimplify_expr): Likewise.
39202 (gimplify_function_tree): Conditionally initialize
39203 asan_poisoned_variables.
39205 2017-01-04 Jeff Law <law@redhat.com>
39207 PR tree-optimizatin/78812
39208 * rtl.h (contains_mem_rtx_p): Prototype.
39209 * ifcvt.c (containts_mem_rtx_p): Move from here to...
39210 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
39211 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
39212 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
39213 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
39215 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39217 * input.c (assert_char_at_range): Default-initialize actual_range.
39219 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39221 * df-scan.c (df_ref_create_structure): Make regno unsigned,
39222 to match the caller.
39224 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39226 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
39227 insns after final jump in test to emit dummy move.
39229 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39231 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
39232 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
39234 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39236 * multiple_target.c (create_dispatcher_calls): Init e_next.
39237 * tree-ssa-loop-split.c (split_loop): Init border.
39238 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
39241 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
39246 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
39247 (altivec_register_operand): Do not return true if the operand
39248 contains a SUBREG mixing SImode and SFmode.
39249 (vsx_register_operand): Likewise.
39250 (vsx_reg_sfsubreg_ok): New predicate.
39251 (vfloat_operand): Do not return true if the operand contains a
39252 SUBREG mixing SImode and SFmode.
39253 (vint_operand): Likewise.
39254 (vlogical_operand): Likewise.
39255 (gpc_reg_operand): Likewise.
39256 (int_reg_operand): Likewise.
39257 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
39258 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
39259 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
39261 (rs6000_emit_move_si_sf_subreg): New helper function.
39262 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
39263 fixup SUBREGs involving SImode and SFmode.
39264 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
39265 numbers for the new peephole2 optimization.
39266 (peephole2 for SFmode unions): New peephole2 to optimize cases in
39267 the GLIBC math library that do AND/IOR/XOR operations on single
39268 precision floating point.
39269 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
39270 target macros to say whether we need to avoid SUBREGs mixing
39272 (TARGET_ALLOW_SF_SUBREG): Likewise.
39273 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
39274 (UNSPEC_SI_FROM_SF): Likewise.
39275 (iorxor): Change spacing.
39276 (and_ior_xor): New iterator for AND, IOR, and XOR.
39277 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
39278 (movdi_from_sf_zero_ext): Likewise.
39279 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
39280 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
39281 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
39282 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
39283 (fms<mode>4): Likewise.
39284 (fnma<mode>4): Likewise.
39285 (fnms<mode>4): Likewise.
39286 (nfma<mode>4): Likewise.
39287 (nfms<mode>4): Likewise.
39289 2017-01-04 Marek Polacek <polacek@redhat.com>
39292 * doc/invoke.texi: Document -Wpointer-compare.
39294 2017-01-04 Jakub Jelinek <jakub@redhat.com>
39296 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
39299 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
39300 descriptions for -gdwarf-5 and emit them as uleb128 instead of
39303 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
39306 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
39307 documentation of the powerpc_popcntb_ok attribute.
39308 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
39309 code to issue warning messages if a requested CPU configuration is
39310 not supported by the binary (assembler and loader) toolchain.
39311 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
39312 made to define a built-in function that has been disabled.
39313 (paired_init_builtins): Add assertion to prevent ICE if attempt is
39314 made to define a built-in function that has been disabled.
39315 (altivec_init_builtins): Add comment explaining why definition
39316 of the DST built-in functions is not preceded by an assertion
39317 check. Add assertions to prevent ICE if attempts are made to
39318 define an altivec predicate or an abs* built-in function that has
39320 (htm_init_builtins): Add comment explaining why definition of the
39321 htm built-in functions is not preceded by an assertion check.
39323 2017-01-04 Jeff Law <law@redhat.com>
39325 PR tree-optimizatin/67955
39326 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
39327 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
39328 the points-to solution does not include pt_null. Use DECL_PT_UID
39331 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
39333 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
39334 Use gen_int_mode instead of gen_lopwart for const_int operands.
39336 2017-01-04 Jakub Jelinek <jakub@redhat.com>
39338 PR tree-optimization/71563
39339 * match.pd: Simplify X << Y into X if Y is known to be 0 or
39340 out of range value - has low bits known to be zero.
39342 2017-01-04 Alan Modra <amodra@gmail.com>
39344 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
39345 * configure: Regenerate.
39346 * config.in: Regenerate.
39348 2017-01-04 Jakub Jelinek <jakub@redhat.com>
39351 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
39352 a substring of the message, but strcmp with the whole message. Ifdef
39353 ENABLE_NLS, translate the message first using dgettext.
39355 2017-01-03 Jeff Law <law@redhat.com>
39357 PR tree-optimizatin/78856
39358 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
39359 (mark_threaded_blocks): Remove code to truncate thread paths that
39360 cross multiple loop headers. Instead invalidate the cached loop
39361 iteration information and handle case of a thread path walking
39362 into an irreducible region.
39364 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
39367 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
39368 assertions. Add support for doing the signbit if the IEEE 128-bit
39369 floating point value is in a GPR.
39370 * config/rs6000/rs6000.md (Fsignbit): Delete.
39371 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
39372 Update the length attribute if the value is in a GPR.
39373 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
39374 the sign or zero extension instruction, since the value is always 0/1.
39375 (signbit<mode>2_dm2): Delete using <Fsignbit>.
39378 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
39379 extracting SImode to a GPR register so that we can generate a
39380 store, limit the vector to be in a traditional Altivec register
39381 for the vextuwrx instruction.
39383 2017-01-03 Ian Lance Taylor <iant@google.com>
39385 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
39387 2017-01-03 Martin Sebor <msebor@redhat.com>
39389 PR tree-optimization/78696
39390 * gimple-ssa-sprintf.c (format_floating): Correct handling of
39391 precision. Use MPFR for %f for greater fidelity. Correct handling
39393 (pass_sprintf_length::compute_format_length): Set width and precision
39394 specified by asrerisk to void_node for vararg functions.
39395 (try_substitute_return_value): Adjust dump output.
39397 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
39399 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
39401 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
39403 * doc/invoke.texi (SPARC options): Document -mlra as the default.
39404 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
39405 -mlra/-mno-lra was passed to the compiler.
39407 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
39409 PR rtl-optimization/65618
39410 * emit-rtl.c (try_split): Move initialization of "before" and
39411 "after" to just before the call to emit_insn_after_setloc.
39413 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
39415 * doc/md.texi (Standard Names): Remove reference to Java frontend.
39417 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
39419 * dwarf2out.c (gen_enumeration_type_die): When
39420 -gno-strict-dwarf, add a DW_AT_encoding attribute.
39422 2017-01-03 Jakub Jelinek <jakub@redhat.com>
39424 PR tree-optimization/78965
39425 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
39426 Change first argument from const call_info & to call_info &. For %n
39427 set info.nowrite to false.
39429 PR middle-end/78901
39430 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
39431 possibly throwing calls.
39433 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
39434 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
39435 and fns handling, rather than in a separate case SSA_NAME.
39437 2017-01-02 Jeff Law <law@redhat.com>
39439 * config/darwin-driver.c (darwin_driver_init): Const-correctness
39440 fixes for first_period and second_period variables.
39442 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
39445 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
39446 (*insvqi_1): New insn pattern.
39447 (*insvqi_1_mem_rex64): Ditto.
39448 (*insvqi_2): Ditto.
39449 (*insvqi_3): Rename from *insvqi.
39451 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
39453 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
39455 * doc/cfg.texi (Edges): Remove reference to Java.
39456 (Maintaining the CFG): Ditto.
39458 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39460 PR middle-end/77674
39461 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
39462 transparent aliases.
39464 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39466 PR middle-end/77484
39467 * predict.def (PRED_CALL): Update hitrate.
39468 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
39469 * predict.c (tree_estimate_probability_bb): Split CALL predictor
39470 into direct/indirect/polymorphic variants.
39472 2017-01-01 Jakub Jelinek <jakub@redhat.com>
39474 Update copyright years.
39476 * gcc.c (process_command): Update copyright notice dates.
39477 * gcov-dump.c (print_version): Ditto.
39478 * gcov.c (print_version): Ditto.
39479 * gcov-tool.c (print_version): Ditto.
39480 * gengtype.c (create_file): Ditto.
39481 * doc/cpp.texi: Bump @copying's copyright year.
39482 * doc/cppinternals.texi: Ditto.
39483 * doc/gcc.texi: Ditto.
39484 * doc/gccint.texi: Ditto.
39485 * doc/gcov.texi: Ditto.
39486 * doc/install.texi: Ditto.
39487 * doc/invoke.texi: Ditto.
39489 Copyright (C) 2017 Free Software Foundation, Inc.
39491 Copying and distribution of this file, with or without modification,
39492 are permitted in any medium without royalty provided the copyright
39493 notice and this notice are preserved.