1 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
3 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
6 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
8 * doc/invoke.texi: Fix some typos.
10 2017-12-16 Martin Sebor <msebor@redhat.com>
12 PR tree-optimization/78918
13 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
14 * builtins.c (check_sizes): Rename...
15 (check_access): ...to this. Rename function arguments for clarity.
16 (check_memop_sizes): Adjust names.
17 (expand_builtin_memchr, expand_builtin_memcpy): Same.
18 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
19 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
20 (check_strncat_sizes, expand_builtin_strncat): Same.
21 (expand_builtin_strncpy, expand_builtin_memset): Same.
22 (expand_builtin_bzero, expand_builtin_memcmp): Same.
23 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
24 (maybe_emit_sprintf_chk_warning): Same.
25 (expand_builtin_strcpy): Adjust.
26 (expand_builtin_stpcpy): Same.
27 (expand_builtin_with_bounds): Detect out-of-bounds accesses
28 in pointer-checking forms of memcpy, memmove, and mempcpy.
29 (gcall_to_tree_minimal, max_object_size): Define new functions.
30 * builtins.h (max_object_size): Declare.
31 * calls.c (alloc_max_size): Call max_object_size instead of
32 hardcoding ssizetype limit.
33 (get_size_range): Handle new argument.
34 * calls.h (get_size_range): Add a new argument.
35 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
36 * doc/invoke.texi (-Wrestrict): Adjust, add example.
37 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
39 (gimple_fold_builtin_memory_chk): Same.
40 (gimple_fold_builtin_stxcpy_chk): New function.
41 * gimple-ssa-warn-restrict.c: New source.
42 * gimple-ssa-warn-restrict.h: New header.
43 * gimple.c (gimple_build_call_from_tree): Propagate location.
44 * passes.def (pass_warn_restrict): Add new pass.
45 * tree-pass.h (make_pass_warn_restrict): Declare.
46 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
48 (handle_builtin_strcat): Same.
49 (strlen_optimize_stmt): Rename...
50 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
51 stpncpy, strncpy, and their checking forms.
53 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
55 PR rtl-optimization/82849
56 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
57 and get_max_loop_iterations_int.
59 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
60 Alan Hayward <alan.hayward@arm.com>
61 David Sherwood <david.sherwood@arm.com>
63 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
64 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
65 size as a poly_uint64.
66 (mode_for_vector, mode_for_int_vector): Take the number of vector
67 elements as a poly_uint64.
68 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
69 size as a poly_uint64.
70 (mode_for_vector, mode_for_int_vector): Take the number of vector
71 elements as a poly_uint64.
73 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
74 Alan Hayward <alan.hayward@arm.com>
75 David Sherwood <david.sherwood@arm.com>
77 * machmode.h (MACRO_MODE): New macro.
78 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
79 * targhooks.c (default_libcall_value, default_secondary_reload)
80 (default_memory_move_cost, default_register_move_cost)
81 (default_class_max_nregs): Likewise.
83 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
84 Alan Hayward <alan.hayward@arm.com>
85 David Sherwood <david.sherwood@arm.com>
87 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
88 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
90 * genautomata.c (main): Likewise.
91 * genconditions.c (write_header): Likewise.
92 * genemit.c (main): Likewise.
93 * genextract.c (print_header): Likewise.
94 * genopinit.c (main): Likewise.
95 * genoutput.c (output_prologue): Likewise.
96 * genpeep.c (main): Likewise.
97 * genpreds.c (write_insn_preds_c): Likewise.
98 * genrecog.c (writer_header): Likewise.
99 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
100 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
101 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
102 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
103 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
104 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
105 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
106 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
107 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
108 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
109 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
110 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
111 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
112 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
113 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
114 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
115 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
116 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
117 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
118 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
119 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
120 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
121 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
122 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
123 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
124 * config/darwin.c (IN_TARGET_CODE): Likewise.
125 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
126 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
127 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
128 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
129 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
130 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
131 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
132 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
133 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
134 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
135 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
136 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
137 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
138 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
139 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
140 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
141 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
142 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
143 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
144 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
145 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
146 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
147 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
148 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
149 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
150 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
151 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
152 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
153 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
154 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
155 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
156 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
157 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
158 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
159 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
160 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
161 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
162 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
163 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
164 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
165 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
166 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
167 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
168 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
169 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
170 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
171 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
172 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
173 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
174 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
175 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
176 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
177 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
178 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
179 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
180 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
181 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
182 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
183 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
184 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
185 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
186 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
187 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
188 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
189 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
190 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
191 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
192 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
193 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
194 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
195 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
196 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
197 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
198 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
199 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
200 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
201 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
202 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
203 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
204 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
205 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
206 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
207 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
208 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
209 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
210 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
211 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
212 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
213 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
214 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
215 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
216 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
217 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
218 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
219 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
220 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
221 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
222 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
223 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
224 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
225 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
226 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
227 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
228 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
229 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
230 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
232 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
233 Alan Hayward <alan.hayward@arm.com>
234 David Sherwood <david.sherwood@arm.com>
236 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
239 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
240 Alan Hayward <alan.hayward@arm.com>
241 David Sherwood <david.sherwood@arm.com>
243 * doc/generic.texi (VEC_SERIES_EXPR): Document.
244 * doc/md.texi (vec_series@var{m}): Document.
245 * tree.def (VEC_SERIES_EXPR): New tree code.
246 * tree.h (build_vec_series): Declare.
247 * tree.c (build_vec_series): New function.
248 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
249 * tree-pretty-print.c (dump_generic_node): Likewise.
250 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
251 * tree-inline.c (estimate_operator_cost): Likewise.
252 * expr.c (expand_expr_real_2): Likewise.
253 * optabs-tree.c (optab_for_tree_code): Likewise.
254 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
255 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
256 * expmed.c (make_tree): Handle VEC_SERIES.
257 * optabs.def (vec_series_optab): New optab.
258 * optabs.h (expand_vec_series_expr): Declare.
259 * optabs.c (expand_vec_series_expr): New function.
260 * tree-vect-generic.c (expand_vector_operations_1): Check that
261 the operands also have vector type.
263 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
264 Alan Hayward <alan.hawyard@arm.com>
265 David Sherwood <david.sherwood@arm.com>
267 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
268 (VEC_COND_EXPR): Add missing @tindex.
269 * doc/md.texi (vec_duplicate@var{m}): Document.
270 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
271 * tree.c (build_vector_from_val): Add stubbed-out handling of
272 variable-length vectors, using VEC_DUPLICATE_EXPR.
273 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
274 * cfgexpand.c (expand_debug_expr): Likewise.
275 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
276 * tree-inline.c (estimate_operator_cost): Likewise.
277 * tree-pretty-print.c (dump_generic_node): Likewise.
278 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
279 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
280 (test_vec_duplicate_folding): New function.
281 (fold_const_c_tests): Call it.
282 * optabs.def (vec_duplicate_optab): New optab.
283 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
284 * optabs.h (expand_vector_broadcast): Declare.
285 * optabs.c (expand_vector_broadcast): Make non-static. Try using
287 * expr.c (store_constructor): Try using vec_duplicate_optab for
289 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
291 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
294 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
295 div/mod latencies a bit.
297 2017-12-15 Jeff Law <law@redhat.com>
299 PR tree-optimization/36550
300 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
301 (mark_threaded_blocks): Rewrite code to avoid block copying when
302 optimizing for size. Don't pessimize blocks which will be
303 copied, but all the statements will be dead.
305 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
307 PR tree-optimization/81165
308 * tree-ssa-threadupdate.c (uses_in_bb): New.
309 (estimate_threading_killed_stmts): New.
310 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
311 * tree-ssa-threadedge.c
312 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
315 2017-12-15 Jeff Law <law@redhat.com>
317 PR tree-optimization/83410
318 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
319 threads when parallelizing loops.
321 2017-12-15 Jakub Jelinek <jakub@redhat.com>
323 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
325 * config/alpha/alpha.c (vms_attribute_table): Swap
326 affects_type_identity and handler fields, adjust comments.
327 * config/mips/mips.c (mips_attribute_table): Likewise.
328 * config/visium/visium.c (visium_attribute_table): Likewise.
329 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
330 * config/microblaze/microblaze.c (microblaze_attribute_table):
332 * config/spu/spu.c (spu_attribute_table): Likewise.
333 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
334 * config/arc/arc.c (arc_attribute_table): Likewise.
335 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
336 * config/v850/v850.c (v850_handle_interrupt_attribute,
337 v850_handle_data_area_attribute): Formatting fixes.
338 (v850_attribute_table): Swap affects_type_identity and handler
339 fields, adjust comments.
340 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
341 * config/arm/arm.c (arm_attribute_table): Likewise.
342 * config/avr/avr.c (avr_attribute_table): Likewise.
343 * config/s390/s390.c (s390_attribute_table): Likewise.
344 * config/sh/sh.c (sh_attribute_table): Likewise.
345 * config/i386/i386.c (ix86_handle_cconv_attribute,
346 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
347 (ix86_attribute_table): Swap affects_type_identity and handler
348 fields, adjust comments.
349 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
350 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
351 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
352 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
353 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
354 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
355 * config/rx/rx.c (rx_attribute_table): Likewise.
356 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
357 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
358 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
359 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
360 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
361 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
362 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
363 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
364 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
365 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
366 * doc/plugins.texi (user_attr): Likewise. Add NULL for
368 * attribs.c (empty_attribute_table): Swap affects_type_identity and
370 (register_scoped_attributes, decl_attributes): Formatting fixes.
372 PR tree-optimization/83269
373 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
374 subtraction in arg0's type if type is signed and arg0 is unsigned.
378 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
379 convert? on @0 instead of convert. Check type of @1, not @0.
380 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
381 convert? on @0 instead of convert. Check type of @1, not @0.
382 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
383 only optimize if either both @1 and @2 types are narrower
384 precision, or both are wider or equal precision, and in the former
385 case only if both have undefined overflow.
387 2017-12-15 Richard Biener <rguenther@suse.de>
390 * internal-fn.def (IFN_NOP): Add.
391 * internal-fn.c (expand_NOP): Do nothing.
392 * lto-streamer-in.c (input_function): Instead of removing
393 sanitizer calls replace them with IFN_NOP calls.
395 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
396 Alan Hayward <alan.hayward@arm.com>
397 David Sherwood <david.sherwood@arm.com>
399 * dse.c (store_info, read_info_type): Replace begin and end with
401 (print_range): New function.
402 (set_all_positions_unneeded, any_positions_needed_p)
403 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
405 (record_store): Likewise. Optimize the case in which all positions
407 (get_stored_val): Replace read_begin and read_end with read_offset
409 (replace_read): Update call accordingly.
411 2017-12-15 Bin Cheng <bin.cheng@arm.com>
413 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
414 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
415 (loop_cand::loop_cand): Initialize above members.
416 (loop_cand::supported_operations): Delete.
417 (loop_cand::can_interchange_p): Inline above function.
418 (loop_cand::classify_simple_reduction): Record number of constant
419 initialized simple reductions.
420 (should_interchange_loops): New parameters. Check stmt cost of loops
422 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
423 Update call to should_interchange_loops.
424 (should_interchange_loop_nest): Update call to
425 should_interchange_loops.
427 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
430 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
431 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
432 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
434 2017-12-15 Julia Koval <julia.koval@intel.com>
436 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
437 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
438 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
439 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
440 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
442 2017-12-15 Julia Koval <julia.koval@intel.com>
444 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
445 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
446 * config/i386/sse.md (vaesenc_<mode>): New pattern.
447 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
448 _mm_aesenc_epi128): New intrinsics.
450 2017-12-15 Julia Koval <julia.koval@intel.com>
452 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
453 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
454 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
455 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
456 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
458 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
460 * gimple-ssa-strength-reduction.c (analyze_increments):
461 Distinguish replacement costs for constant strides from those for
464 2017-12-14 Jakub Jelinek <jakub@redhat.com>
466 * var-tracking.c (variable_tracking_main_1): Formatting fix.
468 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
470 * doc/invoke.texi: Document -Wcast-function-type.
471 * recog.h (stored_funcptr): Change signature.
472 * tree-dump.c (dump_node): Avoid warning.
473 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
475 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
478 * gimple-fold.c (get_range_strlen): Add the handling of non-member
481 2017-12-14 David Malcolm <dmalcolm@redhat.com>
483 PR tree-optimization/83312
484 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
485 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
486 "val" param, and to cope with arbitrary basic blocks.
487 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
488 handle NULL_TREE for "val", dropping "bb" param.
489 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
490 drop "bb" param. Handle NULL_TREE for "val".
491 (find_case_label_for_value): Make "switch_stmt" param const.
492 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
494 (vrp_prop::check_all_array_refs): Reimplement as...
495 (check_array_bounds_dom_walker::before_dom_children): ...this new
496 vfunc. Replace linear search through BB block list, excluding
497 those with non-executable in-edges via dominator walk.
499 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
501 * config/arm/arm.opt (mverbose-cost-dump): New option.
502 * config/arm/arm.c (arm_rtx_costs): Use it.
504 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
507 * reload1.c (emit_input_reload_insns): Skip debug markers.
509 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
511 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
515 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
516 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
517 * config/sh/sh.c: Skip debug insns besides notes.
518 * config/sh/sh.md: Likewise.
519 * config/sh/sh_treg_combine.cc: Likewise.
520 * config/sh/sync.md: Likewise.
522 2017-12-14 Tom de Vries <tom@codesourcery.com>
524 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
525 Add item for weak_undefined.
527 2017-12-14 Richard Biener <rguenther@suse.de>
529 PR tree-optimization/67842
530 * sese.h (bb_in_region): Remove #if 0'ed code.
532 2017-12-14 Richard Biener <rguenther@suse.de>
534 PR tree-optimization/83326
535 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
536 may_be_zero parameter and handle it by not marking the first
537 peeled copy as not exiting the loop.
538 (try_peel_loop): Likewise.
539 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
540 to handle the case of constant or zero iterations and perform
541 loop header copying on-the-fly.
543 2017-12-14 Richard Biener <rguenther@suse.de>
545 PR tree-optimization/83418
546 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
547 Instead of asserting we don't get unfolded comparisons deal with
550 2017-12-14 Jakub Jelinek <jakub@redhat.com>
553 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
554 basic blocks. Assert debug bind insns don't appear outside of bbs,
555 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
558 PR tree-optimization/83198
559 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
560 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
561 value if arg is a REAL_CST with incompatible type.
563 2017-12-14 Sudakshina Das <sudi.das@arm.com>
564 Bin Cheng <bin.cheng@arm.com>
567 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
569 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
572 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
574 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
575 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
577 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
578 Alan Hayward <alan.hayward@arm.com>
579 David Sherwood <david.sherwood@arm.com>
581 * poly-int.h: New file.
582 * poly-int-types.h: Likewise.
583 * coretypes.h: Include them.
584 (POLY_INT_CONVERSION): Define.
585 * target.def (estimated_poly_value): New hook.
586 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
587 * doc/tm.texi: Regenerate.
588 * doc/poly-int.texi: New file.
589 * doc/gccint.texi: Include it.
590 * doc/rtl.texi: Describe restrictions on subreg modes.
591 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
592 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
593 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
594 * targhooks.h (default_estimated_poly_value): Declare.
595 * targhooks.c (default_estimated_poly_value): New function.
596 * target.h (estimated_poly_value): Likewise.
597 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
598 (wi::unary_traits): Delete.
599 (wi::binary_traits::signed_shift_result_type): Define for
600 offset_int << HOST_WIDE_INT, etc.
601 (generic_wide_int::operator <<=): Define for all types and use
602 wi::lshift instead of <<.
603 (wi::hwi_with_prec): Add a default constructor.
604 (wi::ints_for): New class.
605 (operator <<): Define for all wide-int types.
606 (operator /): New function.
607 (operator %): Likewise.
608 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
609 (ASSERT_MAYBE_NE_AT): New macros.
611 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
612 Dominik Vogt <vogt@linux.vnet.ibm.com>
615 * emit-rtl.c (init_emit): Remove ??? comment.
616 * explow.c (get_dynamic_stack_size): Take known alignment of stack
617 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
618 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
619 alignment of 3 virtual registers to BITS_PER_WORD.
621 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
623 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
625 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
626 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
628 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
631 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
633 2017-12-13 Jakub Jelinek <jakub@redhat.com>
635 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
636 with the exception of debug begin stmt markers appear before
640 * final.c (rest_of_handle_final): Call variable_tracking_main only
641 if !flag_var_tracking.
643 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
644 Jakub Jelinek <jakub@redhat.com>
648 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
649 labels debug stmts that can only appear after labels.
651 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
653 PR rtl-optimization/82398
654 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
655 EXPR_USEFULNESS in priority comparison.
657 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
659 PR rtl-optimization/83393
660 * combine.c (move_deaths): If reg_stat points to a too new insn in
661 last_death, do not use it: find the proper insn instead.
663 2017-12-12 Jeff Law <law@redhat.com>
665 PR tree-optimization/83298
666 PR tree-optimization/83362
667 PR tree-optimization/83383
668 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
669 push_value_range a public interface. Add new argument to
670 record_ranges_from_stmt.
671 * gimple-ssa-evrp-analyze.c
672 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
673 Update comments. Handle recording temporary equivalences.
674 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
675 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
676 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
677 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
678 gimple-ssa-evrp-analyze.h.
679 (record_temporary_equivalences_from_phis): Add new argument. When
680 the PHI arg is an SSA_NAME, set the result's range to the range
682 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
684 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
685 Pass it down to children as needed.
686 (thread_outgoing_edges): Likewise.
687 (thread_across_edge): Likewise. Push/pop range state as needed.
688 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
690 2017-12-12 Julia Koval <julia.koval@intel.com>
692 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
693 (PTA_CANNONLAKE): Remove PTA_CLWB.
695 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
698 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
699 (vcondv2div2df): Likewise.
700 (vconduv2dfv2di): Likewise.
702 2017-12-12 Jakub Jelinek <jakub@redhat.com>
704 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
705 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
706 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
707 ATTR_MATHFN_FPROUNDING_ERRNO.
709 2017-12-12 Richard Biener <rguenther@suse.de>
711 PR tree-optimization/83385
712 * graphite-scop-detection.c (get_order, order): Remove.
713 (bb_to_rpo): New global.
715 (build_scops): Sort pbbs in RPO order.
717 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
719 * combine.c (simplify_set): Do not transform subregs to zero_extends
720 if the destination mode is a vector mode.
722 2017-12-12 Jakub Jelinek <jakub@redhat.com>
724 PR tree-optimization/83359
725 * tree-cfg.h (fold_loop_internal_call): Declare.
726 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
727 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
728 (find_loop_dist_alias): New function.
729 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
730 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
733 PR tree-optimization/80631
734 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
735 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
736 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
737 hardcoding zero as the value if COND_EXPR is never true. For
738 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
739 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
740 hardcoding MAX_EXPR as the reduction operation.
741 (is_nonwrapping_integer_induction): Allow negative step.
742 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
743 vect_create_epilog_for_reduction, if no value is suitable, don't
744 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
746 2017-12-12 Richard Biener <rguenther@suse.de>
748 PR tree-optimization/81889
749 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
750 range info from the non-wrapping IV instead of just the range
753 2017-12-12 Julia Koval <julia.koval@intel.com>
755 * config.gcc: Add vaesintrin.h.
756 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
758 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
759 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
761 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
762 * config/i386/immintrin.h: Include vaesintrin.h.
763 * config/i386/sse.md (vaesdec_<mode>): New pattern.
764 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
765 _mm_aesdec_epi128): New intrinsics.
767 2017-12-12 Julia Koval <julia.koval@intel.com>
769 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
770 OPTION_MASK_ISA_VAES_UNSET): New.
771 (ix86_handle_option): Handle -mvaes.
772 * config/i386/cpuid.h: Define bit_VAES.
773 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
774 * config/i386/i386-c.c (__VAES__): New.
775 * config/i386/i386.c (ix86_target_string): Add -mvaes.
776 (ix86_valid_target_attribute_inner_p): Ditto.
777 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
778 * config/i386/i386.opt: Add -mvaes.
779 * doc/invoke.texi: Ditto.
781 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
783 * debug.h (gcc_debug_hooks): Add inline_entry.
784 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
785 * debug.c (do_nothing_debug_hooks): Likewise.
786 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
787 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
788 (dwarf2_lineno_debug_hooks): Likewise.
790 * common.opt (gstatement-frontiers): New, setting
791 debug_nonbind_markers_p.
792 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
793 * toplev.c (process_options): Autodetect value for debug statement
795 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
796 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
798 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
799 markers. Integrate source bind into debug stmt expand loop.
800 (pass_expand::execute): Check debug marker limit. Avoid deep
801 TER and expand debug locations for debug bind insns only.
802 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
804 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
805 * final.c (reemit_insn_block_notes): Take current block from
806 nonbind markers. Declare note where it's first set.
807 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
808 begin stmt markers if enabled.
809 (notice_source_line): Handle nonbind markers. Fail if their
810 location is unknown or that of builtins.
811 (rest_of_handle_final): Convert begin stmt markers to notes if
812 var-tracking didn't run.
813 (rest_of_clean_state): Skip begin stmt markers.
814 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
816 * function.c (allocate_struct_function): Set begin_stmt_markers.
817 * function.h (struct function): Add debug_marker_count counter
818 and debug_nonbind_markers flag.
819 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
820 * gimple-low.c (lower_function_body): Adjust
821 debug_nonbind_markers.
822 (lower_stmt): Drop or skip gimple debug stmts.
823 (lower_try_catch): Skip debug stmts.
824 * gimple.c (gimple_build_debug_begin_stmt): New.
825 (gimple_copy): Increment debug_marker_count if copying one.
826 * gimple.h (gimple_build_debug_begin_stmt): Declare.
827 * gimplify.c (rexpr_location): New.
828 (rexpr_has_location): New.
829 (warn_switch_unreachable_r): Handle gimple debug stmts.
830 (shortcut_cond_r): Call expr_location.
832 (find_goto_label): New.
833 (shortcut_cond_expr): Call expr_has_location, expr_location, and
835 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
837 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
838 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
839 (LANG_HOOKS_INITIALIZER): ... this.
840 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
841 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
843 (update_ebb_live_info): Skip debug insn markers.
844 * lra.c (debug_insn_static_data): Rename to...
845 (debug_bind_static_data): ... this.
846 (debug_marker_static_data): New.
847 (lra_set_insn_recog_data): Select one of the above depending
849 (lra_update_isn_regno_info): Don't assume debug insns have
851 (push_insns): Skip debug insns.
852 * lto-streamer-in.c (input_function): Drop debug stmts
853 depending on active options. Adjust debug_nonbind_markers.
854 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
855 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
856 begin stmt marker notes.
857 (print_insn): Likewise.
858 * recog.c (extract_insn): Recognize rtl for debug markers.
859 * rtl.def (DEBUG_MARKER): New.
860 * tree-inline.c: Include params.h.
861 (remap_gimple_stmt): Handle nonbind markers.
862 (maybe_move_debug_stmts_to_successors): Likewise.
863 (copy_debug_stmt): Likewise.
864 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
865 markers regardless of no side effects.
866 (tsi_link_before): Don't update container's side effects when adding
868 (tsi_link_after): Likewise.
869 (expr_first): Skip begin stmt markers.
870 (expr_last): Likewise.
871 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
872 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
873 Disregard nonbind markers.
874 * tree.c (make_node_stat): Don't set side effects for begin stmt
876 (build1_stat): Likewise.
877 * tree.def (DEBUG_BEGIN_STMT): New.
878 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
879 * var-tracking.c (delete_debug_insns): Renamed to...
880 (delete_vta_debug_insns): ... this.
881 (reemit_marker_as_note): New.
882 (vt_initialize): Reemit markers.
883 (delete_vta_debug_insns): Likewise.
884 (vt_debug_insns_local): Reemit or delete markers.
885 (variable_tracking_main_1): Likewise.
886 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
887 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
888 (gimple_debug_nonbind_marker_p): New.
889 (gimple_build_debug_bind): Adjust.
890 (gimple_build_debug_begin_stmt): New.
891 * doc/invoke.texi (max-debug-marker-count): New param.
892 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
893 (debug_parameter_ref, debug_marker): New.
894 (NOTE_INSN_BEGIN_STMT): New.
895 (DEBUG_INSN): Describe begin stmt markers.
897 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
898 without debug insns, we wouldn't, but clean up debug insns
899 after a control flow insn nevertheless.
901 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
902 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
903 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
904 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
905 (rtl_tidy_fallthru_edge): Likewise.
906 (rtl_verify_fallthru): Likewise.
907 (rtl_verify_bb_layout): Likewise.
908 (skip_insns_after_block): Likewise.
909 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
910 * dwarf2out.c: Include print-rtl.h.
911 (dwarf2out_next_real_insn): New.
912 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
913 Dump debug binds in asm comments.
914 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
915 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
916 callers to use gsi_start_nondebug_bb instead.
917 (gsi_after_labels): Skip gimple debug stmts.
918 (gsi_start_nondebug): New.
919 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
920 (proper_loop_form_for_interchange): Adjust.
921 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
922 * gimple.h (gimple_seq_last_nondebug_stmt): New.
923 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
924 (collect_fallthrough_labels): Likewise.
925 (should_warn_for_implicit_fallthrough): Likewise.
926 (warn_implicit_fallthrough_r): Likewise.
927 (expand_FALLTHROUGH_r): Likewise.
928 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
929 (graphite_copy_stmts_from_block): Skip nonbind markers.
930 * haifa-sched.c (sched_extend_bb): Skip debug insns.
931 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
932 * jump.c (clean_barriers): Skip debug insns.
933 * omp-expand.c (expand_parallel_call): Skip debug insns.
934 (expand_task_call): Likewise.
935 (remove_exit_barrier): Likewise.
936 (expand_omp_taskreg): Likewise.
937 (expand_omp_for_init_counts): Likewise.
938 (expand_omp_for_generic): Likewise.
939 (expand_omp_for_static_nochunk): Likewise.
940 (expand_omp_for_static_chunk): Likewise.
941 (expand_omp_simd): Likewise.
942 (expand_omp_taskloop_for_outer): Likewise.
943 (expand_omp_taskloop_for_inner): Likewise.
944 (expand_oacc_for): Likewise.
945 (expand_omp_sections): Likewise.
946 (expand_omp_single): Likewise.
947 (expand_omp_synch): Likewise.
948 (expand_omp_atomic_load): Likewise.
949 (expand_omp_atomic_store): Likewise.
950 (expand_omp_atomic_fetch_op): Likewise.
951 (expand_omp_atomic_pipeline): Likewise.
952 (expand_omp_atomic_mutex): Likewise.
953 (expand_omp_target): Likewise.
954 (grid_expand_omp_for_loop): Likewise.
955 (grid_expand_target_grid_body): Likewise.
956 (build_omp_regions_1): Likewise.
957 * omp-low.c (check_combined_parallel): Skip debug stmts.
958 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
959 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
960 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
962 * tree-cfg.c (make_blobs_1): Skip debug stmts.
963 (make_edges): Likewise.
964 (cleanup_dead_labels): Likewise.
965 (gimple_can_merge_blocks_p): Likewise.
966 (stmt_starts_bb_p): Likewise.
967 (gimple_block_label): Likewise.
968 (gimple_redirect_edge_and_branch): Likewise.
969 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
971 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
973 * tree-pretty-print (print_declaration): Omit initializer in slim
975 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
977 (eliminate_unnecessary_stmts): Stabilize block removal order.
978 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
979 * var-tracking.c (get_first_insn): New.
980 (vt_emit_notes): Call it.
981 (vt_initialize): Walk any insns before the first BB.
982 (delete_debug_insns): Likewise.
984 * gimple.h (enum gimple_debug_subcode): Add
985 GIMPLE_DEBUG_BEGIN_STMT.
986 (gimple_debug_begin_stmt_p): New.
987 (gimple_debug_nonbind_marker_p): New.
988 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
989 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
990 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
991 * insn-notes.def (BEGIN_STMT): New.
992 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
993 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
994 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
995 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
996 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
997 (INSN_DEBUG_MARKER_KIND): New.
998 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
999 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
1000 (INSN_VAR_LOCATION_PTR): New.
1001 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
1003 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
1004 deps only if debug bind insns are enabled.
1005 (pass_expand::execute): Avoid deep TER and expand
1006 debug locations for debug bind insns only.
1007 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
1008 debug stmts special handling down to debug bind stmts.
1009 * combine.c (try_combine): Narrow debug insns special handling
1010 down to debug bind insns.
1011 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
1012 Narrow debug insns preexisting special handling down to debug
1014 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
1015 handling down to debug bind insns.
1016 * function.c (instantiate_virtual_regs): Skip debug markers,
1017 adjust handling of debug binds.
1018 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
1019 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
1020 * haifa-sched.c (schedule_insn): Narrow special handling of debug
1021 insns to debug bind insns.
1022 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
1023 special handling of debug stmts to debug bind stmts.
1024 * ipa-split.c (split_function): Likewise.
1025 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
1026 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
1028 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
1029 * regrename.c (build_def_use): Likewise.
1030 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
1031 (pass_cprop_hardreg): Narrow special casing of debug insns to
1033 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
1034 * reload1.c (reload): Likewise.
1035 * sese.c (sese_insert_phis_for_liveouts): Narrow special
1036 casing of debug stmts to debug bind stmts.
1037 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
1038 * ssa-iterators.h (num_imm_uses): Likewise.
1039 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
1040 debug stmts to debug bind stmts.
1041 * tree-inline.c (tree_function_versioning): Narrow special casing
1042 of debug stmts to debug bind stmts.
1043 * tree-loop-distribution.c (generate_loops_for_partition):
1044 Narrow special casing of debug stmts to debug bind stmts.
1045 * tree-sra.c (analyze_access_subtree): Narrow special casing
1046 of debug stmts to debug bind stmts.
1047 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
1048 stmts to debug bind stmts.
1049 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
1050 casing of debug stmts to debug bind stmts.
1051 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
1052 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
1053 casing of debug stmts to debug bind stmts.
1054 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
1056 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
1057 debug stmts to debug bind stmts.
1058 (gimple_replace_ssa_lhs): Likewise.
1059 (insert_debug_temp_for_var_def): Likewise.
1060 (insert_debug_temps_for_defs): Likewise.
1061 (reset_debug_uses): Likewise.
1062 * tree-ssanames.c (release_ssa_name_fn): Likewise.
1063 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
1064 (adjust_debug_stmts): Likewise.
1065 (adjust_phi_and_debug_stmts): Likewise.
1066 (vect_do_peeling): Likewise.
1067 * tree-vect-loop.c (vect_transform_loop): Likewise.
1068 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
1069 * var-tracking.c (adjust_mems): Narrow special casing of debug
1070 insns to debug bind insns.
1071 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
1072 (compute_bb_dataflow, vt_find_locations): Likewise.
1073 (vt_expand_loc, emit_notes_for_changes): Likewise.
1074 (vt_init_cfa_base): Likewise.
1075 (vt_emit_notes): Likewise.
1076 (vt_initialize): Likewise.
1077 (vt_finalize): Likewise.
1079 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
1080 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
1081 (next_nonnote_nondebug_insn_bb): New.
1082 (prev_nonnote_nondebug_insn_bb): New.
1083 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
1084 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
1085 (prev_nonnote_nondebug_insn_bb): Declare.
1086 (next_nonnote_nondebug_insn_bb): Declare.
1087 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
1088 * cfgrtl.c (get_last_bb_insn): Likewise.
1089 * lra.c (push_insns): Likewise.
1091 2017-12-11 David Malcolm <dmalcolm@redhat.com>
1094 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
1096 * selftest.c (selftest::test_runner::test_runner): New ctor.
1097 (selftest::test_runner::~test_runner): New dtor.
1098 * selftest.h (class selftest::test_runner): New class.
1100 2017-12-11 Carl Love <cel@us.ibm.com>
1102 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
1103 vec_extract_fp32_from_shortl]): Add #defines.
1104 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
1105 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
1106 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
1107 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
1108 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
1109 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
1110 * doc/extend.texi: Add documentation for the added builtins.
1112 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
1114 PR rtl-optimization/80693
1115 PR rtl-optimization/81019
1116 PR rtl-optimization/81020
1117 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
1118 are not mentioned in i3. Place the REG_UNUSED note on i2,
1119 possibly modified to REG_DEAD, if it did not originate in i3.
1121 2017-12-11 Jakub Jelinek <jakub@redhat.com>
1123 * recog.c (store_data_bypass_p_1): New function.
1124 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
1125 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
1127 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
1129 PR rtl-optimization/83361
1130 * ifcvt.c (if_convert): Call fixup_partitions.
1132 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
1134 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1135 early folding of splat_u{8,16,32}.
1137 2017-12-11 Jakub Jelinek <jakub@redhat.com>
1139 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
1140 output_operand_lossage first argument with capital letter.
1141 (aarch64_override_options): Don't start error and sorry first argument
1142 with capital letter.
1144 2017-12-11 Andi Kleen <ak@linux.intel.com>
1146 PR gcov-profile/83355
1147 * auto-profile.c (string_table::get_index_by_decl): Don't
1148 recurse when abstract origin points to itself.
1150 2017-12-11 Bin Cheng <bin.cheng@arm.com>
1152 PR tree-optimization/83320
1153 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
1154 (prune_datarefs_not_in_loop): Ditto.
1156 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
1158 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
1160 2017-12-10 Jakub Jelinek <jakub@redhat.com>
1162 PR tree-optimization/83337
1163 * gimple-loop-interchange.cc (compute_access_stride): Handle
1164 bitfield DRs properly.
1166 2017-12-09 Jakub Jelinek <jakub@redhat.com>
1168 PR tree-optimization/83338
1169 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
1170 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
1171 vector of unsigned integers to vector of signed integers.
1173 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
1175 PR rtl-optimization/83317
1176 * lra-constraints.c (process_address_1): Add insn code check.
1178 2017-12-08 Michael Matz <matz@suse.de>
1180 Fix PR tree-optimization/83323
1181 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
1182 head-controlled loops and loop BBs.
1183 * common.opt (funroll-and-jam): Remove, instead ...
1184 (floop-unroll-and-jam): ... reuse this option.
1185 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
1186 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
1187 (-floop-unroll-and-jam): ... this option.
1189 2017-12-08 Jakub Jelinek <jakub@redhat.com>
1191 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
1192 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
1193 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
1195 2017-12-08 Richard Biener <rguenther@suse.de>
1198 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
1199 handle accesses outside of zero-sized vars.
1201 2017-12-08 Martin Jambor <mjambor@suse.cz>
1203 PR tree-optimization/83141
1204 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
1205 test for MEM_REFs implicitely changing types with padding. Remove
1207 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
1209 2017-12-08 Jakub Jelinek <jakub@redhat.com>
1211 * config/arc/arc.c (arc_attribute_table): Add exclusions to
1213 * config/avr/avr.c (avr_attribute_table): Likewise.
1214 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
1215 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
1216 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
1217 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
1218 of struct attribute_spec.
1219 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1221 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
1224 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
1226 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
1229 * profile-count.c (profile_count::from_gcov_type): Move from
1230 profile-count.h; handle overflow.
1231 * profile-count.h (profile_count::from_gcov_type): Move offline.
1233 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
1235 PR rtl-optimization/83304
1236 * combine.c (move_deaths): If we do not know where a register died,
1239 2017-12-08 Richard Biener <rguenther@suse.de>
1241 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
1242 Provide -fopt-info-loop feedback when we interchange in a nest.
1244 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
1246 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
1247 for armv6 ARM CPU IDs.
1249 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
1251 * common/config/arm/arm-common.c: Include <algorithm>.
1252 (INCLUDE_VECTOR): Define.
1253 (compare_opt_names): New function.
1254 (arm_rewrite_selected_arch): Only strip out extensions that can be
1255 expressed through -mfpu. Sort the remaining extensions
1258 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
1260 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
1261 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
1262 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
1263 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
1265 2017-06-08 Tristan Gingold <gindold@adacore.com>
1268 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
1269 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
1271 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1273 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
1274 of struct attribute_spec.
1276 2017-12-08 Julia Koval <julia.koval@intel.com>
1278 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
1279 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
1280 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
1281 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
1282 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
1283 _mm_maskz_dpwssds_epi32): Ditto.
1285 2017-12-08 Richard Biener <rguenther@suse.de>
1287 PR tree-optimization/81303
1288 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
1289 conditions try to create a comparison vector type matching
1290 the data vector type.
1291 (vectorizable_condition): Adjust.
1292 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
1293 Leave invariant conditions alone in case we can vectorize those.
1295 2017-12-08 Julia Koval <julia.koval@intel.com>
1297 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
1298 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
1299 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
1300 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
1301 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
1303 2017-12-08 Julia Koval <julia.koval@intel.com>
1305 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
1306 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
1307 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
1308 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
1309 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
1310 _mm_maskz_dpbusds_epi32): New intrinsics.
1312 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
1314 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
1316 (ld<bh>io_signed): Likewise.
1317 (st<bhw>io): Likewise.
1318 * config/nios2/predicates.md (ldstio_memory_operand): Allow
1319 SMALL_INT12 constant integer operand.
1321 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
1323 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
1325 (nios2_large_constant_allowed): ...this. Adjust uses.
1326 (nios2_plus_symbolic_constant_p): Rename to...
1327 (nios2_plus_large_constant_p): ...this. Adjust uses.
1328 (nios2_legitimate_address_p): Correct CONST_INT handling.
1329 (nios2_symbolic_memory_operand_p): Rename to...
1330 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
1331 (nios2_large_constant_p): Check for large constant integers too.
1332 (nios2_split_large_constant): Handle constant integers.
1333 (nios2_split_symbolic_memory_operand): Rename to...
1334 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
1335 (nios2_legitimize_constant_address): Handle constant integers.
1336 (r0rel_constant_p): Handle small constant integers.
1337 (nios2_print_operand_address): Handle r0-relative integer addresses.
1338 * config/nios2/nios2-protos.h: Adjust for renamed functions.
1339 * config/nios2/nios2.md: Adjust for renamed functions.
1341 2017-12-07 Andrew Waterman <andrew@sifive.com>
1343 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
1344 (TARGET_HAVE_SRODATA_SECTION): New define.
1345 (riscv_select_section): New function.
1347 2017-12-08 Joseph Myers <joseph@codesourcery.com>
1348 Alexander Monakov <amonakov@ispras.ru>
1349 Jakub Jelinek <jakub@redhat.com>
1352 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
1354 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1356 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
1357 a bool success value. Don't call output_operand_lossage here.
1358 (aarch64_print_ldpstp_address): Return a bool success value.
1359 (aarch64_print_operand_address): Call output_addr_const if
1360 aarch64_print_address_internal fails.
1361 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
1362 'y'; call output_operand_lossage instead. Call output_operand_lossage
1363 if aarch64_print_ldpstp_address fails.
1365 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1367 * tree-vector-builder.h
1368 (tree_vector_builder::binary_encoded_nelts): Declare.
1369 * tree-vector-builder.c
1370 (tree_vector_builder::binary_encoded_nelts): New function.
1371 * fold-const.c (negate_expr_p): Likewise.
1372 (operand_equal_p, fold_checksum_tree): Likewise.
1373 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
1374 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
1375 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
1376 (uniform_vector_p): Likewise.
1377 * varasm.c (const_hash_1, compare_constant): Likewise.
1378 * tree-ssa-ccp.c: Include tree-vector-builder.h.
1379 (valid_lattice_transition): Operate directly on the VECTOR_CST
1381 * ipa-icf.c: Include tree-vector-builder.h.
1382 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
1383 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
1385 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1387 * tree.c (build_vector): Delete.
1388 * tree.h (build_vector): Make static and move into the self-testing
1391 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1393 * vector-builder.h (vector_builder::derived): New const overload.
1394 (vector_builder::elt): New function.
1395 * tree-vector-builder.h (tree_vector_builder::type): New function.
1396 (tree_vector_builder::apply_step): Declare.
1397 * tree-vector-builder.c (tree_vector_builder::apply_step): New
1399 * gimple-fold.h (tree_vector_builder): Declare.
1400 (gimple_build_vector): Take a tree_vector_builder instead of a
1401 type and vector of elements.
1402 * gimple-fold.c (gimple_build_vector): Likewise.
1403 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
1405 (get_initial_defs_for_reduction): Likewise.
1406 (vectorizable_induction): Likewise.
1408 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1410 * tree-vector-builder.h
1411 (tree_vector_builder::new_binary_operation): Declare.
1412 * tree-vector-builder.c
1413 (tree_vector_builder::new_binary_operation): New function.
1414 * fold-const.c (fold_relational_const): Use it.
1415 (const_binop): Likewise. Check that both input vectors have
1416 the same number of elements, thus excluding things like WIDEN_SUM.
1417 Check whether it is possible to operate directly on the encodings
1420 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1422 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
1423 new_unary_operation, operating only on the encoded elements.
1424 (const_unop): Likewise.
1425 (exact_inverse): Likewise.
1426 (distributes_over_addition_p): New function.
1427 (const_binop): Use tree_vector_builder and new_unary_operation
1428 for combinations of VECTOR_CST and INTEGER_CST. Operate only
1429 on the encoded elements unless the encoding is strided and the
1430 operation does not distribute over addition.
1431 (fold_convert_const): Use tree_vector_builder and
1432 new_unary_operation. Operate only on the encoded elements
1433 for truncating integer conversions, or for non-stepped encodings.
1435 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1437 * config/sparc/sparc.c: Include tree-vector-builder.h.
1438 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
1439 * expmed.c: Include tree-vector-builder.h.
1440 (make_tree): Use tree_vector_builder instead of build_vector.
1441 * fold-const.c: Include tree-vector-builder.h.
1442 (const_binop): Use tree_vector_builder instead of build_vector.
1443 (const_unop): Likewise.
1444 (native_interpret_vector): Likewise.
1445 (fold_vec_perm): Likewise.
1446 (fold_ternary_loc): Likewise.
1447 * gimple-fold.c: Include tree-vector-builder.h.
1448 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
1450 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
1451 (simplify_vector_constructor): Use tree_vector_builder instead
1453 * tree-vect-generic.c: Include tree-vector-builder.h.
1454 (add_rshift): Use tree_vector_builder instead of build_vector.
1455 (expand_vector_divmod): Likewise.
1456 (optimize_vector_constructor): Likewise.
1457 * tree-vect-loop.c: Include tree-vector-builder.h.
1458 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
1459 of build_vector. Explicitly use a stepped encoding for
1461 * tree-vect-slp.c: Include tree-vector-builder.h.
1462 (vect_get_constant_vectors): Use tree_vector_builder instead
1464 (vect_transform_slp_perm_load): Likewise.
1465 (vect_schedule_slp_instance): Likewise.
1466 * tree-vect-stmts.c: Include tree-vector-builder.h.
1467 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
1468 (vect_gen_perm_mask_any): Likewise.
1469 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
1470 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
1472 (build_vector_from_val): Likewise. Explicitly use a duplicate
1475 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
1477 * doc/generic.texi (VECTOR_CST): Describe new representation of
1479 * vector-builder.h: New file.
1480 * tree-vector-builder.h: Likewise.
1481 * tree-vector-builder.c: Likewise.
1482 * Makefile.in (OBJS): Add tree-vector-builder.o.
1483 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
1484 * tree-core.h (tree_base): Add a vector_cst field to the u union.
1485 (tree_vector): Change the number of elements to
1486 vector_cst_encoded_nelts.
1487 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
1488 (VECTOR_CST_ELTS): Delete.
1489 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
1490 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
1491 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
1492 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
1493 (VECTOR_CST_ENCODED_ELT): Likewise.
1494 (vector_cst_encoded_nelts): New function.
1495 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
1496 VECTOR_CST_NELTS_PER_PATTERN as arguments.
1497 (vector_cst_int_elt, vector_cst_elt): Declare.
1498 * tree.c: Include tree-vector-builder.h.
1499 (tree_code_size): Abort if passed VECTOR_CST.
1500 (tree_size): Update for new VECTOR_CST layout.
1501 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
1502 VECTOR_CST_NELTS_PER_PATTERN as arguments.
1503 (build_vector): Use tree_vector_builder.
1504 (vector_cst_int_elt, vector_cst_elt): New functions.
1505 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
1506 encoded elements and then create the vector in the canonical form.
1507 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
1508 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
1509 (tree_c_tests): Call test_vector_cst_patterns.
1510 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
1512 (hash_tree): Likewise.
1513 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
1514 (streamer_write_tree_header): Likewise.
1515 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
1516 (streamer_alloc_tree): Likewise. Update call to make_vector.
1517 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
1519 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1521 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
1522 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
1523 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
1525 2017-12-07 Bin Cheng <bin.cheng@arm.com>
1526 Richard Biener <rguenther@suse.de>
1528 PR tree-optimization/81303
1529 * Makefile.in (gimple-loop-interchange.o): New object file.
1530 * common.opt (floop-interchange): Reuse the option from graphite.
1531 * doc/invoke.texi (-floop-interchange): Ditto. New document for
1532 -floop-interchange and mention it for -O3.
1533 * opts.c (default_options_table): Enable -floop-interchange at -O3.
1534 * gimple-loop-interchange.cc: New file.
1535 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
1536 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
1537 * passes.def (pass_linterchange): New pass.
1538 * timevar.def (TV_LINTERCHANGE): New time var.
1539 * tree-pass.h (make_pass_linterchange): New declaration.
1540 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
1541 interchange. Record IV before/after increment in new parameters.
1542 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
1543 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
1545 (check_reduction_path): ...New function here.
1546 * tree-vectorizer.h (check_reduction_path): New declaration.
1548 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
1551 PR rtl-optimization/80818
1552 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
1553 always early clobbered.
1554 * lra-lives.c (process_bb_lives): Check input hard regs for early
1555 clobbered non-operand hard reg.
1557 2017-12-07 Jakub Jelinek <jakub@redhat.com>
1560 * tree-cfg.c (verify_gimple_assign_binary): Don't require
1561 types_compatible_p, just that TYPE_MODE is the same.
1563 2017-12-07 Martin Sebor <msebor@redhat.com>
1566 * attribs.c (empty_attribute_table): Initialize new member of
1567 struct attribute_spec.
1568 (decl_attributes): Add argument. Handle mutually exclusive
1569 combinations of attributes.
1570 (selftests::test_attribute_exclusions): New function.
1571 (selftests::attribute_c_tests): Ditto.
1572 * attribs.h (decl_attributes): Add default argument.
1573 * selftest.h (attribute_c_tests): Declare.
1574 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
1575 * tree-core.h (attribute_spec::exclusions, exclude): New type and
1577 * doc/extend.texi (Common Function Attributes): Update const and pure.
1578 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
1579 of struct attribute_spec.
1580 * config/arc/arc.c (arc_attribute_table): Same.
1581 * config/arm/arm.c (arm_attribute_table): Same.
1582 * config/avr/avr.c ( avr_attribute_table): Same.
1583 * config/bfin/bfin.c (bfin_attribute_table): Same.
1584 * config/cr16/cr16.c (cr16_attribute_table): Same.
1585 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
1586 * config/h8300/h8300.c (h8300_attribute_table): Same.
1587 * config/i386/i386.c (ix86_attribute_table): Same.
1588 * config/ia64/ia64.c (ia64_attribute_table): Same.
1589 * config/m32c/m32c.c (m32c_attribute_table): Same.
1590 * config/m32r/m32r.c (m32r_attribute_table): Same.
1591 * config/m68k/m68k.c (m68k_attribute_table): Same.
1592 * config/mcore/mcore.c (mcore_attribute_table): Same.
1593 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
1594 * config/mips/mips.c (mips_attribute_table): Same.
1595 * config/msp430/msp430.c (msp430_attribute_table): Same.
1596 * config/nds32/nds32.c (nds32_attribute_table): Same.
1597 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
1598 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
1599 * config/rl78/rl78.c (rl78__attribute_table): Same.
1600 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
1601 * onfig/rx/rx.c (rx_attribute_table): Same.
1602 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
1603 * config/sh/sh.c (sh_attribute_table): Same.
1604 * config/sparc/sparc.c (sparc_attribute_table): Same.
1605 * config/spu/spu.c (spu_attribute_table): Same.
1606 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
1607 * config/v850/v850.c (v850_attribute_table): Same.
1608 * config/visium/visium.c (visium_attribute_table): Same.
1610 2017-12-07 Tamar Christina <tamar.christina@arm.com>
1613 * config/arm/arm.c (INCLUDE_STRING): Define.
1614 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
1615 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
1618 2017-12-07 Michael Matz <matz@suse.de>
1620 Add unroll and jam pass
1622 * gimple-loop-jam.c: New file.
1623 * Makefile.in (OBJS): Add gimple-loop-jam.o.
1624 * common.opt (funroll-and-jam): New option.
1625 * opts.c (default_options_table): Add unroll-and-jam at -O3.
1626 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
1627 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
1628 * passes.def: Add pass_loop_jam.
1629 * timevar.def (TV_LOOP_JAM): Add.
1630 * tree-pass.h (make_pass_loop_jam): Declare.
1631 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
1632 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
1633 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
1634 to flow_loop_tree_node_add.
1635 (duplicate_subloops, copy_loops_to): Append to sibling list.
1636 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
1637 * doc/invoke.texi (-funroll-and-jam): Document new option.
1638 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
1640 2017-12-07 Richard Biener <rguenther@suse.de>
1642 PR tree-optimization/83296
1643 PR tree-optimization/67769
1644 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
1645 flow sensitive info in an unrelated BB.
1646 (value_replacement): Use reset_flow_sensitive_info.
1647 (minmax_replacement): Reset flow sensitive info on the def
1648 we move. Do not reset flow sensitive info in the whole BB
1649 we move the stmt to.
1650 (abs_replacement): Likewise.
1652 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1655 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1656 rs6000_cpu to the given -mcpu=, or to the default processor.
1658 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1660 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
1661 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
1662 instead of rs6000_cpu_attr.
1663 (rs6000_adjust_cost): Ditto.
1664 (is_microcoded_insn): Ditto.
1665 (rs6000_adjust_priority): Ditto.
1666 (rs6000_issue_rate): Ditto.
1667 (rs6000_use_sched_lookahead): Ditto.
1668 (rs6000_use_sched_lookahead_guard): Ditto.
1669 (rs6000_sched_reorder): Ditto.
1670 (force_new_group): Ditto.
1671 * config/rs6000/rs6000.md (cpu attribute): Ditto.
1672 (group_ending_nop): Ditto.
1674 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1676 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
1677 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
1678 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
1679 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
1681 (rs6000_reassociation_width): Ditto.
1682 (rs6000_emit_epilogue): Ditto.
1683 (rs6000_adjust_cost): Ditto.
1684 (is_microcoded_insn): Ditto.
1685 (is_cracked_insn): Ditto.
1686 (rs6000_adjust_priority): Ditto.
1687 (rs6000_sched_reorder): Ditto.
1688 (rs6000_sched_reorder2): Ditto.
1689 (insn_must_be_first_in_group): Ditto.
1690 (insn_must_be_last_in_group): Ditto.
1691 (rs6000_register_move_cost): Ditto.
1692 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
1695 2017-12-07 Julia Koval <julia.koval@intel.com>
1697 * config.gcc: Add vaesintrin.h.
1698 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
1700 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
1701 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
1703 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1704 * config/i386/immintrin.h: Include vaesintrin.h.
1705 * config/i386/sse.md (vaesdec_<mode>): New pattern.
1706 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
1707 _mm_aesdec_epi128): New intrinsics.
1709 2017-12-06 David Malcolm <dmalcolm@redhat.com>
1711 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
1712 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
1713 c-family/c-spellcheck.cc.
1714 (best_macro_match::best_macro_match): Likewise.
1716 (struct edit_distance_traits<cpp_hashnode *>): Move to
1717 c-family/c-spellcheck.h.
1718 (class best_macro_match): Likewise.
1720 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1722 PR tree-optimization/83293
1723 * gimple-ssa-strength-reduction.c (insert_initializers): Use
1724 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
1725 might insert into empty bb.
1728 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
1729 simplify for plus with :c added, and pointer_plus without that.
1730 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
1731 with undefined overflow and the conversion is not widening,
1732 perform negation in utype and only convert to type afterwards.
1733 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
1734 simplify for plus with :c added, and pointer_plus without that.
1735 If type is integral with undefined overflow and the conversion is
1736 not widening, perform minus in utype and only convert to type
1737 afterwards. Move the last pointer_diff_expr simplify into the
1740 2017-12-06 Martin Sebor <msebor@redhat.com>
1742 PR tree-optimization/82646
1743 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
1744 strncpy, not maxlen.
1746 2017-12-06 Martin Sebor <msebor@redhat.com>
1748 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
1751 PR tree-optimization/83075
1752 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
1753 strncat/strncpy don't change length of source string.
1755 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
1758 2017-11-29 Martin Aberg <maberg@gaisler.com>
1760 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1761 to prevent b2bst errata sequence.
1762 (sqrtdf2_fix): Likewise.
1764 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1766 PR tree-optimization/81945
1767 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
1768 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
1769 to dest_cfun has orig_loop_num set, either remap it to the new
1770 loop number if the loop got moved too, or clear it.
1772 2017-12-05 Steve Ellcey <sellcey@cavium.com>
1774 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
1776 (thunderx2t99_nothing): New insn reservation.
1777 (thunderx2t99_mrs): New insn reservation.
1778 (thunderx2t99_multiple): New insn reservation.
1779 (thunderx2t99_alu_basi): Add bfx to reservation.
1780 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
1782 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1785 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
1787 2017-12-05 Bin Cheng <bin.cheng@arm.com>
1789 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
1790 tree-ssa-pre.c::remove_dead_inserted_code.
1791 * tree-ssa-dce.h: New file.
1792 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
1793 (remove_dead_inserted_code): Move and rename to function
1794 tree-ssa-dce.c::simple_dce_from_worklist.
1795 (pass_pre::execute): Update use.
1797 2017-12-05 Richard Biener <rguenther@suse.de>
1799 PR tree-optimization/83277
1800 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
1801 to code-gen liveout vars.
1803 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
1805 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
1806 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
1807 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
1810 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
1812 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
1814 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
1815 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
1818 2017-12-05 Richard Biener <rguenther@suse.de>
1820 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
1821 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
1822 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
1823 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
1825 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1827 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
1829 2017-12-05 Martin Liska <mliska@suse.cz>
1830 Jakub Jelinek <jakub@redhat.com>
1832 * doc/invoke.texi: Document the options.
1833 * flag-types.h (enum sanitize_code): Add
1834 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1835 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
1836 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1837 * opts.c: Define new sanitizer options.
1838 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
1839 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
1841 2017-12-05 Julia Koval <julia.koval@intel.com>
1843 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
1844 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
1845 (ix86_handle_option): Handle -mavx512vnni.
1846 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
1847 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
1848 * config/i386/i386-c (__AVX512VNNI__): New.
1849 * config/i386/i386.c (ix86_target_string): Handle new option.
1850 (ix86_valid_target_attribute_inner_p): Handle new option.
1851 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
1852 * config/i386/i386.opt (mavx512vnni): New option.
1854 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1857 * athlon.md: Disable for generic.
1858 * haswell.md: Enable for generic.
1859 * i386.c (ix86_sched_init_global): Add core hooks for generic.
1860 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
1862 (ix86_adjust_cost): Move generic to haswell path.
1864 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
1866 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
1867 instead of MEM_P in a couple more places. Fix formatting issues.
1869 2017-12-04 Jim Wilson <jimw@sifive.com>
1871 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
1872 instead of GP_REG_LAST-1.
1873 (riscv_adjust_libcall_cfi_prologue): Likewise.
1874 (riscv_adjust_libcall_cri_epilogue): Likewise.
1875 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
1878 2017-12-04 Luis Machado <luis.machado@linaro.org>
1880 * ipa-pure-const.c (check_decl): Add missing newline.
1881 (state_from_flags): Likewise.
1883 2017-12-04 Jeff Law <law@redhat.com>
1885 PR tree-optimizatin/78496
1886 * gimple-ssa-evrp-analyze.h
1887 (evrp_range_analyzer::get_vr_values): Simplify.
1888 * gimple-ssa-evrp-analyze.c: Corresponding changes.
1889 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
1890 and gimple-ssa-evrp-analyze.h.
1891 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
1892 (simplify_stmt_for_jump_threading): Copy a blob of code from
1893 tree-vrp.c to use ranges to simplify statements.
1894 (dom_opt_dom_walker::before_dom_children): Call
1895 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
1896 (dom_opt_dom_walker::after_dom_children): Similarly for
1897 evrp_range_analyzer::leave.
1898 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
1901 * gimple-ssa-evrp-analyze.c
1902 (evrp_range_analyzer::extract_range_from_stmt): Always use
1903 vr_values::update_value_range so preexisting range info is
1904 medged with new range info, even if the new range is VR_VARYING.
1906 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1908 * combine.c: Adjust comment.
1909 (use_crosses_set_p): Delete.
1910 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
1911 (try_combine): Ditto.
1913 2017-12-04 Richard Biener <rguenther@suse.de>
1915 PR tree-optimization/83255
1916 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
1917 Re-add zero-iteration check.
1919 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1921 PR rtl-optimization/83245
1922 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
1923 hard registers as earlyclobber, also if not in an asm.
1925 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1929 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1932 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1933 rs6000_cpu based on cpu_index, not tune_index.
1935 2017-12-04 Richard Biener <rguenther@suse.de>
1937 PR tree-optimization/83238
1938 * graphite-scop-detection.c (scop_detection::merge_sese): Make
1939 code match comment, rejecting invalid SESE regions.
1941 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
1943 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
1944 require base operand is a REG_POINTER prior to reload on targets
1945 with non-equivalent space registers.
1947 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1949 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
1950 (update_specialized_profile): Fix updating of counts.
1951 (perhaps_add_new_callers): Likewise.
1953 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1956 * x86-tune.def: Remove obsolette FIXMEs.
1957 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
1958 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
1959 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
1961 (X86_TUNE_PAD_RETURNS): Disable for generic.
1963 2017-12-02 Jakub Jelinek <jakub@redhat.com>
1965 PR tree-optimization/83170
1966 PR tree-optimization/83241
1967 * gimple-ssa-store-merging.c
1968 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
1969 gimple_vuse (ins_stmt) in case it has changed.
1970 (imm_store_chain_info::output_merged_store): Likewise.
1972 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
1976 * tree-cfg.c (pass_warn_function_return::execute): Handle
1977 __builtin_ubsan_handle_missing_return like __builtin_unreachable
1978 with BUILTINS_LOCATION.
1982 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
1983 is BLKmode for vector field with vector raw mode, use TYPE_MODE
1984 instead of DECL_MODE.
1986 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
1987 last argument to rtx pointer.
1988 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
1989 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
1990 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
1991 depending on the chosen ISAs.
1992 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
1993 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
1994 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
1996 * config/i386/sse.md (mov<mode>_internal): Likewise.
1997 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
1999 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2001 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
2003 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2005 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
2006 parameters from prototype.
2007 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
2008 parameters. Don't print a comment.
2009 (emit_fusion_gpr_load): Adjust.
2010 (emit_fusion_load_store): Adjust.
2011 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
2012 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
2013 comment on the second line.
2015 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2018 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
2019 rs6000_cpu based on cpu_index, not tune_index.
2021 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2023 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
2024 which_alternative instead of which_alternative + 1.
2025 (output_asm_insn): Print an extra tab if the template is short.
2027 2017-12-01 Jim Wilson <jimw@sifive.com>
2029 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
2031 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
2032 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
2034 * doc/tm.texi: Regenerate.
2036 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
2039 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
2040 whether we can allocate pseudos before trying to fix an address.
2041 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
2042 memory address is indexed or indirect.
2043 (floatuns_<mode>si2_hw2): Likewise.
2045 2017-12-01 Jason Merrill <jason@redhat.com>
2047 * Makefile.in (TAGS): Add c-family/*.cc.
2049 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
2051 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
2052 (call_value_insn): Likewise.
2053 (sibcall_insn): Likewise.
2054 (sibcall_value_insn): Likewise.
2055 (movsi_aarch64): Likewise.
2056 (movdi_aarch64): Likewise.
2057 (add_losym_): Likewise.
2058 (ldr_got_small_): Likewise.
2059 (ldr_got_small_sidi): Likewise.
2060 (ldr_got_small_28k_): Likewise.
2061 (ldr_got_small_28k_sidi): Likewise.
2062 * config/aarch64/aarch64.c (aarch64_print_address_internal):
2063 Move output_addr_const to symbolic case. Add error check.
2065 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2067 * config/s390/predicates.md (plus16_Q_operand): New predicate.
2068 * config/s390/s390.md: Disable MVC merging peephole if it would
2069 disable operand forwarding.
2070 (new peephole2): Split MVCs if it would turn them into up to 2
2073 2017-12-01 Richard Biener <rguenther@suse.de>
2075 PR tree-optimization/83232
2076 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
2077 detection of same access. Instead of breaking the group here
2078 do not consider the duplicate. Add comment explaining real fix.
2080 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
2082 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
2084 2017-12-01 Sudakshina Das <sudi.das@arm.com>
2086 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
2088 2017-12-01 Jakub Jelinek <jakub@redhat.com>
2090 * function.h (struct function): Remove cilk_frame_decl,
2091 is_cilk_function and calls_cilk_spawn fields.
2092 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
2094 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
2095 cilk_elemental field.
2096 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
2097 * target.def: Adjust comment.
2098 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
2099 Don't test cilk_elemental.
2101 PR tree-optimization/83233
2102 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
2103 bswap_stat name for the struct.
2106 * tree.h (SWITCH_BREAK_LABEL_P): Define.
2107 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
2108 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
2109 SWITCH_BREAK_LABEL_P set on the label.
2110 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
2111 added for default case if it was missing and not all cases covered.
2112 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
2113 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
2115 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
2116 to avoid -Wimplicit-fallthrough warning.
2117 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
2118 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
2120 PR tree-optimization/83221
2121 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
2123 (init_reassoc): Formatting fix.
2126 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
2127 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
2131 * tree-cfg.c: Include asan.h.
2132 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
2134 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2136 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
2138 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2140 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
2141 vpcmpeqd instruction.
2143 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2145 * config/i386/i386.c (standard_sse_constant_opcode): Fix
2146 registers type for 128bit mode.
2148 2017-11-30 Jakub Jelinek <jakub@redhat.com>
2150 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
2152 * gengtype-state.c (read_state_pair): Likewise.
2153 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
2154 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
2155 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
2156 * ifcvt.c (noce_try_store_flag_constants): Likewise.
2157 * tree-ssa-ccp.c (ccp_finalize): Likewise.
2158 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
2159 * builtins.c (fold_builtin_3): Likewise.
2160 * graphite-scop-detection.c
2161 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
2162 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
2165 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
2168 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
2169 parameter. Return true if ignore_decl_rtl_set_p is true and other
2170 conditions are satisfied.
2171 * asan.h (asan_protect_global): Add new parameter.
2172 * varasm.c (categorize_decl_for_section): Pass true as second parameter
2173 to asan_protect_global calls.
2175 2017-11-30 Jim Wilson <jimw@sifive.com>
2177 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
2178 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
2179 -mstrict-align, add info on default value. Delete redundant lines for
2180 -mabi. Add missing -mexplicit-relocs docs.
2182 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2184 * config/arc/arc.md (trap): New pattern.
2186 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2188 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
2189 instruction to end into a delay slot.
2190 * config/arc/arc.md (cond_delay_insn): Check if the instruction
2191 can be placed into a delay slot against reg_note.
2193 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2195 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
2196 labels number of usages.
2198 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2200 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
2202 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
2204 2017-11-30 Jakub Jelinek <jakub@redhat.com>
2207 * internal-fn.c (expand_mul_overflow): Optimize unsigned
2208 multiplication by power of 2 constant into two shifts + comparison.
2210 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
2213 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
2215 2017-11-30 Richard Biener <rguenther@suse.de>
2217 PR tree-optimization/83202
2218 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2219 allow_peel argument and guard peeling.
2220 (canonicalize_loop_induction_variables): Likewise.
2221 (canonicalize_induction_variables): Pass false.
2222 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
2223 peeling from cunrolli.
2225 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
2227 * combine.c (try_combine): Print a message to dump file whenever
2228 I0, I1, or I2 cannot be combined into I3.
2230 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
2232 PR rtl-optimization/83156
2233 PR rtl-optimization/82621
2234 * combine.c (try_combine): Don't split an I2 if one of the dests is
2235 set again before I3. Allow unused dests.
2237 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
2239 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
2241 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
2243 PR rtl-optimization/80818
2244 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
2245 recursively. Use insn code for clobber.
2246 (lra_set_insn_recog_data): Pass the new arg to
2247 collect_non_operand_hard_regs.
2248 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
2250 (lra_update_insn_regno_info): Pass insn to
2251 add_regs_to_insn_regno_info.
2253 2017-11-29 Jim Wilson <jimw@sifive.com>
2254 Andrew Waterman <andrew@sifive.com>
2256 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
2257 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
2258 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
2259 use SINGLE_SHIFT_COST.
2260 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
2261 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
2263 2017-11-29 Julia Koval <julia.koval@intel.com>
2265 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
2266 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
2267 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
2268 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
2269 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
2270 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
2271 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
2272 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
2273 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
2274 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
2275 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
2276 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
2277 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
2278 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
2279 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
2280 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
2281 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
2282 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
2283 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
2284 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
2285 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
2286 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
2287 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
2288 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
2289 __builtin_ia32_vpshldv_v2di_mask,
2290 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
2291 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
2292 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
2294 2017-11-29 Julia Koval <julia.koval@intel.com>
2296 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
2297 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
2298 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
2299 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
2300 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
2301 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
2302 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
2303 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
2304 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
2305 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
2306 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
2307 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
2308 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
2309 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
2310 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
2311 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
2312 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
2313 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
2314 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2315 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
2316 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
2318 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2320 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
2321 movsi_pic_gotdata_op instruction as a load for the UT699 errata
2324 2017-11-29 Martin Aberg <maberg@gaisler.com>
2326 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
2327 to prevent b2bst errata sequence.
2328 (sqrtdf2_fix): Likewise.
2330 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2332 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
2333 (div_sqrt_insn_p): New function.
2334 (sparc_do_work_around_errata): Insert NOP instructions to
2335 prevent sequences that could trigger the TN-0013 errata for
2336 certain LEON3 processors.
2337 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
2338 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
2339 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
2340 (in_branch_delay): Prevent div and sqrt in delay slot if
2342 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
2344 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2346 * config/sparc/sparc.c (atomic_insn_p): New function.
2347 (sparc_do_work_around_errata): Insert NOP instructions to
2348 prevent sequences that could trigger the TN-0010 errata for
2350 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
2351 instruction referable in atomic_insns_p.
2353 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2355 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
2356 (atomic_compare_and_swap_leon3_1): Likewise.
2359 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2361 * config/sparc/sparc.c (fpop_insn_p): New function.
2362 (sparc_do_work_around_errata): Insert NOP instructions to
2363 prevent sequences that could trigger the TN-0012 errata for
2365 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
2366 * config/sparc/sparc.md (fix_gr712rc): New attribute.
2367 (in_branch_annul_delay): Prevent floating-point instructions
2368 in delay slot of annulled integer branch.
2370 2017-11-29 Richard Biener <rguenther@suse.de>
2372 PR tree-optimization/83202
2373 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
2375 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
2376 nodes vectorized to the same stmts multiple times.
2377 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
2378 (vect_analyze_slp_instance): Adjust.
2379 (scalar_stmts_to_slp_tree_map_t): New typedef.
2380 (vect_schedule_slp_instance): Add a map recording the SLP node
2381 representing the vectorized stmts for a set of scalar stmts.
2382 Avoid code-generating redundancies.
2383 (vect_schedule_slp): Allocate map and pass it down.
2385 2017-11-29 Nathan Sidwell <nathan@acm.org>
2388 * tree.c (build_complex_type): Fix canonicalization. Only fill in
2391 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
2393 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
2394 cases for printing LDP/STP memory addresses.
2395 (aarch64_print_address_internal): Renamed from
2396 aarch64_print_operand_address, added parameter, add Pmode check.
2397 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
2398 (aarch64_print_operand_address): Indirect to
2399 aarch64_print_address_internal.
2400 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
2401 'y' operand output specifier.
2403 2017-11-29 Jakub Jelinek <jakub@redhat.com>
2406 * tree.c (build_simple_mem_ref_loc): Handle
2407 get_addr_base_and_unit_offset returning a MEM_REF.
2410 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
2413 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
2416 2017-11-28 David Malcolm <dmalcolm@redhat.com>
2418 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
2420 (layout::move_to_column): Likewise.
2422 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2425 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
2426 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
2428 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2429 Martin Jambor <mjambor@suse.cz>
2432 * tree.h (expr_type_first_operand_type_p): Declare
2433 * tree.c (expr_type_first_operand_type_p): New function.
2434 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
2435 (ipa_value_from_jfunc): Adjust declaration.
2436 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
2437 Use it as result type for arithmetics, unless it is NULL in which case
2438 be more conservative.
2439 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
2440 ipa_get_jf_pass_through_result.
2441 (propagate_vals_across_pass_through): Likewise.
2442 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
2443 is to propagate_vals_across_pass_through.
2444 (propagate_constants_across_call): Pass PARM_TYPE to
2445 propagate_scalar_across_jump_function.
2446 (find_more_scalar_values_for_callers_subset): Pass parameter type to
2447 ipa_value_from_jfunc.
2448 (cgraph_edge_brings_all_scalars_for_node): Likewise.
2449 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
2450 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
2451 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
2452 target_type, pass it to ipa_value_from_jfunc.
2453 (update_indirect_edges_after_inlining): Pass parameter type to
2454 try_make_edge_direct_simple_call.
2456 2017-11-28 Jeff Law <law@redhat.com>
2458 * gimple-ssa-evrp-analyze.c
2459 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
2460 refine ranges if scev_initialized_p returns true.
2461 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
2463 2017-11-28 Julia Koval <julia.koval@intel.com>
2465 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
2466 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
2467 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
2468 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
2469 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
2470 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
2471 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
2472 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
2473 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
2474 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
2475 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
2476 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
2477 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
2478 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
2479 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
2480 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
2481 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
2482 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
2483 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
2484 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
2485 __builtin_ia32_vpshrd_v2di_mask): New builtins.
2486 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
2488 2017-11-28 Julia Koval <julia.koval@intel.com>
2490 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
2491 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
2492 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
2493 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
2494 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
2495 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
2496 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
2497 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
2498 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
2499 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
2500 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
2501 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
2502 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
2503 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
2504 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
2505 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
2506 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
2507 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
2508 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
2509 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
2510 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
2511 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
2512 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
2513 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
2514 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
2515 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
2516 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
2517 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
2518 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
2519 __builtin_ia32_vpshld_v2di_mask): New builtins.
2520 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2521 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
2523 2017-11-28 Richard Biener <rguenther@suse.de>
2525 PR tree-optimization/80776
2526 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
2528 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
2530 (evrp_range_analyzer::record_ranges_from_incoming_edges):
2531 If the incoming edge is an effective fallthru because the other
2532 edge only reaches a __builtin_unreachable () then record ranges
2533 derived from the controlling condition in SSA info.
2534 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
2535 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
2537 2017-11-28 Olivier Hainque <hainque@adacore.com>
2539 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
2540 on mingw build hosts.
2542 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2545 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
2547 2017-11-28 Richard Biener <rguenther@suse.de>
2550 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
2551 copies generated from memcpy use a character array as reference
2554 2017-11-28 Julia Koval <julia.koval@intel.com>
2555 Sebastian Peryt <sebastian.peryt@intel.com>
2557 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
2558 c-family/c-cilkplus.o, c-family/array-notation-common.o,
2559 cilk-common.o, cilk.h, cilk-common.c): Remove.
2561 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
2562 * builtins.c (is_builtin_name): Remove cilkplus condition.
2563 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
2564 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
2565 cilk-builtins.def, cilkplus.def): Remove.
2566 * cif-code.def (CILK_SPAWN): Remove.
2567 * cilk-builtins.def: Delete.
2568 * cilk-common.c: Ditto.
2570 * cilkplus.def: Ditto.
2571 * config/darwin.h (fcilkplus): Delete.
2572 * cppbuiltin.c: Ditto.
2573 * doc/extend.texi: Remove cilkplus doc.
2574 * doc/generic.texi: Ditto.
2575 * doc/invoke.texi: Ditto.
2576 * doc/passes.texi: Ditto.
2577 * gcc.c (fcilkplus): Remove.
2578 * gengtype.c (cilk.h): Remove.
2579 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
2581 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
2583 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
2585 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
2586 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
2587 cilkplus conditions.
2588 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
2589 inline_read_section): Ditto.
2590 * ipa-inline-analysis.c (cilk.h): Remove.
2591 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
2592 * lto-wrapper.c (merge_and_complain, append_compiler_options,
2593 append_linker_options): Remove condition for fcilkplus.
2594 * lto/lto-lang.c (cilk.h): Remove.
2595 (lto_init): Remove condition for fcilkplus.
2596 * omp-expand.c (expand_cilk_for_call): Delete.
2597 (expand_omp_taskreg, expand_omp_for_static_chunk,
2598 expand_omp_for): Remove cilkplus
2600 (expand_cilk_for): Delete.
2601 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
2602 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
2603 execute_lower_omp, diagnose_sb_0): Ditto.
2604 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
2605 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2606 * tree-nested.c: Ditto.
2607 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
2608 (dump_generic_node): Ditto.
2609 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2610 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
2612 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
2614 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2616 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
2617 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
2618 for vector mode and !TARGET_SIMD.
2620 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2622 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
2624 * tree.h (SWITCH_LABELS): Remove.
2625 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
2626 assert SWITCH_BODY is non-NULL.
2627 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
2629 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
2631 PR tree-optimization/80788
2632 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
2633 has TREE_OVERFLOW set, call drop_tree_overflow.
2635 2017-11-28 Richard Biener <rguenther@suse.de>
2637 PR tree-optimization/83158
2638 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
2640 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
2643 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
2644 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
2646 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
2648 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
2649 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
2651 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2654 * varasm.c (compare_constant): Take the mode of the constants into
2655 account when comparing floating point constants.
2657 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
2659 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
2660 from explicit instantiation of debug_helper.
2661 * vec.h (DEFINE_DEBUG_VEC): Ditto.
2663 2017-11-27 Richard Biener <rguenther@suse.de>
2665 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
2668 2017-11-27 Richard Biener <rguenther@suse.de>
2670 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
2671 (type_hash_canon): Also clear int_cst_hash_table entry for
2673 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
2675 2017-11-27 Tamar Christina <tamar.christina@arm.com>
2677 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2678 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2679 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2681 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2683 * hash-map.h (gt_cleare_cache): Avoid UB.
2685 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
2687 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
2688 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
2689 (decide_unroll_runtime_iterations): Likewise.
2690 (decide_unroll_stupid): Likewise.
2692 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2695 * config/i386/i386.md: Add a loop with incssp.
2697 2017-11-27 Martin Jambor <mjambor@suse.cz>
2699 PR tree-optimization/81248
2700 * tree-sra.c (splice_param_accesses): Remove size check.
2701 (decide_one_param_reduction): Fix size check.
2702 * gimple-pretty-print.c (dump_profile): Silence warning.
2703 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
2705 2017-11-27 Jakub Jelinek <jakub@redhat.com>
2708 * dbxout.c (lastlineno): New variable.
2709 (dbx_debug_hooks): Use dbxout_switch_text_section as
2710 switch_text_section debug hook.
2711 (dbxout_function_end): Switch to current_function_section
2712 rather than function_section. If crtl->has_bb_partition,
2713 output just one N_FUN, depending on in_cold_section_p.
2714 (dbxout_source_line): Remember last lineno in lastlineno.
2715 (dbxout_switch_text_section): New function.
2716 (dbxout_function_decl): Adjust dbxout_block caller.
2717 (dbx_block_with_cold_children): New function.
2718 (dbxout_block): Return true if any LBRAC/RBRAC have been
2719 emitted. Use dbx_block_with_cold_children at depth == 0
2720 in second partition. Add PARENT_BLOCKNUM argument, pass
2721 it optionally adjusted to children. Output LBRAC/RBRAC
2722 around recursive call only if the block is in the current
2723 partition, if not and anything was output, emit empty
2725 * final.c (final_scan_insn): Compute cold_function_name
2726 before calling switch_text_section debug hook. Call
2727 that hook even if dwarf2out_do_frame if not emitting
2731 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
2732 TREE_READONLY decls.
2734 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2736 PR rtl-optimization/82488
2737 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
2739 2017-11-26 Julia Koval <julia.koval@intel.com>
2741 * config/i386/i386.c (processor_target_table): Add skylake_cost for
2743 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
2746 2017-11-26 Julia Koval <julia.koval@intel.com>
2748 * config/i386/driver-i386.c (host_detect_local_cpu):
2749 Detect skylake-avx512.
2751 2017-11-26 Julia Koval <julia.koval@intel.com>
2753 * config.gcc: Add -march=cannonlake.
2754 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
2755 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
2756 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
2757 (PTA_CANNONLAKE): New.
2758 (processor_target_table): Add cannonlake.
2759 (ix86_option_override_internal): Ditto.
2760 (fold_builtin_cpu): Ditto.
2761 (get_builtin_code_for_version): Handle cannonlake.
2762 (M_INTEL_COREI7_CANNONLAKE): New.
2763 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
2764 * doc/invoke.texi: Add -march=cannonlake.
2766 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
2768 * plugin.c (add_new_plugin): Use platform-specific library extensions.
2769 (try_init_one_plugin): Alternative implementation for MinGW.
2770 * Makefile.in (plugin_implib): New.
2771 (gengtype-lex.c): Fix broken AIX workaround.
2772 * configure: Regenerate.
2773 * doc/plugins.texi: Document support for MinGW.
2775 2017-11-25 Jakub Jelinek <jakub@redhat.com>
2777 PR rtl-optimization/81553
2778 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
2779 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
2780 is a shift where C1 has different mode than the whole shift, use C1's
2781 mode for MULT rather than the shift's mode.
2784 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
2785 builtins not enabled in the currently selected ISA.
2787 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
2789 PR tree-optimization/71026
2790 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
2791 (insert_reciprocals): Change to insert reciprocals before a division
2792 by a square and to insert the square of a reciprocal.
2793 (execute_cse_reciprocals_1): Change to consider division by a square.
2794 (register_division_in): Add importance parameter.
2796 2017-11-24 Richard Biener <rguenther@suse.de>
2798 PR tree-optimization/82402
2799 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
2800 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
2802 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
2804 * match.pd (0-ptr): New transformation.
2806 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
2809 * ipa-inline.c (inline_small_functions): Set current badnes correctly
2810 when skipping checking.
2812 2017-11-24 Richard Biener <rguenther@suse.de>
2814 PR tree-optimization/83128
2815 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
2816 (vn_reference_lookup_3): Likewise.
2818 2017-11-24 Jakub Jelinek <jakub@redhat.com>
2821 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
2822 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
2825 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
2826 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
2827 check for the latter with gimple_call_builtin_p. Do not handle
2828 BUILT_IN_STPNCPY_CHK which is not a pass through call.
2830 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
2832 * config/arm/arm_neon.h: Fix pragma GCC push_options before
2835 2017-11-23 Julia Koval <julia.koval@intel.com>
2837 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
2838 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
2839 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
2840 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
2841 _mm512_maskz_expandloadu_epi16): New intrinsics.
2842 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
2843 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
2844 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
2845 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
2846 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
2847 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
2848 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
2849 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
2850 _mm256_maskz_expandloadu_epi8): New intrinsics.
2851 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
2852 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
2853 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
2854 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
2855 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
2856 * config/i386/sse.md (VI248_VLBW): New iterator.
2857 (expand<mode>_mask, expand<mode>_maskz): New patterns.
2859 2017-11-23 Julia Koval <julia.koval@intel.com>
2861 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
2862 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
2863 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
2864 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
2865 _mm512_mask_compressstoreu_epi16): New.
2866 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
2867 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
2868 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
2869 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
2870 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
2871 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
2872 _mm256_mask_compressstoreu_epi8): New.
2873 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
2874 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
2875 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
2876 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
2877 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
2878 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
2879 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
2880 __builtin_ia32_compresshi128_mask,
2881 __builtin_ia32_compressstoreuqi512_mask,
2882 __builtin_ia32_compressstoreuhi512_mask,
2883 __builtin_ia32_compressstoreuqi256_mask,
2884 __builtin_ia32_compressstoreuqi128_mask,
2885 __builtin_ia32_compressstoreuhi256_mask,
2886 __builtin_ia32_compressstoreuhi128_mask): New builtins.
2887 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
2889 (ix86_expand_special_args_builtin): Handle new types.
2890 (s4fma_expand): Handle new builtin array.
2891 * config/i386/immintrin.h: Include new headers.
2892 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
2893 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
2895 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2897 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
2898 fixes. Declare temp and g variables at the top in order to avoid
2899 {} in most of the cases.
2901 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
2903 * match.pd (ptr-0): New transformation.
2905 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2907 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
2908 (struct addr_mode_cost_table): New.
2909 (struct tune_params): Add field addr_mode_costs.
2910 * config/arm/arm.c (generic_addr_mode_costs): New.
2911 (arm_slowmul_tune): Initialise addr_mode_costs field.
2912 (arm_fastmul_tune): Likewise.
2913 (arm_strongarm_tune): Likewise.
2914 (arm_xscale_tune): Likewise.
2915 (arm_9e_tune): Likewise.
2916 (arm_marvell_pj4_tune): Likewise.
2917 (arm_v6t2_tune): Likewise.
2918 (arm_cortex_tune): Likewise.
2919 (arm_cortex_a8_tune): Likewise.
2920 (arm_cortex_a7_tune): Likewise.
2921 (arm_cortex_a15_tune): Likewise.
2922 (arm_cortex_a35_tune): Likewise.
2923 (arm_cortex_a53_tune): Likewise.
2924 (arm_cortex_a57_tune): Likewise.
2925 (arm_exynosm1_tune): Likewise.
2926 (arm_xgene1_tune): Likewise.
2927 (arm_cortex_a5_tune): Likewise.
2928 (arm_cortex_a9_tune): Likewise.
2929 (arm_cortex_a12_tune): Likewise.
2930 (arm_cortex_a73_tune): Likewise.
2931 (arm_v7m_tune): Likewise.
2932 (arm_cortex_m7_tune): Likewise.
2933 (arm_v6m_tune): Likewise.
2934 (arm_fa726te_tune): Likewise.
2935 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
2937 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2939 * config/arm/arm.c (arm_mem_costs): New function.
2940 (arm_rtx_costs_internal): Use arm_mem_costs.
2942 2017-11-23 Mark Wielaard <mark@klomp.org>
2944 * dwarf2out.c (init_sections_and_labels): Use generation to create
2945 unique ranges_section_label and ranges_base_label. Return generation.
2946 (output_rnglists): Add generation argument. Use generation to create
2947 unique ranges labels.
2948 (dwarf2out_finish): Get generation from init_sections_and_labels
2949 and pass generation to output_rnglists.
2951 2017-11-23 Mike Stump <mikestump@comcast.net>
2952 Eric Botcazou <ebotcazou@adacore.com>
2954 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
2955 * cfgloop.h (struct loop): Add unroll field.
2956 * function.h (struct function): Add has_unroll bitfield.
2957 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
2958 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
2959 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
2960 (pass_rtl_unroll_loops::gate): Likewise.
2961 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
2962 for which loop->unroll==1.
2963 (decide_unroll_constant_iterations): Use note for consistency and
2964 take loop->unroll into account. Return early if loop->unroll is set.
2965 Fix thinko in existing test.
2966 (decide_unroll_runtime_iterations): Use note for consistency and
2967 take loop->unroll into account.
2968 (decide_unroll_stupid): Likewise.
2969 * lto-streamer-in.c (input_cfg): Read loop->unroll.
2970 * lto-streamer-out.c (output_cfg): Write loop->unroll.
2971 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
2973 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
2974 (print_loop): Print loop->unroll if set.
2975 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
2976 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
2977 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
2979 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
2980 loop->unroll is set and smaller than the trip count. Otherwise bypass
2981 entirely the heuristics if loop->unroll is set. Remove dead note.
2982 Fix off-by-one bug in other note.
2983 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
2984 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
2986 (tree_unroll_loops_completely): Make static.
2987 (pass_complete_unroll::execute): Use correct type for variable.
2988 (pass_complete_unrolli::execute): Fix formatting.
2989 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
2991 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
2993 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
2994 enable when TARGET_PREFER_AVX128 is set.
2996 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2998 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
3000 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
3001 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
3003 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3005 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
3006 frequencies. Fix estimation of aggregate parameters.
3008 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3010 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
3013 2017-11-23 Tom de Vries <tom@codesourcery.com>
3015 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
3017 2017-11-23 Tom de Vries <tom@codesourcery.com>
3019 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
3022 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
3025 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
3026 sibcall_value_pcrel_fdpic): Use local variable instead of
3028 (calli_tbr_rel): Add missing operand 2.
3029 (call_valuei_tbr_rel): Add missing operand 3.
3031 2017-11-23 Jakub Jelinek <jakub@redhat.com>
3034 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
3035 bitpos/bitsize covering the whole destination, use store_expr only if
3036 the complex mode is the same. Otherwise, use expand_normal and if
3037 it returns CONCAT, subreg each part separately instead of trying to
3038 subreg the whole result.
3040 2017-11-23 Richard Biener <rguenther@suse.de>
3042 PR tree-optimization/23094
3043 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
3044 come here from walking over backedges in the first iteration.
3045 (vn_reference_lookup_3): Skip clobbers that store the same value.
3047 2017-11-23 Richard Biener <rguenther@suse.de>
3049 PR tree-optimization/81403
3050 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
3051 a block we need a leader relative to.
3052 (phi_translate_1): For nary processing require a leader from
3053 get_representative_for given we run expression simplification
3054 using match-and-simplify. Remove previous fix.
3056 2017-11-22 Jeff Law <law@redhat.com>
3058 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
3059 Use new method allocate_value_range rather than accessing the
3060 vrp_value_range_pool data member directly.
3061 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
3062 to use extract_range_from_stmt method to avoid need for
3063 extract_range_from_assignment method.
3064 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
3065 method rather than setting values_propgated data member directly.
3066 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
3067 and values propagated data members and extract_range_from_assignment
3068 method. Reorder private data members to conform to standards.
3069 Add new methods set_lattice_propagation_complete and
3070 allocate_value_range.
3072 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
3074 PR rtl-optimization/83030
3075 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
3076 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
3077 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
3078 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
3079 CROSSING_JUMP_P flag is already set before setting it.
3080 * cfgrtl.c (fixup_partition_crossing): Likewise.
3081 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
3084 2017-11-22 Jakub Jelinek <jakub@redhat.com>
3086 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
3087 Handle the case where both arguments are using gen_const_vec_series.
3089 2017-11-22 David Malcolm <dmalcolm@redhat.com>
3092 * pretty-print.c (pp_format): Move quoting implementation to
3093 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
3094 to pass address of "quote" local.
3095 (pp_begin_quote): New function.
3096 (pp_end_quote): New function.
3097 * pretty-print.h (printer_fn): Convert penultimate param from bool
3099 (pp_begin_quote): New decl.
3100 (pp_end_quote): New decl.
3101 * tree-diagnostic.c (default_tree_printer): Convert penultimate
3102 param from bool to bool *.
3103 * tree-diagnostic.h (default_tree_printer): Likewise.
3105 2017-11-22 Jeff Law <law@redhat.com>
3107 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
3108 Initialize vr_values.
3109 (evrp_range_analyzer::try_find_new_range): Call methods attached to
3110 vr_values via vr_values class instance rather than delegators.
3111 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
3112 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
3113 (evrp_range_analyzer::push_value_range): Likewise.
3114 (evrp_range_analyzer::pop_value_range): Likewise.
3115 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
3116 most delegators. Those remaining are exposed as public interfaces.
3117 Make vr_values a pointer and private.
3118 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
3120 (evrp_range_analyzer::get_vr_value): New method.
3121 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
3122 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
3123 (class evrp_dom_walker): Attach evrp_folder class, initialize
3124 it in the ctor. Remove temporary delegators.
3125 (evrp_dom_walker::before_dom_children): Call methods in attached
3126 evrp_range_analyzer class via class instance pointer. Use
3127 free value_range_constant_singleton to remove need for
3128 op_with_constant_singleton_value delegator method. Do not
3129 create a vrp_prop class instance for every call! Narrow
3130 scope of a couple variables.
3131 (evrp_dom_walker::cleanup): Call methods in attached
3132 evrp_range_analyzer class via class instance pointer.
3133 * vr-values.h (class vr_values): Privatize many methods and
3136 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3138 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
3139 * cfgexpand.c (expand_debug_expr): Remove handling for them.
3140 * expr.c (expand_expr_real_2): Likewise.
3141 * fold-const.c (const_unop): Likewise.
3142 * optabs-tree.c (optab_for_tree_code): Likewise.
3143 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3144 * tree-inline.c (estimate_operator_cost): Likewise.
3145 * tree-pretty-print.c (dump_generic_node): Likewise.
3146 (op_code_prio): Likewise.
3147 (op_symbol_code): Likewise.
3148 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
3149 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
3150 * internal-fn.c (direct_internal_fn_optab): New function.
3151 (direct_internal_fn_array, direct_internal_fn_supported_p
3152 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
3153 * fold-const-call.c (fold_const_reduction): New function.
3154 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
3156 * tree-vect-loop.c: Include internal-fn.h.
3157 (reduction_code_for_scalar_code): Rename to...
3158 (reduction_fn_for_scalar_code): ...this and return an internal
3160 (vect_model_reduction_cost): Take an internal_fn rather than
3162 (vect_create_epilog_for_reduction): Likewise. Build calls rather
3164 (vectorizable_reduction): Use internal functions rather than tree
3165 codes for the reduction operation. Update calls to the functions
3167 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
3168 Use calls to internal functions rather than REDUC tree codes.
3169 * config/aarch64/aarch64-simd.md: Update comment accordingly.
3171 2017-11-22 Olivier Hainque <hainque@adacore.com>
3173 * config/vxworks.c (vxworks_override_options): Pick default
3174 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
3175 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
3176 DWARF_GNAT_ENCODINGS_DEFAULT.
3177 * config/vxworksae.h: Likewise.
3179 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
3181 PR tree-optimization/83104
3182 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
3185 2017-11-22 Marek Polacek <polacek@redhat.com>
3186 H.J. Lu <hongjiu.lu@intel.com>
3187 Jason Merrill <jason@redhat.com>
3192 * calls.c (initialize_argument_information): Call
3193 warn_parameter_passing_abi target hook.
3194 (store_one_arg): Use 0 for empty record size. Don't push 0 size
3195 argument onto stack.
3196 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
3197 * common.opt: Update -fabi-version description.
3198 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
3199 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
3201 (ix86_is_empty_record): New function.
3202 (ix86_warn_parameter_passing_abi): New function.
3203 (TARGET_EMPTY_RECORD_P): Redefine.
3204 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
3205 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
3206 * doc/tm.texi: Regenerated.
3207 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
3208 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
3209 * dwarf2out.c (get_ultimate_context): Move to tree.c.
3210 * explow.c (hard_function_value): Call arg_int_size_in_bytes
3211 instead of int_size_in_bytes.
3212 * expr.c (copy_blkmode_to_reg): Likewise.
3213 * function.c (aggregate_value_p): Return 0 for empty types.
3214 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
3216 (locate_and_pad_parm): Call arg size_in_bytes instead
3218 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
3219 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
3220 * target.def (empty_record_p, warn_parameter_passing_abi): New target
3222 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
3223 (std_gimplify_va_arg_expr): Skip empty records. Call
3224 arg_size_in_bytes instead size_in_bytes.
3225 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
3226 * tree-core.h (tree_type_common): Add empty_flag.
3227 (tree_decl_common): Update comments.
3228 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
3230 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
3231 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
3233 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
3234 * tree.c (default_is_empty_type): New function.
3235 (default_is_empty_record): New function.
3236 (arg_int_size_in_bytes): New function.
3237 (arg_size_in_bytes): New function.
3238 (get_ultimate_context): New function.
3239 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
3240 TRANSLATION_UNIT_WARN_EMPTY_P.
3241 (default_is_empty_record, arg_int_size_in_bytes,
3242 arg_size_in_bytes, get_ultimate_context): Declare.
3244 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
3246 * config/arm/arm.c (cmse_clear_registers): New function.
3247 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
3248 code by call to cmse_clear_registers.
3249 (cmse_nonsecure_entry_clear_before_return): Likewise.
3251 2017-11-22 Tamar Christina <tamar.christina@arm.com>
3253 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
3254 (vdot_s32, vdotq_s32): New.
3255 (vdot_lane_u32, vdotq_lane_u32): New.
3256 (vdot_lane_s32, vdotq_lane_s32): New.
3259 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3262 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
3263 for unsigned values with fewer HWIs than the precision.
3264 (test_overflow): New function.
3265 (wide_int_cc_tests): Call it.
3267 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3269 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
3270 has the same mode class as Pmode.
3272 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
3274 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
3275 padding_bits_to_clear_ptr.
3276 (cmse_nonsecure_entry_clear_before_return): Likewise.
3278 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
3280 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
3281 auto_sbitap instead of integer bitfield to control register needing
3284 2017-11-22 Jakub Jelinek <jakub@redhat.com>
3286 PR tree-optimization/83044
3287 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
3288 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
3289 that need the upper bound. Subtract offset from
3290 get_addr_base_and_unit_offset only if positive and subtract it
3291 before division by eltsize rather than after it.
3294 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
3295 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
3296 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
3300 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
3301 expand_binop with *mul_widen_optab, make sure at least one of the
3302 operands doesn't have VOIDmode.
3305 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
3307 PR rtl-optimization/82044
3308 PR tree-optimization/82042
3309 * dse.c (record_store): Check for overflow.
3310 (check_mem_read_rtx): Properly check for overflow if width == -1, call
3311 add_wild_read instead of clear_rhs_from_active_local_stores on
3312 overflow and log it into dump_file.
3314 2017-11-22 Richard Biener <rguenther@suse.de>
3316 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
3317 fake edges to exit when looking for a place to insert.
3318 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
3320 (insert_into_preds_of_block): Commit edge insertion immediately,
3321 assert that doesn't require new BBs.
3322 (fini_pre): Release expressions.
3323 (pass_pre::execute): Shuffle things around a bit, if the fn
3324 is too large do not compute AVAIL either as this is really the
3327 2017-11-22 Richard Biener <rguenther@suse.de>
3329 PR tree-optimization/83089
3330 * tree-if-conv.c (pass_if_conversion::execute): If anything
3331 changed reset SCEV and free the number of iteration estimates.
3333 2017-11-21 Martin Sebor <msebor@redhat.com>
3335 PR tree-optimization/82945
3336 * calls.h (warn_nonstring_bound): Remove unused function.
3338 2017-11-21 Martin Sebor <msebor@redhat.com>
3340 PR tree-optimization/82945
3341 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
3342 * calls.h (maybe_warn_nonstring_arg): Declare new function.
3343 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
3345 (initialize_argument_information): Call maybe_warn_nonstring_arg.
3346 * calls.h (get_attr_nonstring_decl): Declare new function.
3347 * doc/extend.texi (attribute nonstring): Update.
3348 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
3349 get_attr_nonstring_decl and handle it.
3350 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
3351 detection of nul-termination.
3352 (strlen_to_stridx): Change to a pointer.
3353 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
3354 (pass_strlen::execute): Same.
3356 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
3358 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
3359 for the new option -mprefer-vector-width=[none|128|256|512].
3360 * config/i386/i386.c (ix86_target_string): remove old style options
3361 -mprefer-avx256 and make -mprefer-avx128 as alias.
3362 (ix86_option_override_internal): Apply defaults for the
3363 -mprefer-vector-width=[128|256] option.
3364 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
3365 Implement macros to work with -mprefer-vector-width=.
3366 * config/i386/i386.opt: Implemented option
3367 -mprefer-vector-width=[none|128|256|512].
3368 * doc/invoke.texi: Documentation for
3369 -mprefer-vector-width=[none|128|256|512].
3371 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
3373 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
3375 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
3377 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
3379 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
3381 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
3383 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
3384 * expr.c (expand_expr_real_2): Likewise.
3385 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
3386 fold_binary_loc): Likewise.
3387 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
3388 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
3389 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
3390 MINUS_EXPR transformations.
3391 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
3392 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
3393 * tree-inline.c (estimate_operator_cost): Likewise.
3394 * tree-pretty-print.c (dump_generic_node, op_code_prio,
3395 op_symbol_code): Likewise.
3396 * tree-vect-stmts.c (vectorizable_operation): Likewise.
3397 * vr-values.c (extract_range_from_binary_expr): Likewise.
3398 * varasm.c (initializer_constant_valid_p_1): Likewise.
3399 * tree.def: New tree code POINTER_DIFF_EXPR.
3401 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
3403 * config/i386/i386.md (*bswap<mode>2_movbe): Add
3404 integer suffix to movbe mnemonic.
3405 (*bswaphi2_movbe): Ditto.
3406 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
3408 2017-11-21 Jakub Jelinek <jakub@redhat.com>
3411 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
3412 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
3413 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
3415 2017-11-21 Martin Liska <mliska@suse.cz>
3417 * tree-inline.c (expand_call_inline): Remove not needed
3420 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
3421 Jakub Jelinek <jakub@redhat.com>
3424 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
3425 Remove static keyword from f variable.
3427 2017-11-21 Jakub Jelinek <jakub@redhat.com>
3429 PR tree-optimization/83086
3430 * gimple-ssa-store-merging.c
3431 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
3432 rather than n.base_addr.
3434 2017-11-21 Martin Liska <mliska@suse.cz>
3436 PR rtl-optimization/82044
3437 PR tree-optimization/82042
3438 * dse.c (check_mem_read_rtx): Check for overflow.
3440 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
3442 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
3445 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
3447 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
3448 and its corresponding call arg location note.
3450 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
3451 Andrew Burgess <andrew.burgess@embecosm.com>
3453 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
3455 (arc_return_slot_offset): Likewise.
3456 (arc_eh_return_address_location): New declaration.
3457 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
3458 (MUST_SAVE_REGISTER): Add exception handler case.
3459 (MUST_SAVE_RETURN_ADDR): Likewise.
3460 (arc_frame_pointer_required): Likewise.
3461 (arc_frame_pointer_needed): New function.
3462 (arc_compute_frame_size): Changed.
3463 (arc_expand_prologue): Likewise.
3464 (arc_expand_epilogue): Likewise.
3465 (arc_initial_elimination_offset): Likewise.
3466 (arc_return_slot_offset): Delete.
3467 (arc_eh_return_address_location): New function.
3468 (arc_builtin_setjmp_frame_value): Likewise.
3469 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
3470 (EH_RETURN_STACKADJ_RTX): Define.
3471 (EH_RETURN_HANDLER_RTX): Likewise.
3472 * config/arc/arc.md (eh_return): Delete.
3474 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
3476 * print-tree.h (debug_vec_tree): Remove prototype.
3477 * gdbinit.in (pvt): Remove macro.
3479 2017-11-21 Jakub Jelinek <jakub@redhat.com>
3481 PR tree-optimization/83047
3482 * gimple-ssa-store-merging.c
3483 (imm_store_chain_info::output_merged_store): If the loads with the
3484 same vuse are in different basic blocks, for load_gsi pick a load
3485 location that is dominated by the other loads.
3488 * config/i386/i386.c (ix86_memmodel_check): Start
3489 -Winvalid-memory-model diagnostics with lowercase letter.
3492 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
3493 set in_cold_section_p to the partition containing loc_list->first.
3494 When seeing loc_list->last_before_switch node, update secname and
3495 perform range_across_switch second partition handling only after that.
3498 * run-rtl-passes.c: Include debug.h.
3499 (run_rtl_passes): Call debug_hooks->assembly_start.
3500 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
3504 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
3505 OPTAB_DIRECT in calls to expand_simple_binop.
3507 2017-11-20 David Malcolm <dmalcolm@redhat.com>
3510 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
3512 2017-11-20 David Malcolm <dmalcolm@redhat.com>
3515 * spellcheck.h (best_match::blithely_get_best_candidate): New
3518 2017-11-20 Jakub Jelinek <jakub@redhat.com>
3520 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
3521 with lower case letter.
3523 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
3525 * config/i386/i386.md (bswaphi2): New expander.
3526 (*bswaphi2_movbe): New insn pattern.
3527 (bswaphi -> rorhi pepehole2): New peephole pattern.
3529 2017-11-20 Jeff Law <law@redhat.com>
3531 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
3532 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
3533 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
3534 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
3535 gimple-ssa-evrp-analyze.h.
3537 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
3538 set BB_VISITED here.
3539 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
3541 2017-11-20 Bin Cheng <bin.cheng@arm.com>
3543 * tree-predcom.c: Add general comment on Store-Store chains.
3544 (split_data_refs_to_components): Postpone clearing eliminate_store_p
3546 (get_chain_last_ref_at): Rename into...
3547 (get_chain_last_write_at): ...this.
3548 (get_chain_last_write_before_load): New function.
3549 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
3550 CT_STORE_STORE when write reference is added.
3551 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
3552 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
3553 (initialize_root_vars_store_elim_1): Ditto.
3554 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
3555 definition is created.
3556 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
3557 chain by replacing it with dominant stored value.
3559 2017-11-20 Bin Cheng <bin.cheng@arm.com>
3561 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
3563 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
3565 * vr-values.c (extract_range_from_binary_expr): Use a full range
3568 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
3570 * config/arm/arm.md (R4_REGNUM): Define constant.
3571 (nonsecure_call_internal): Remove r4 clobber.
3572 (nonsecure_call_value_internal): Likewise.
3573 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
3574 clobber and resequence match_operands.
3575 (nonsecure_call_value_reg_thumb1_v5): Likewise.
3576 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
3577 (nonsecure_call_value_reg_thumb2): Likewise.
3579 2017-11-20 Jakub Jelinek <jakub@redhat.com>
3581 PR tree-optimization/78821
3582 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
3583 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
3584 to the address of the base rather than the base itself.
3585 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
3586 (find_bswap_or_nop_finalize): New function.
3587 (find_bswap_or_nop): Use it.
3588 (bswap_replace): Return a tree rather than bool, change first
3589 argument from gimple * to gimple_stmt_iterator, allow inserting
3590 into an empty sequence, allow ins_stmt to be NULL - then emit
3591 all stmts into gsi. Fix up MEM_REF address gimplification.
3592 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
3593 (struct store_immediate_info): Add N and INS_STMT non-static
3595 (store_immediate_info::store_immediate_info): Initialize them
3596 from newly added ctor args.
3597 (merged_store_group::apply_stores): Formatting fixes. Sort by
3599 (stmts_may_clobber_ref_p): For stores call also
3600 refs_anti_dependent_p.
3601 (gather_bswap_load_refs): New function.
3602 (imm_store_chain_info::try_coalesce_bswap): New method.
3603 (imm_store_chain_info::coalesce_immediate_stores): Use it.
3604 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
3605 (imm_store_chain_info::output_merged_store): Fail if number of
3606 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
3607 and NOP_EXPR rhs_code.
3608 (pass_store_merging::process_store): Compute n and ins_stmt, if
3609 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
3610 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
3612 (pass_store_merging::execute): Calculate dominators.
3614 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
3615 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
3616 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
3617 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3618 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
3619 class pass_optimize_bswap, bswap_replace,
3620 pass_optimize_bswap::execute): Moved to ...
3621 * gimple-ssa-store-merging.c: ... this file.
3622 Include optabs-tree.h.
3623 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
3624 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3625 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
3626 anonymous namespace, remove static keywords.
3627 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
3628 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
3630 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
3633 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
3635 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
3637 * vec.h (debug_helper): New function.
3638 (DEFINE_DEBUG_VEC): New macro.
3639 * hash-set.h (debug_helper): New function.
3640 (DEFINE_DEBUG_HASH_SET): New macro.
3641 * cfg.c (debug_slim (edge)): New function.
3642 Call DEFINE_DEBUG_VEC for edges.
3643 Call DEFINE_DEBUG_HASH_SET for edges.
3644 * cfghooks.c (debug_slim (basic_block)): New function.
3645 Call DEFINE_DEBUG_VEC for basic blocks.
3646 Call DEFINE_DEBUG_HASH_SET for basic blocks.
3647 * print-tree.c (debug_slim): New function to handle trees.
3648 Call DEFINE_DEBUG_VEC for trees.
3649 Call DEFINE_DEBUG_HASH_SET for trees.
3650 (debug (vec<tree, va_gc>) &): Remove.
3651 (debug (<vec<tree, va_gc>) *): Remove.
3652 * print-rtl.c (debug_slim): New function to handle const_rtx.
3653 Call DEFINE_DEBUG_VEC for rtx_def.
3654 Call DEFINE_DEBUG_VEC for rtx_insn.
3655 Call DEFINE_DEBUG_HASH_SET for rtx_def.
3656 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
3657 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
3658 (debug (vec<rtx_insn *> *ptr): Remove.
3659 (debug_insn_vector): Remove.
3660 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
3662 2017-11-20 Tom de Vries <tom@codesourcery.com>
3664 PR rtl-optimization/82020
3665 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
3666 IF_THEN_ELSE condition.
3668 2017-11-19 Jeff Law <law@redhat.com>
3670 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
3671 of degenerates resulting from ignoring an edge.
3673 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3676 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
3678 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3681 * profile-count.c (profile_count::to_sreal_scale): Fix return value
3682 for uninitialied counts.
3684 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3687 * tree-inline.c (estimate_num_insns): Set to 1 at least.
3689 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3692 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
3694 2017-11-19 Tom de Vries <tom@codesourcery.com>
3696 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
3699 2017-11-19 Tom de Vries <tom@codesourcery.com>
3701 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
3704 2017-11-19 Tom de Vries <tom@codesourcery.com>
3706 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
3709 2017-11-19 Tom de Vries <tom@codesourcery.com>
3711 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
3713 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
3715 2017-11-19 Tom de Vries <tom@codesourcery.com>
3717 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
3719 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
3720 semicolon after MCORE_EXPORT_NAME call.
3722 2017-11-19 Tom de Vries <tom@codesourcery.com>
3725 * vmsdbgout.c (vmsdbgout_early_finish): New function.
3726 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
3728 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
3730 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
3731 (cgraph_edge::clone): Cleanup updating of profile.
3732 * ipa-cp.c (update_profiling_info): Likewise.
3733 * ipa-inline-transform.c (inline_transform): Likewise.
3734 * ipa-inline.c (inline_small_functions): Add missing space to dump.
3735 * ipa-split.c (execute_split_functions): Do not split when function
3737 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
3738 * profile-count.c (profile_count::dump): Add global0.
3739 (profile_count::to_cgraph_frequency): Do not ICE when entry is
3741 (profile_count::to_sreal_scale): Likewise.
3742 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
3743 (profile_count::combine_with_ipa_count): New function.
3744 * profile-count.h (profile_guessed_global0adjusted): New.
3745 (profile_count::adjusted_zero): New.
3746 (profile_count::global0adjusted): New.
3747 (profile_count::combine_with_ipa_count): New.
3748 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
3749 correct profile of return block of split functions.
3750 (copy_cfg_body): Remove unused profile_count.
3751 (copy_body): Likewise.
3752 (expand_call_inline): Update.
3753 (tree_function_versioning): Update.
3755 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
3757 * hash-set.h (hash_set::empty): New.
3758 * tree-ssa-threadbackward.h: Delete.
3759 * tree-ssa-threadbackward.c (class thread_jumps): New.
3760 Move max_threaded_paths into class.
3761 (fsm_find_thread_path): Remove arguments that are now in class.
3762 (profitable_jump_thread_path): Rename to...
3763 (thread_jumps::profitable_jump_thread_path): ...this.
3764 (convert_and_register_jump_thread_path): Rename to...
3765 (thread_jumps::convert_and_register_current_path): ...this.
3766 (check_subpath_and_update_thread_path): Rename to...
3767 (thread_jumps::check_subpath_and_update_thread_path): ...this.
3768 (register_jump_thread_path_if_profitable): Rename to...
3769 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
3770 (handle_phi): Rename to...
3771 (thread_jumps::handle_phi): ...this.
3772 (handle_assignment): Rename to...
3773 (thread_jumps::handle_assignment): ...this.
3774 (fsm_find_control_statement_thread_paths): Rename to...
3775 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
3776 (find_jump_threads_backwards): Rename to...
3777 (thread_jumps::find_jump_threads_backwards): ...this.
3778 Initialize path local data.
3779 (pass_thread_jumps::execute): Call find_jump_threads_backwards
3780 from within thread_jumps class.
3781 (pass_early_thread_jumps::execute): Same.
3783 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3785 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
3787 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3789 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
3790 to canonical location.
3792 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3795 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
3797 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
3799 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
3801 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
3802 rcrt1.o%s/grcrt1.o%s for -static-pie.
3804 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3806 * i386.c (ix86_multiplication_cost, ix86_division_cost,
3807 ix86_shift_rotate_cost): Break out from ...
3808 (ix86_rtx_costs): ... here.
3809 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
3812 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3814 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
3815 when entry block was promoted unlikely.
3816 (estimate_bb_frequencies): Increase frequency scale.
3817 * profile-count.h (profile_count): Export precision info.
3819 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3821 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
3822 disturbin profile of entry block.
3824 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3826 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
3829 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3831 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
3832 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
3834 2017-11-17 Jeff Law <law@redhat.com>
3836 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
3837 from evrp_dom_walker class. Various methods moved into new class.
3838 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
3839 (evrp_range_analyzer::enter): New method.
3840 (evrp_range_analyzer::leave): New method.
3841 (evrp_dom_walker): Remove delegators no longer needed by this class.
3842 Replace vr_values data member with evrp_range_analyzer
3844 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
3845 method extracted from evrp_dom_walker::before_dom_children.
3846 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
3847 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
3849 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
3850 Add private copy constructor and move assignment operators.
3851 Privatize methods and class data where trivially possible.
3852 (evrp_dom_walker::cleanup): New function, extracted from
3853 execute_early_vrp. Simplify access to class data.
3855 * vr-values.h (get_output_for_vrp): Prototype.
3856 * vr-values.c (get_output_for_vrp): New function extracted from
3857 vrp_visit_assignment_or_call and extract_range_from_stmt.
3858 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
3860 2017-11-17 Luis Machado <luis.machado@linaro.org>
3862 * config/aarch64/aarch64.c
3863 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
3864 (qdf24xx_tunings) <autoprefetcher_model>: Set to
3865 tune_params::AUTOPREFETCHER_WEAK.
3867 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3870 * config/arm/arm.c (arm_valid_target_attribute_rec):
3871 Parse "arch=" and "+<ext>".
3872 (arm_valid_target_attribute_tree): Re-init global options.
3873 (arm_option_override): Make non-static.
3874 (arm_options_perform_arch_sanity_checks): Make errors fatal.
3875 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
3876 (__ARM_FEATURE_CRC32): Support undef.
3877 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
3878 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
3880 2017-11-17 David Malcolm <dmalcolm@redhat.com>
3882 * gdbinit.in (break-on-diagnostic): New command.
3884 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3886 * config/i386/i386.c (ix86_expand_epilogue): Change simple
3887 return to indirect jump for EH return if control-flow
3888 protection is enabled. Change explicit 'false' argument in
3889 pro_epilogue_adjust_stack with a value of
3891 * config/i386/i386.md (simple_return_indirect_internal):
3892 Remove SImode restriction to support 64-bit.
3894 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3896 * combine.c (added_notes_insn): New.
3897 (try_combine): Handle added_notes_insn like added_links_insn.
3898 Rewrite return value code.
3899 (distribute_notes): Set added_notes_insn to the earliest insn we added
3902 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3904 PR rtl-optimization/82621
3905 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
3906 dest of one of those SETs is unused.
3908 2017-11-17 Richard Biener <rguenther@suse.de>
3911 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
3912 * tree-pretty-print.c (dump_generic_node): Handle
3913 annot_expr_parallel_kind.
3914 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
3915 * gimplify.c (gimple_boolify): Likewise.
3917 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3919 * config.gcc (extra_headers): Add cet.h for x86 targets.
3920 * config/i386/cet.h: New file.
3921 * doc/install.texi: Add --enable-cet/--disable-cet.
3923 2017-11-17 Richard Biener <rguenther@suse.de>
3925 PR tree-optimization/83017
3926 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
3927 (gen_parallel_loop): Properly count iterations.
3928 (parallelize_loops): Handle loop->can_be_parallel independent
3929 of flag_loop_parallelize_all. Make static profitability test match
3931 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
3932 * invoke.texi (parloops-min-per-thread): Document.
3934 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
3936 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
3937 upstreaming review comments.
3939 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3941 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
3942 with fast unaligned access.
3943 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
3945 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
3947 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
3948 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
3950 2017-11-17 Richard Biener <rguenther@suse.de>
3952 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
3953 folding of references.
3955 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
3958 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
3959 of replacing call to strncmp with corresponding call to strcmp when
3962 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
3964 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
3965 option prefer-avx256 for skylake-avx512 configuration.
3966 * config/i386/i386.c (ix86_option_override_internal): Ditto.
3967 (get_builtin_code_for_version): Ditto.
3969 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3970 Monk Chiang <sh.chiang04@gmail.com>
3972 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
3973 (FIXED_REGISTERS): Reserve more register numbers.
3974 (CALL_USED_REGISTERS): Likewise.
3975 (REG_ALLOC_ORDER): Likewise.
3976 (REG_CLASS_CONTENTS): Likewise.
3977 (REGISTER_NAMES): Likewise.
3979 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3980 Kito Cheng <kito.cheng@gmail.com>
3982 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
3984 * config/nds32/iterators.md: Add vector mode iterators and attributes.
3986 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
3988 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
3989 parameter list for vec_splats.
3991 2017-11-16 Joseph Myers <joseph@codesourcery.com>
3993 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
3995 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
3997 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
4000 * doc/install.texi: Document incompatibility of Perl >=5.6.26
4001 with the required version of automake 1.11.6.
4003 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
4005 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
4007 (DU_C2_3_power9): Correct reservation combinations.
4008 (FP_DIV_power9, VEC_DIV_power9): New.
4009 (power9-alu): Split out rotate/shift...
4010 (power9-rot): ...to here, correct dispatch resource.
4011 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
4013 (power9-fp): Correct latency.
4014 (power9-sdiv): Add div/sqrt resource.
4015 (power9-ddiv): Correct latency, add div/sqrt resource.
4016 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
4017 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
4019 (power9-qpdiv, power9-qpmul): Adjust resource usage.
4021 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
4023 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
4024 switch statement mapping KF built-ins to TF built-ins if we don't
4025 have the proper ISA 3.0 assembler support.
4027 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4029 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
4030 (lower_emutls_function_body): Do not compute it.
4032 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4034 * ipa-split.c (split_bb_info): Turn time to sreal.
4035 (split_point): Likewise.
4036 (dump_split_point): Likewise.
4037 (fine_split_points): Likewise.
4038 (execute_split_functions): Only zero split_bbs; turn time to sreals.
4040 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4042 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
4045 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4047 * predict.c (combine_predictions_for_bb): Preserve zero predicted
4049 (expensive_function_p): Remove useless assert.
4050 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
4052 2017-11-16 Martin Sebor <msebor@redhat.com>
4054 PR tree-optimization/82588
4055 PR tree-optimization/82583
4056 * tree-vrp.c (check_array_ref): Handle flexible array members,
4057 string literals, and inner indices.
4058 (search_for_addr_array): Add detail to diagnostics.
4060 2017-11-16 Nathan Sidwell <nathan@acm.org>
4064 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
4065 (SET_DECL_ASSEMBLER_NAME): Forward to
4066 overwrite_decl_assembler_name.
4067 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
4068 (overwrite_decl_assembler_name): Declare.
4069 * tree.c (overwrite_decl_assembler_name): New.
4070 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
4071 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
4072 (LANG_HOOKS_INITIALIZER): Add it.
4073 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
4074 * langhooks.c (lhd_set_decl_assembler_name): Use
4075 SET_DECL_ASSEMBLER_NAME.
4076 (lhd_overwrite_decl_assembler_name): Default implementation.
4078 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
4079 Jackson Woodruff <jackson.woodruff@arm.com>
4081 PR tree-optimization/71026
4082 * match.pd: Canonicalize constant multiplies in division.
4084 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
4086 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
4087 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
4088 Remove OPT_fomit_frame_pointer entry.
4089 * common/config/arc/arc-common.c: Likewise.
4090 * common/config/arm/arm-common.c: Likewise.
4091 * common/config/avr/avr-common.c: Likewise.
4092 * common/config/c6x/c6x-common.c: Likewise.
4093 * common/config/cr16/cr16-common.c: Likewise.
4094 * common/config/cris/cris-common.c: Likewise.
4095 * common/config/epiphany/epiphany-common.c: Likewise.
4096 * common/config/fr30/fr30-common.c: Likewise.
4097 * common/config/frv/frv-common.c: Likewise.
4098 * common/config/ia64/ia64-common.c: Likewise.
4099 * common/config/iq2000/iq2000-common.c: Likewise.
4100 * common/config/lm32/lm32-common.c: Likewise.
4101 * common/config/m32r/m32r-common.c: Likewise.
4102 * common/config/mcore/mcore-common.c: Likewise.
4103 * common/config/microblaze/microblaze-common.c: Likewise.
4104 * common/config/mips/mips-common.c: Likewise.
4105 * common/config/mmix/mmix-common.c: Likewise.
4106 * common/config/mn10300/mn10300-common.c: Likewise.
4107 * common/config/nios2/nios2-common.c: Likewise.
4108 * common/config/pa/pa-common.c: Likewise.
4109 * common/config/pdp11/pdp11-common.c: Likewise.
4110 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
4111 * common/config/riscv/riscv-common.c: Likewise.
4112 * common/config/rs6000/rs6000-common.c: Likewise.
4113 * common/config/rx/rx-common.c: Likewise.
4114 * common/config/s390/s390-common.c: Likewise.
4115 * common/config/sh/sh-common.c: Likewise.
4116 * common/config/sparc/sparc-common.c: Likewise.
4117 * common/config/tilegx/tilegx-common.c: Likewise.
4118 * common/config/tilepro/tilepro-common.c: Likewise.
4119 * common/config/v850/v850-common.c: Likewise.
4120 * common/config/visium/visium-common.c: Likewise.
4121 * common/config/xstormy16/xstormy16-common.c: Likewise.
4122 * common/config/xtensa/xtensa-common.c: Likewise.
4123 * invoke.texi (-fomit-frame-pointer): Update documentation.
4125 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4127 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
4129 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4131 * cfg.c (scale_bbs_frequencies_int,
4132 cale_bbs_frequencies_gcov_type): Remove.
4133 * cfg.h (scale_bbs_frequencies_int,
4134 cale_bbs_frequencies_gcov_type): Remove.
4136 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4138 * tree-ssa-loop-manip.c
4139 (scale_dominated_blocks_in_loop): Update to profile counts.
4140 (tree_transform_and_unroll_loop): Likewise.
4142 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4144 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
4145 scale_bbs_frequencies_int.
4147 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4149 * final.c (compute_alignments): Use counts rather than frequencies.
4151 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4153 * cfgloopanal.c: Include sreal.h
4154 (average_num_loop_insns): Use counts and sreal for accounting.
4156 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4158 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
4161 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4162 Kito Cheng <kito.cheng@gmail.com>
4164 * config/nds32/constraints.md: Provide more constraints.
4165 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
4166 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
4167 support constraints usage.
4169 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4171 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
4173 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4174 Kito Cheng <kito.cheng@gmail.com>
4176 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
4177 * config/nds32/nds32.opt: Refine the layout.
4178 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
4179 TARGET_EXT_STRING): Support new options.
4180 * config/nds32/nds32.h: Likewise.
4181 * config/nds32/nds32.md: Likewise.
4182 * config/nds32/nds32-predicates.c: Likewise.
4183 * config/nds32/constraints.md: Likewise.
4184 * common/config/nds32/nds32-common.c: Likewise.
4186 2017-11-16 Julia Koval <julia.koval@intel.com>
4189 * config/i386/gfniintrin.h: Add sse check.
4190 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
4192 2017-11-16 Julia Koval <julia.koval@intel.com>
4194 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
4195 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
4196 (ix86_handle_option): Handle -mavx512vbmi2.
4197 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
4198 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4199 * config/i386/i386-c.c (__AVX512VBMI2__): New.
4200 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
4201 (ix86_valid_target_attribute_inner_p): Ditto.
4202 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
4203 * config/i386/i386.opt (mavx512vbmi2): New option.
4204 * doc/invoke.texi: Add new option.
4206 2017-11-16 Julia Koval <julia.koval@intel.com>
4208 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
4209 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
4210 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
4211 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
4212 _mm512_gf2p8mul_epi8): New intrinsics.
4213 * config/i386/i386-builtin-types.def
4214 (V64QI_FTYPE_V64QI_V64QI): New type.
4215 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
4216 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
4217 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
4218 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
4219 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
4220 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4222 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
4224 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
4225 explicitly as a stream of bytes.
4227 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4229 * config/rs6000/altivec.h (vec_xst_be): New #define.
4230 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
4231 and externalize from *altivec_vperm_<mode>_internal.
4232 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
4234 (XL_BE_V8HI): Likewise.
4235 (XL_BE_V4SI): Likewise.
4236 (XL_BE_V4SI): Likewise.
4237 (XL_BE_V2DI): Likewise.
4238 (XL_BE_V4SF): Likewise.
4239 (XL_BE_V2DF): Likewise.
4240 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
4241 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
4242 all array entries with these keys: VSX_BUILTIN_VEC_XL,
4243 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
4244 VSX_BUILTIN_VEC_XST_BE.
4245 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
4246 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
4248 (altivec_init_builtins): Replace conditional calls to def_builtin
4249 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
4250 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
4251 with unconditional calls. Remove calls to def_builtin for
4252 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
4253 __builtin_vec_xst_be.
4254 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
4255 to define_expand, and add alternate RTL generation for P8.
4256 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
4257 vsx_ld_elemrev_v8hi.
4258 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
4259 add alternate RTL generation for P8.
4260 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
4261 vsx_ld_elemrev_v16qi.
4262 (vsx_st_elemrev_v8hi): Convert define_insn
4263 to define_expand, and add alternate RTL generation for P8.
4264 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
4265 vsx_st_elemrev_v8hi.
4266 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
4267 add alternate RTL generation for P8.
4268 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
4269 vsx_st_elemrev_v16qi.
4271 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
4274 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
4275 TARGET_AVX512ER check.
4276 (ix86_option_override_internal): Set MASK_VZEROUPPER if
4277 neither -mzeroupper nor -mno-zeroupper is used and
4278 TARGET_EMIT_VZEROUPPER is set.
4279 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
4280 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
4282 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
4284 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4285 folding of vector compares.
4286 (fold_build_vec_cmp): New helper function.
4287 (fold_compare_helper): New helper function.
4288 (builtin_function_type): Add compare builtins to the list of functions
4289 having unsigned arguments. Cosmetic updates to comment indentation.
4290 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
4291 the not+eq combination.
4293 2017-11-15 Bin Cheng <bin.cheng@arm.com>
4295 PR tree-optimization/82726
4296 PR tree-optimization/70754
4297 * tree-predcom.c (order_drefs_by_pos): New function.
4298 (combine_chains): Move code setting has_max_use_after to...
4299 (try_combine_chains): ...here. New parameter. Sort combined chains
4300 according to position information.
4301 (tree_predictive_commoning_loop): Update call to above function.
4302 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
4304 2017-11-15 Bin Cheng <bin.cheng@arm.com>
4306 PR tree-optimization/82726
4308 2017-01-23 Bin Cheng <bin.cheng@arm.com>
4310 PR tree-optimization/70754
4311 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
4312 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
4313 combined stmt before it if not NULL.
4314 (combine_chains): Process refs reversely and compute dominance point
4318 2017-02-23 Bin Cheng <bin.cheng@arm.com>
4320 PR tree-optimization/79663
4321 * tree-predcom.c (combine_chains): Process refs in reverse order
4322 only for ZERO length chains, and add explaining comment.
4324 2017-11-15 Tamar Christina <tamar.christina@arm.com>
4326 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
4327 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
4328 * doc/invoke.texi (ARM Options): Add armv8.3-a.
4330 2017-11-15 Tamar Christina <tamar.christina@arm.com>
4332 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
4334 2017-11-15 Martin Liska <mliska@suse.cz>
4336 * tree-cfg.c (pass_warn_function_return::execute):
4337 Compare warn_return_type for greater than zero.
4339 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
4343 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
4344 to return true on Xeon and not on Xeon Phi.
4345 (ix86_check_avx256_register): Changed to ...
4346 (ix86_check_avx_upper_register): ... this. Add extra check for
4347 VALID_AVX512F_REG_OR_XI_MODE.
4348 (ix86_avx_u128_mode_needed): Changed
4349 ix86_check_avx256_register to ix86_check_avx_upper_register.
4350 (ix86_check_avx256_stores): Changed to ...
4351 (ix86_check_avx_upper_stores): ... this. Changed
4352 ix86_check_avx256_register to ix86_check_avx_upper_register.
4353 (ix86_avx_u128_mode_after): Changed
4354 avx_reg256_found to avx_upper_reg_found. Changed
4355 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
4356 (ix86_avx_u128_mode_entry): Changed
4357 ix86_check_avx256_register to ix86_check_avx_upper_register.
4358 (ix86_avx_u128_mode_exit): Ditto.
4359 * config/i386/i386.h: (host_detect_local_cpu): New define.
4361 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
4363 * config/arm/xgene1.md (xgene1): Split into automatons
4364 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
4365 (xgene1_f_load): Adjust reservations and/or types.
4366 (xgene1_f_store): Likewise.
4367 (xgene1_load_pair): Likewise.
4368 (xgene1_store_pair): Likewise.
4369 (xgene1_fp_load1): Likewise.
4370 (xgene1_load1): Likewise.
4371 (xgene1_store1): Likewise.
4372 (xgene1_move): Likewise.
4373 (xgene1_alu): Likewise.
4374 (xgene1_simd): Likewise.
4375 (xgene1_bfm): Likewise.
4376 (xgene1_neon_load1): Likewise.
4377 (xgene1_neon_store1): Likewise.
4378 (xgene1_neon_logic): Likewise.
4379 (xgene1_neon_st1): Likewise.
4380 (xgene1_neon_ld1r): Likewise.
4381 (xgene1_alu_cond): Added.
4382 (xgene1_shift_reg): Likwise.
4383 (xgene1_bfx): Likewise.
4384 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
4386 2017-11-15 Jakub Jelinek <jakub@redhat.com>
4389 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
4391 (can_widen_mult_without_libcall): New function.
4392 (expand_mul_overflow): If only checking unsigned mul overflow,
4393 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
4394 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
4395 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
4396 (expand_DIVMOD): Formatting fix.
4397 * expmed.h (expand_mult): Add NO_LIBCALL argument.
4398 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
4399 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
4401 PR tree-optimization/82977
4402 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
4403 constructed temporary to strlen_to_stridx.put.
4405 2017-11-15 Martin Liska <mliska@suse.cz>
4407 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
4408 * configure: Regenerate.
4410 2017-11-15 Martin Liska <mliska@suse.cz>
4413 * config/sh/sh-mem.cc: Use proper probability for
4416 2017-11-14 Jeff Law <law@redhat.com>
4418 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
4419 the red zone for stack_clash_protection_final_dynamic_probe targets
4420 when the total dynamic stack size is zero bytes.
4422 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
4423 blocks is post order.
4425 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
4427 * dumpfile.h (TDF_COMPARE_DEBUG): New.
4428 * final.c (rest_of_clean_state): Set it for the
4429 -fcompare-debug dump.
4430 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
4431 class when TDF_COMPARE_DEBUG is set.
4433 * dwarf2out.c (gen_producer_string): Discard
4436 2017-11-15 Joseph Myers <joseph@codesourcery.com>
4439 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
4440 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
4441 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
4442 (__TGMATH_REAL_2_3): Remove macros.
4443 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
4444 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
4446 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
4448 (remquo): Define using __TGMATH_REAL_3.
4450 2017-11-14 Jeff Law <law@redhat.com>
4452 * vr-values.c: New file with contents extracted from tree-vrp.c.
4453 * Makefile.in (OBJS): Add vr-values.o
4454 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
4455 (set_value_range, set_and_canonicalize_value_range): Likewise.
4456 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
4457 (value_range_constant_singleton, symbolic_range_p): Likewise.
4458 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
4459 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
4460 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
4461 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
4462 (find_case_label_range, find_case_label_index): Likewise.
4463 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
4464 (range_int_cst_singleton_p, value_inside_range): Likewise.
4465 (get_single_symbol): Likewise.
4466 (switch_update): Move structure definition here.
4467 (to_remove_edges, to_update_switch_stmts): Provide externs.
4468 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
4469 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
4470 (vrp_val_is_min, set_value_range): Likewise.
4471 (set_and_canonicalize_value_range, copy_value_range): Likewise.
4472 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
4473 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
4474 (range_is_nonnull, range_int_cst_p): Likewwise.
4475 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
4476 (get_single_symbol, operand_less_p): Likewise
4477 (compare_values_warnv, compare_values): Likewise.
4478 (value_inside_range, value_range_constant_singleton): Likewise.
4479 (zero_nonzero_bitgs_from_vr): Likewise.
4480 (extract_range_from_binary_expr_1): Likewise.
4481 (overflow_comparison_p): Likewise.
4482 (to_remove_edges, to_update_switch_stmts): Likewise.
4483 (find_case_label-index, find_case_label_range): Likewise.
4484 (switch_update, set_value_range_to_nonnegative): Remove.
4485 (set_value_range_to_truthvalue): Likewise.
4486 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
4487 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
4488 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
4489 (find_case_label_ranges, test_for_singularity): Likewise.
4490 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
4491 (x_vr_values): Move to its remaining use site.
4493 2017-11-10 Jeff Law <law@redhat.com>
4495 * vr-values.h (VR_INITIALIZER): Move #define here.
4496 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
4497 * Makefile.in (OBJS): Add tree-evrp.o
4498 * tree-vrp.h (assert_info): Move structure definition here.
4499 (set_value_range_to_varying): Prototype.
4500 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
4501 (infer_value_range, register_edge_assert_for): Likewise.
4502 (stmt_interesting_for_vrp): Likewise.
4503 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
4504 (set_value_range_to_varying): No longer static.
4505 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
4506 (infer_value_range, register_edge_assert_for): Likewise.
4508 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
4510 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
4511 generate the XXBRD instruction.
4513 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
4514 (rs6000_builtin_type_compatible): Treat _Float128 and long double
4515 as being compatible if -mabi=ieeelongdouble.
4516 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
4517 to setup float128 built-ins with hardware support.
4518 (BU_FLOAT128_HW_2): Likewise.
4519 (BU_FLOAT128_HW_3): Likewise.
4520 (BU_FLOAT128_HW_VSX_1): Likewise.
4521 (BU_FLOAT128_HW_VSX_2): Likewise.
4522 (scalar_extract_expq): Change float128 built-in functions to
4523 accommodate having both KFmode and TFmode functions. Use the
4524 KFmode variant as the default.
4525 (scalar_extract_sigq): Likewise.
4526 (scalar_test_neg_qp): Likewise.
4527 (scalar_insert_exp_q): Likewise.
4528 (scalar_insert_exp_qp): Likewise.
4529 (scalar_test_data_class_qp): Likewise.
4530 (sqrtf128_round_to_odd): Delete processing the round to odd
4531 built-in functions as special built-in functions, and define them
4532 as float128 built-ins. Use the KFmode variant as the default.
4533 (truncf128_round_to_odd): Likewise.
4534 (addf128_round_to_odd): Likewise.
4535 (subf128_round_to_odd): Likewise.
4536 (mulf128_round_to_odd): Likewise.
4537 (divf128_round_to_odd): Likewise.
4538 (fmaf128_round_to_odd): Likewise.
4539 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
4540 support for KFmode and TFmode xststdcqp calls.
4541 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
4542 point, switch the built-in handlers for the get/set float128
4543 exponent, get float128 mantissa, float128 test built-ins, and the
4544 float128 round to odd built-in functions. Eliminate creating the
4545 float128 round to odd built-in functions as special built-ins.
4546 (rs6000_init_builtins): Eliminate special creation of the float128
4547 round to odd built-in functions.
4548 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
4549 function insns to support both TFmode and KFmode variants.
4550 (xsxsigqp_<mode>): Likewise.
4551 (xsiexpqpf_<mode>): Likewise.
4552 (xsiexpqp_<mode>): Likewise.
4553 (xststdcqp_<mode>): Likewise.
4554 (xststdcnegqp_<mode>): Likewise.
4555 (xststdcqp_<mode>): Likewise.
4557 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4559 * tree-ssa-threadupdate.c (compute_path_counts): Remove
4560 unused path_in_freq_ptr parameter.
4561 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
4563 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4565 * ipa-inline.c (edge_badness): Dump sreal frequency.
4566 (compute_inlined_call_time): Match natural implementaiton ...
4567 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
4568 forgotten division by CGRAPH_FREQ_BASE.
4570 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4572 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
4573 Solaris 11. Update comment.
4574 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
4576 * config/sol2.h (STARTFILE_SPEC): Likewise.
4577 * configure: Regenerate.
4579 2017-11-14 Carl Love <cel@us.ibm.com>
4581 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
4582 le_ and be_ prefixes to swap* variables. Remove
4583 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
4586 2017-11-14 Jason Merrill <jason@redhat.com>
4588 Support GTY((cache)) on hash_map.
4589 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
4590 (ggc_cache_remove): Override it instead of ggc_mx.
4591 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
4592 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
4593 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
4594 (simple_cache_map_traits): Override maybe_mx.
4595 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
4596 (hash_map): Friend gt_cleare_cache.
4597 (gt_cleare_cache): New.
4598 * tree.h (tree_cache_traits): New hash_map traits class.
4599 (tree_cache_map): New typedef.
4601 2017-11-14 Richard Biener <rguenther@suse.de>
4603 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
4604 paramter and handling.
4605 (cleanup_control_flow_bb): Likewise.
4606 (cleanup_control_flow_pre): New helper performing a DFS walk
4607 to call cleanup_control_flow_bb in PRE order.
4608 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
4609 via cleanup_control_flow_pre.
4611 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
4613 * config/aarch64/aarch64-simd.md
4614 (aarch64_simd_bsl<mode>_internal): Remove DImode.
4615 (*aarch64_simd_bsl<mode>_alt): Likewise.
4616 (aarch64_simd_bsldi_internal): New.
4617 (aarch64_simd_bsldi_alt): Likewise.
4619 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4621 * tracer.c (better_p): Do not compare frequencies.
4622 * reg-stack.c (better_edge): Likewise.
4623 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
4626 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4628 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
4629 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
4630 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
4631 * lto-streamer-in.c (input_function): Use update_max_bb_count.
4632 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
4633 * predict.c (maybe_hot_frequency_p): Inline to ...
4634 (maybe_hot_count_p): ... here; rewrite to counts.
4635 (counts_to_freqs): Rename to ...
4636 (update_max_bb_count): ... this one.
4637 (expensive_function_p): Use counts.
4638 (estimate_bb_frequencies): Update.
4639 (rebuild_frequencies): Update.
4640 * predict.h (counts_to_freqs): Rename to ...
4641 (update_max_bb_count): ... this one.
4642 * profile.c (compute_branch_probabilities): Add debug info
4643 * tree-inline.c (expand_call_inline): Update debug info.
4644 (optimize_inline_calls): Use update_max_bb_count..
4645 (tree_function_versioning): Use update_max_bb_count..
4646 * value-prof.c (gimple_value_profile_transformations):
4647 Do not use update_max_bb_count.
4649 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4651 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
4652 always use frequencies.
4654 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4656 * bb-reorder.c: Remove frequencies from comments.
4657 (better_edge_p): Use profile counts.
4658 (find_traces): Dump profile counts.
4659 (rotate_loop): Use profile counts.
4660 (find_traces_1_round): Likewise.
4661 (connect_better_edge_p): Use counts instead of probabilities for
4663 (copy_bb_p): Drop early check for non-0 frequency.
4664 (sanitize_hot_paths): Update comments.
4666 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4668 * ipa-split.c (struct split_point): Add count.
4669 (consider_split): Do not compute incoming frequency; compute incoming
4670 count and store it to split_point.
4671 (split_function): Set count of the call to split part correctly.
4673 2017-11-13 Carl Love <cel@us.ibm.com>
4675 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
4677 2017-11-13 Tom Tromey <tom@tromey.com>
4679 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
4681 2017-11-13 Carl Love <cel@us.ibm.com>
4683 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
4684 Add support for builtins:
4685 unsigned int vec_first_{,miss}_match_{,or_eos}index,
4686 vector {un,}signed {char,int,short},
4687 vector {un,}signed {char,int,short}) arguments.
4688 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
4689 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
4690 Add BU_P9V_AV_2 expansions for the builtins.
4691 * config/rs6000/altivec.h (vec_first_match_index,
4692 vec_first_mismatch_index, vec_first_match_or_eos_index,
4693 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
4694 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
4695 new extern declaration.
4696 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
4697 * config/rs6000/vsx.md (first_match_index_<mode>,
4698 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
4699 first_mismatch_or_eos_index_<mode>): Add define expand.
4700 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
4701 * doc/extend.texi: Update the built-in documenation file for the new
4704 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
4706 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
4707 into the min/max operations for _Float<N> and _Float<N>X types.
4709 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4712 * dwarf2out.c (do_eh_frame): New static variable.
4713 (dwarf2out_begin_prologue): Set it.
4714 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
4716 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4718 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
4720 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
4722 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4725 * debug.h (dwarf2out_do_eh_frame): Declare.
4726 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
4727 (dwarf2out_do_frame): Use it.
4728 (dwarf2out_do_cfi_asm): Likewise.
4729 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
4730 (dwarf2out_assembly_start): Likewise.
4731 (dwarf2out_begin_prologue): Fix comment.
4732 * toplev.c (compile_file): Always call dwarf2out_frame_finish
4733 if the target needs either debug or unwind DWARF2 info.
4734 * lto-opts.c (lto_write_options): Do not save -fexceptions,
4735 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
4736 -ftrapping-math, -ftrapv and -fwrapv.
4738 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4740 * cgraph.c (cgraph_edge::sreal_frequency): New function.
4741 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
4742 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
4743 (estimate_edge_size_and_time): Likewise.
4744 (ipa_merge_fn_summary_after_inlining): Likewise.
4745 * ipa-inline.c (cgraph_freq_base_rec): Remove.
4746 (compute_uninlined_call_time): Use sreal_frequency.
4747 (compute_inlined_call_time): Likewise.
4748 (ipa_inline): Do not initialize cgraph_freq_base_rec.
4749 * profile-count.c: Include sreal.h.
4750 (profile_count::to_sreal_scale): New.
4751 * profile-count.h: Forward declare sreal.
4752 (profile_count::to_sreal_scale): Declare.
4754 2017-11-13 Nathan Sidwell <nathan@acm.org>
4756 * diagnostic.c (maybe_line_and_column): New.
4757 (diagnostic_get_location_text): Use it.
4758 (diagnostic_report_current_module): Likewise.
4759 (test_diagnostic_get_location_text): Add tests.
4761 2017-11-13 Luis Machado <luis.machado@linaro.org>
4763 * doc/md.texi (Specifying processor pipeline description): Fix
4764 incorrect latency for the div instruction example.
4766 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4768 PR tree-optimization/78821
4769 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
4770 that bit_not_p is the same.
4771 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
4772 (split_group): Count precisely bit_not_p bits in each statement.
4773 (invert_op): New function.
4774 (imm_store_chain_info::output_merged_store): Use invert_op to
4775 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
4776 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
4778 2017-11-13 Martin Liska <mliska@suse.cz>
4780 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
4781 (struct source_info): Likewise.
4782 (add_branch_counts): Likewise.
4783 (add_line_counts): Likewise.
4784 (function_summary): Likewise.
4785 (output_intermediate_line): Likewise.
4786 (generate_results): Likewise.
4788 2017-11-13 Martin Liska <mliska@suse.cz>
4790 * gcov.c (struct block_info): Remove typedef for block_t.
4791 (struct line_info): Likewise.
4792 (line_info::has_block): Likewise.
4793 (EXIT_BLOCK): Likewise.
4794 (unblock): Likewise.
4795 (circuit): Likewise.
4796 (get_cycles_count): Likewise.
4797 (process_file): Likewise.
4798 (read_graph_file): Likewise.
4799 (solve_flow_graph): Likewise.
4800 (find_exception_blocks): Likewise.
4801 (add_line_counts): Likewise.
4802 (accumulate_line_info): Likewise.
4803 (output_line_details): Likewise.
4805 2017-11-13 Martin Liska <mliska@suse.cz>
4807 * gcov.c (struct arc_info): Remove typedef for arc_t.
4808 (struct line_info): Likewise.
4809 (add_branch_counts): Likewise.
4810 (output_branch_count): Likewise.
4811 (function_info::~function_info): Likewise.
4812 (circuit): Likewise.
4813 (output_intermediate_line): Likewise.
4814 (read_graph_file): Likewise.
4815 (solve_flow_graph): Likewise.
4816 (find_exception_blocks): Likewise.
4817 (add_line_counts): Likewise.
4818 (accumulate_line_info): Likewise.
4819 (output_line_details): Likewise.
4820 (output_function_details): Likewise.
4822 2017-11-13 Martin Liska <mliska@suse.cz>
4824 * gcov.c (struct function_info): Remove typedef for function_t.
4825 (struct source_info): Likewise.
4826 (source_info::get_functions_at_location): Likewise.
4827 (solve_flow_graph): Likewise.
4828 (find_exception_blocks): Likewise.
4829 (add_line_counts): Likewise.
4830 (output_intermediate_file): Likewise.
4831 (process_file): Likewise.
4832 (generate_results): Likewise.
4833 (release_structures): Likewise.
4834 (read_graph_file): Likewise.
4835 (read_count_file): Likewise.
4836 (accumulate_line_counts): Likewise.
4837 (output_lines): Likewise.
4839 2017-11-13 Martin Liska <mliska@suse.cz>
4841 * gcov.c (function_info::function_info): Remove num_counts
4842 and add vector<gcov_type>.
4843 (function_info::~function_info): Use the vector.
4844 (process_file): Likewise.
4845 (read_graph_file): Likewise.
4846 (read_count_file): Likewise.
4847 (solve_flow_graph): Likewise.
4849 2017-11-13 Martin Liska <mliska@suse.cz>
4851 * gcov.c (function_info::is_artificial): New function.
4852 (process_file): Erase all artificial early.
4853 (generate_results): Skip as all artificial are already
4856 2017-11-13 Martin Liska <mliska@suse.cz>
4858 * gcov.c (read_graph_file): Store to global vector of functions.
4859 (read_count_file): Iterate the vector.
4860 (process_file): Likewise.
4861 (generate_results): Likewise.
4862 (release_structures): Likewise.
4864 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4866 PR tree-optimization/82954
4867 * gimple-ssa-store-merging.c
4868 (imm_store_chain_info::coalesce_immediate_stores): If
4869 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
4871 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
4873 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
4874 Upddate call to ENDIAN_LANE_N.
4875 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
4876 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
4877 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
4878 and use aarch64_endian_lane_rtx.
4879 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
4881 2017-11-12 Tom de Vries <tom@codesourcery.com>
4883 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
4885 2017-11-12 Tom de Vries <tom@codesourcery.com>
4887 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
4889 2017-11-12 Tom de Vries <tom@codesourcery.com>
4891 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
4893 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4894 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4895 * defaults.h (ASM_OUTPUT_LABELREF): Same.
4897 2017-11-11 Martin Sebor <msebor@redhat.com>
4900 * doc/extend.texi (attribute nonstring): Remove spurious argument.
4903 * prefic.c (translate_name): Replace strncpy with memcpy to
4904 avoid -Wstringop-truncation.
4906 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4908 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
4910 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4912 * predict.c (maybe_hot_frequency_p): Do not use cfun.
4914 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4916 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
4919 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
4921 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
4922 enable generating XXBRH if the value is in a vector register.
4923 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
4924 value is in a vector register.
4925 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
4926 register to register bswap64's instead of doing the GPR sequence
4927 used on previous machines.
4928 (bswapdi2_xxbrd): New insn.
4929 (bswapdi2_reg): Disallow on ISA 3.0.
4930 (register to register bswap64 splitter): Do not split the insn on
4931 ISA 3.0 systems that use XXBRD.
4933 2017-11-10 Martin Sebor <msebor@redhat.com>
4936 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
4938 (find_subframework_file): Same.
4940 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4942 * auto-profile.c (afdo_indirect_call): Drop frequency.
4943 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
4944 (cgraph_node::create_edge): Drop frequency argument.
4945 (cgraph_node::create_indirect_edge): Drop frequency argument.
4946 (cgraph_edge::make_speculative): Drop frequency arguments.
4947 (cgraph_edge::resolve_speculation): Do not update frequencies
4948 (cgraph_edge::dump_edge_flags): Do not dump frequency.
4949 (cgraph_node::dump): Check consistency in IPA mode.
4950 (cgraph_edge::maybe_hot_p): Use IPA counter.
4951 (cgraph_edge::verify_count_and_frequency): Rename to ...
4952 (cgraph_edge::verify_count): ... this one; drop frequency checking.
4953 (cgraph_node::verify_node): Update.
4954 * cgraph.h (struct cgraph_edge): Drop frequency.
4955 (cgraph_edge::frequency): New function.
4956 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
4958 (cgraph_edge::rebuild_edges): Likewise.
4959 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
4960 (duplicate_thunk_for_node): Do not pass frequency.
4961 (cgraph_node::create_clone): Scale only counts.
4962 (cgraph_node::create_virtual_clone): Do not pass frequency.
4963 (cgraph_node::create_edge_including_clones): Do not pass frequency.
4964 (cgraph_node::create_version_clone): Do not pass frequency.
4965 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
4966 (cgraph_node::expand_thunk): Do not pass frequency.
4967 (cgraph_node::create_wrapper): Do not pass frequency.
4968 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
4970 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
4971 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
4972 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
4973 (gather_caller_stats): Use frequency function.
4974 (ipcp_cloning_candidate_p): Use frequency function.
4975 (ipcp_propagate_stage): Use frequency function.
4976 (get_info_about_necessary_edges): Use frequency function.
4977 (update_profiling_info): Update only IPA profile.
4978 (update_specialized_profile): Use frequency functoin.
4979 (perhaps_add_new_callers): Update only IPA profile.
4980 * ipa-devirt.c (ipa_devirt): Use IPA profile.
4981 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
4982 (dump_ipa_call_summary): Use frequency function.
4983 (estimate_edge_size_and_time): Use frequency function.
4984 (ipa_merge_fn_summary_after_inlining): Use frequency function.
4985 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
4986 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
4987 (update_noncloned_counts): ... ths one; scale counts only.
4988 (clone_inlined_nodes): Do not scale frequency.
4989 (inline_call): Do not pass frequency.
4990 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
4991 (compute_inlined_call_time): Use IPA profile.
4992 (want_inline_small_function_p): Use IPA profile.
4993 (want_inline_self_recursive_call_p): Use IPA profile.
4994 (edge_badness): Use IPA profile.
4995 (lookup_recursive_calls): Use IPA profile.
4996 (recursive_inlining): Do not pass frequency.
4997 (resolve_noninline_speculation): Do not update frequency.
4998 (inline_small_functions): Collect max of IPA profile.
4999 (dump_overall_stats): Dump IPA porfile.
5000 (dump_inline_stats): Dump IPA porfile.
5001 (ipa_inline): Collect IPA stats.
5002 * ipa-inline.h (clone_inlined_nodes): Update prototype.
5003 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
5004 (ipa_propagate_frequency): Use frequency function.
5005 (ipa_profile): Cleanup.
5006 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
5007 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
5008 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
5009 (input_node): Do not stream frequency.
5010 (input_edge): Do not stream frequency.
5011 (merge_profile_summaries): Scale only IPA profiles.
5012 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
5013 * predict.c (drop_profile): Do not recompute frequency.
5014 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
5015 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
5016 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
5017 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
5018 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
5019 * tree-inline.c (copy_bb): Do not scale frequency.
5020 (expand_call_inline): Do not scale frequency.
5021 (tree_function_versioning): Do not scale frequency.
5022 * ubsan.c (ubsan_create_edge): Do not pass frequency.
5024 2017-11-10 Julia Koval <julia.koval@intel.com>
5026 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
5027 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
5028 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
5029 (_mm256_maskz_gf2p8affine_epi64_epi8)
5030 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
5031 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
5032 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
5033 (__builtin_ia32_vgf2p8affineqb_v32qi)
5034 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
5035 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
5037 2017-11-10 Tamar Christina <tamar.christina@arm.com>
5041 (arm_option_override): Refactor.
5042 (arm_option_reconfigure_globals): New.
5043 (arm_options_perform_arch_sanity_checks): New.
5044 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
5046 (arm_options_perform_arch_sanity_checks): Likewise
5048 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
5050 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
5051 (power9-qpmul): New.
5052 * rs6000/rs6000.md ("type" attr): Add qmul.
5053 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
5054 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
5055 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
5057 2017-11-10 Martin Sebor <msebor@redhat.com>
5060 * builtins.c (compute_objsize): Handle arrays that
5061 compute_builtin_object_size likes to fail for. Make extern.
5062 * builtins.h (compute_objsize): Declare.
5063 (check_strncpy_sizes): New function.
5064 (expand_builtin_strncpy): Call check_strncpy_sizes.
5065 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
5066 -Wstringop-truncation.
5067 (gimple_fold_builtin_strncat): Same.
5068 * gimple.c (gimple_build_call_from_tree): Set call location.
5069 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
5070 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
5071 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
5072 (handle_builtin_strlen): Use strlen_to_stridx.
5073 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
5075 Use strlen_to_stridx.
5076 (pass_strlen::execute): Release strlen_to_stridx.
5077 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
5078 (-Wstringop-truncation): Document new option.
5080 2017-11-10 Martin Liska <mliska@suse.cz>
5082 PR gcov-profile/82702
5083 * gcov.c (main): Handle intermediate files in a different
5085 (get_gcov_intermediate_filename): New function.
5086 (output_gcov_file): Remove support of intermediate files.
5087 (generate_results): Allocate intermediate file.
5088 (release_structures): Clean-up properly fn_end.
5089 (output_intermediate_file): Start iterating with line 1.
5091 2017-11-10 Jakub Jelinek <jakub@redhat.com>
5093 PR tree-optimization/82929
5094 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
5095 ops_swapped_p non-static data member.
5096 (store_immediate_info::store_immediate_info): Clear it.
5097 (imm_store_chain_info::coalesce_immediate_stores): If swapping
5098 ops set ops_swapped_p.
5099 (count_multiple_uses): Handle ops_swapped_p.
5101 2017-11-10 Martin Liska <mliska@suse.cz>
5103 * coverage.c (coverage_init): Stream information about
5104 support of has_unexecuted_blocks.
5105 * doc/gcov.texi: Document that.
5106 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
5107 * gcov.c (read_graph_file): Likewise.
5108 (output_line_beginning): Fix a small issue with
5111 2017-11-10 Bin Cheng <bin.cheng@arm.com>
5113 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
5114 reference of trivial component.
5116 2017-11-10 Jakub Jelinek <jakub@redhat.com>
5119 * gimple-ssa-store-merging.c
5120 (pass_store_merging::terminate_all_aliasing_chains): For
5121 gimple_store_p stmts also call refs_output_dependent_p.
5123 PR rtl-optimization/82913
5124 * compare-elim.c (try_merge_compare): Punt if def_insn is not
5127 2017-11-09 Jeff Law <law@redhat.com>
5129 * vr-values.h: New file with vr_values class.
5130 * tree-vrp.c: Include vr-values.h
5131 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
5132 data objects into the vr_values class.
5133 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
5134 (get_value_range): Make it a member function within vr_values class.
5135 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
5136 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
5137 (op_with_constant_singleton_value_range): Likewise.
5138 (extract_range_for_var_from_comparison_expr): Likewise.
5139 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
5140 (extract_range_from_binary_expr): Likewise.
5141 (extract_range_from_unary_expr): Likewise.
5142 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
5143 (check_for_binary_op_overflow, extract_range_basic): Likewise.
5144 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
5145 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
5146 (compare_name_with_value, compare_names): Likewise.
5147 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
5148 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
5149 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
5150 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
5151 (extract_range_from_phi_node): Likewise.
5152 (simplify_truth_ops_using_ranges): Likewise.
5153 (simplify_div_or_mod_using_ranges): Likewise.
5154 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
5155 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
5156 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
5157 (simplify_float_conversion_using_ranges): Likewise.
5158 (simplify_internal_call_using_ranges): Likewise.
5159 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
5160 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
5161 poitner via x_vr_values for calls into gimple folder.
5162 (vrp_initialize_lattice): Make this the vr_values ctor.
5163 (vrp_free_lattice): Make this the vr_values dtor.
5164 (set_vr_value): New function.
5165 (class vrp_prop): Add vr_values data member. Add various member
5166 functions as well as member functions that delegate to vr_values.
5167 (check_array_ref): Make a member function within vrp_prop class.
5168 (search_for_addr_array, vrp_initialize): Likewise.
5169 (vrp_finalize): Likewise. Revamp to avoid direct access to
5170 vr_value, values_propagated, etc.
5171 (check_array_bounds): Extract vrp_prop class instance pointer from
5172 walk info structure. Use it to call member functions.
5173 (check_all_array_refs): Make a member function within vrp_prop class.
5174 Smuggle class instance pointer via walk info structure.
5175 (x_vr_values): New local static.
5176 (vrp_valueize): Use x_vr_values to get class instance.
5177 (vr_valueize_1): Likewise.
5178 (class vrp_folder): Add vr_values data member. Add various member
5179 functions as well as member functions that delegate to vr_values.
5180 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
5181 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
5182 class instance from vr_values. Use it to call member functions.
5183 (vrp_dom_walker): Add vr_values data member.
5184 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
5185 instance via x_vr_values.
5186 (identify_jump_threads): Accept vr_values as argument. Store
5187 it into the walker structure.
5188 (evrp_dom_walker): Add vr_values class data member. Add various
5190 (evrp_dom_walker::try_find_new_range): Use vr_values data
5191 member to access the memory allocator.
5192 (evrp_dom_walker::before_dom_children): Store vr_values class
5193 instance into the vrp_folder class.
5194 (evrp_dom_walker::push_value_range): Rework to avoid direct
5195 access to num_vr_values and vr_value.
5196 (evrp_dom_walker::pop_value_range): Likewise.
5197 (execute_early_vrp): Remove call to vrp_initialize_lattice.
5198 Use vr_values to get to dump_all_value_ranges member function.
5199 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
5200 and simplify_cond_using_ranges_2 via vrp_prop class instance.
5201 Pass vr_values class instance down to identify_jump_threads.
5202 Remove call to vrp_free_lattice.
5203 (debug_all_value_ranges): Remove.
5205 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
5206 (vrp_folder): Likewise.
5208 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
5209 Get it from the existing bitmap instead.
5210 (vrp_intersect_ranges_1): Likewise.
5212 2017-11-09 Jakub Jelinek <jakub@redhat.com>
5214 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
5216 (store_immediate_info::store_immediate_info): Add bitnotp argument,
5217 set bit_not_p to it.
5218 (imm_store_chain_info::coalesce_immediate_stores): Break group
5219 if bit_not_p is different.
5220 (count_multiple_uses, split_group,
5221 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
5222 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
5223 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
5224 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
5225 to store_immediate_info ctor.
5227 2017-11-09 Jim Wilson <jimw@sifive.com>
5229 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
5230 (scan_prog_file): Likewise.
5232 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
5234 * bb-reorder.c (max_entry_frequency): Remove.
5235 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
5236 connect_traces, push_to_next_round_p): Remove prototypes.
5237 (find_traces_1_round): Use counts only.
5238 (push_to_next_round_p): Likewise.
5239 (find_traces): Likewise.
5240 (rotate_loop): Likewise.
5241 (find_traces_1_round): Likewise.
5242 (connect_traces): Likewise.
5243 (edge_order): Likewise.
5245 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
5247 * config/arm/arm.c (output_return_instruction): Add comments to
5248 indicate requirement for cmse_nonsecure_entry return to account
5249 for the size of clearing instruction output here.
5250 (thumb_exit): Likewise.
5251 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
5252 return in hardfloat mode.
5254 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
5256 * config/rs6000/rs6000.c (machine_function): Add a bool,
5257 "toc_is_wrapped_separately".
5258 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
5259 if it wasn't explicitly set or unset, we are optimizing for speed, and
5260 doing separate shrink-wrapping.
5261 (rs6000_get_separate_components): Enable the TOC component if
5262 saving the TOC register in the prologue.
5263 (rs6000_components_for_bb): Handle the TOC component.
5264 (rs6000_emit_prologue_components): Store the TOC register where needed.
5265 (rs6000_set_handled_components): Mark TOC as handled, if handled.
5266 (rs6000_emit_prologue): Don't save the TOC if that is already done.
5268 2017-11-09 Martin Jambor <mjambor@suse.cz>
5270 * ipa-param-manipulation.c: New file.
5271 * ipa-param-manipulation.h: Likewise.
5272 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
5273 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
5274 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
5275 (ipa_parm_adjustment): Likewise.
5276 (ipa_parm_adjustment_vec): Likewise.
5277 (ipa_get_vector_of_formal_parms): Moved declaration to
5278 ipa-param-manipulation.h.
5279 (ipa_get_vector_of_formal_parm_types): Likewise.
5280 (ipa_modify_formal_parameters): Likewise.
5281 (ipa_modify_call_arguments): Likewise.
5282 (ipa_combine_adjustments): Likewise.
5283 (ipa_dump_param_adjustments): Likewise.
5284 (ipa_modify_expr): Likewise.
5285 (ipa_get_adjustment_candidate): Likewise.
5286 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
5287 ipa-param-manipulation.c.
5288 (ipa_get_vector_of_formal_parm_types): Likewise.
5289 (ipa_modify_formal_parameters): Likewise.
5290 (ipa_modify_call_arguments): Likewise.
5291 (ipa_modify_expr): Likewise.
5292 (get_ssa_base_param): Likewise.
5293 (ipa_get_adjustment_candidate): Likewise.
5294 (index_in_adjustments_multiple_times_p): Likewise.
5295 (ipa_combine_adjustments): Likewise.
5296 (ipa_dump_param_adjustments): Likewise.
5297 * tree-sra.c: Also include ipa-param-manipulation.h
5298 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
5301 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5302 Alan Hayward <alan.hayward@arm.com>
5303 David Sherwood <david.sherwood@arm.com>
5305 * doc/sourcebuild.texi (vect_masked_store): Document.
5307 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5308 Alan Hayward <alan.hayward@arm.com>
5309 David Sherwood <david.sherwood@arm.com>
5311 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
5313 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5314 Alan Hayward <alan.hayward@arm.com>
5315 David Sherwood <david.sherwood@arm.com>
5317 * doc/sourcebuild.texi (vect_variable_length): Document.
5319 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5320 Alan Hayward <alan.hayward@arm.com>
5321 David Sherwood <david.sherwood@arm.com>
5323 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
5325 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5326 Alan Hayward <alan.hayward@arm.com>
5327 David Sherwood <david.sherwood@arm.com>
5329 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
5331 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5332 Alan Hayward <alan.hayward@arm.com>
5333 David Sherwood <david.sherwood@arm.com>
5335 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
5336 previously undocumented selectors.
5337 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
5339 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5341 * doc/rtl.texi (const_vector): Say that elements can be
5342 const_wide_ints too.
5343 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
5344 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
5345 (gen_vec_duplicate): Use it instead of CONSTANT_P.
5346 * optabs.c (expand_vector_broadcast): Likewise.
5348 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5349 Alan Hayward <alan.hayward@arm.com>
5350 David Sherwood <david.sherwood@arm.com>
5352 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
5353 scaled index even if the unscaled address was invalid.
5354 Don't increase the complexity of using a scale in that case.
5356 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5357 Alan Hayward <alan.hayward@arm.com>
5358 David Sherwood <david.sherwood@arm.com>
5360 * doc/rtl.texi: Rewrite the subreg rules so that they partition
5361 the inner register into REGMODE_NATURAL_SIZE bytes rather than
5362 UNITS_PER_WORD bytes.
5363 * emit-rtl.c (validate_subreg): Divide subregs into blocks
5364 based on REGMODE_NATURAL_SIZE of the inner mode.
5365 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
5366 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
5367 * expmed.c (lowpart_bit_field_p): Divide the value up into
5368 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
5369 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
5370 whether something is likely to occupy more than one register.
5372 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
5375 * ipa-inline-transform.c (update_noncloned_frequencies): Use
5376 profile_count::adjust_for_ipa_scaling.
5377 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
5378 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
5380 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
5382 2017-11-09 Jakub Jelinek <jakub@redhat.com>
5384 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
5385 (split_group): Add total_orig and total_new arguments, estimate the
5386 number of statements related to the store group without store merging
5387 and with store merging.
5388 (imm_store_chain_info::output_merged_store): Adjust split_group
5389 callers, punt if estimated number of statements with store merging
5390 is not smaller than estimated number of statements without it.
5392 (handled_load): Remove has_single_use checks.
5393 (pass_store_merging::process_store): Likewise.
5395 2017-11-09 Richard Biener <rguenther@suse.de>
5397 PR tree-optimization/82902
5398 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
5400 2017-11-09 Martin Liska <mliska@suse.cz>
5403 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
5406 2017-11-09 Martin Liska <mliska@suse.cz>
5408 PR tree-optimization/82669
5409 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
5411 2017-11-09 Martin Liska <mliska@suse.cz>
5413 PR gcov-profile/48463
5414 * coverage.c (coverage_begin_function): Output also end locus
5415 of a function and information whether the function is
5417 * gcov-dump.c (tag_function): Parse and print the information.
5418 * gcov.c (INCLUDE_MAP): Add include.
5419 (INCLUDE_SET): Likewise.
5420 (struct line_info): Move earlier in the source file because
5421 of vector<line_info> in function_info structure.
5422 (line_info::line_info): Likewise.
5423 (line_info::has_block): Likewise.
5424 (struct source_info): Add new member index.
5425 (source_info::get_functions_at_location): New function.
5426 (function_info::group_line_p): New function.
5427 (output_intermediate_line): New function.
5428 (output_intermediate_file): Use the mentioned function.
5429 (struct function_start): New.
5430 (struct function_start_pair_hash): Likewise.
5431 (process_file): Add code that identifies group functions.
5432 Assign lines either to global or function scope.
5433 (generate_results): Skip artificial functions.
5434 (find_source): Assign index for each source file.
5435 (read_graph_file): Read new flag artificial and end_line.
5436 (add_line_counts): Assign it either to global of function scope.
5437 (accumulate_line_counts): Isolate core of the function to
5438 accumulate_line_info and call it for both function and global
5440 (accumulate_line_info): New function.
5441 (output_line_beginning): Fix GNU coding style.
5442 (print_source_line): New function.
5443 (output_line_details): Likewise.
5444 (output_function_details): Likewise.
5445 (output_lines): Iterate both source (global) scope and function
5447 (struct function_line_start_cmp): New class.
5448 * doc/gcov.texi: Reflect changes in documentation.
5450 2017-11-09 Jakub Jelinek <jakub@redhat.com>
5453 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
5454 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
5455 and similarly for not instead of neg.
5457 2017-11-08 Andi Kleen <ak@linux.intel.com>
5459 * config/i386/i386.opt: Add -mforce-indirect-call.
5460 * config/i386/predicates.md: Check for flag_force_indirect_call.
5461 * doc/invoke.texi: Document -mforce-indirect-call
5463 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
5465 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
5467 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
5468 config/riscv/riscv.c (predict.h): New include.
5469 (riscv_slow_unaligned_access_p): No longer static.
5470 (riscv_block_move_straight): Add require.
5471 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
5473 2017-11-08 Jakub Jelinek <jakub@redhat.com>
5476 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
5477 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
5478 nonimmediate_operand predicate for operand 1 instead of
5481 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5483 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
5485 * config/aarch64/constraints.md (Uml): New constraint.
5486 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
5489 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5491 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
5492 of two vec_duplicates into a vec_concat.
5494 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5496 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
5497 Simplify vec_merge of vec_duplicate and vec_concat.
5498 * config/aarch64/constraints.md (Utq): New constraint.
5499 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
5502 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5504 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
5505 Simplify vec_merge of vec_duplicate and const_vector.
5506 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
5508 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
5509 mode iterator. Update predicate on operand 1 to
5510 handle non-const_vec constants. Delete constraints.
5511 (*aarch64_combinez_be<mode>): Likewise for operand 2.
5513 2017-11-08 Jakub Jelinek <jakub@redhat.com>
5515 PR tree-optimization/78821
5516 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
5518 (store_operand_info::store_operand_info): Initialize it to false.
5519 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
5520 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
5521 store in the group, and if chain_info is non-NULL, to ignore altogether
5523 (compatible_load_p): Fail if bit_not_p does not match.
5524 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
5525 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
5526 (pass_store_merging::process_store): Adjust
5527 terminate_all_aliasing_chains calls to pass NULL in all current spots,
5528 call terminate_all_aliasing_chains newly when adding a store into
5529 a chain with non-NULL chain_info.
5531 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
5533 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
5535 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
5537 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5539 (aarch64_layout_frame): Initialise emit_frame_chain.
5540 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
5541 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
5543 2017-11-08 Martin Liska <mliska@suse.cz>
5545 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
5546 of gimple_call_internal_p.
5548 2017-11-07 Tom de Vries <tom@codesourcery.com>
5550 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
5552 2017-11-07 Tom de Vries <tom@codesourcery.com>
5554 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5557 2017-11-08 Martin Liska <mliska@suse.cz>
5560 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
5562 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
5564 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
5566 2017-11-07 Jakub Jelinek <jakub@redhat.com>
5569 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
5570 (*cmp<mode>_ccz_1): New insn with $k alternative.
5573 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
5574 RTX_COMM_COMPARE as commutative as well.
5575 (ix86_binary_operator_ok): Formatting fix.
5576 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
5577 *<code><mode>3<mask_name><round_saeonly_name>,
5578 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
5579 *<s>mul<mode>3_highpart<mask_name>,
5580 *vec_widen_umult_even_v16si<mask_name>,
5581 *vec_widen_umult_even_v8si<mask_name>,
5582 *vec_widen_umult_even_v4si<mask_name>,
5583 *vec_widen_smult_even_v16si<mask_name>,
5584 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
5585 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
5586 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
5587 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
5588 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
5589 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
5590 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
5591 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
5592 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
5593 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
5594 ix86_binary_operator_ok. Formatting fixes.
5595 (*<plusminus_insn><mode>3<mask_name><round_name>,
5596 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
5599 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
5601 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
5602 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
5603 eq and ne if TARGET_ISEL.
5604 (cmp): New code_iterator.
5605 (UNS, UNSU_, UNSIK): New code_attrs.
5606 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
5607 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
5609 ("eq<mode>3"): ... this.
5610 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
5612 ("ne<mode>3"): ... this.
5614 2017-11-07 Julia Koval <julia.koval@intel.com>
5617 * common/config/i386/i386-common.c
5618 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
5619 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
5620 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
5621 * config/i386/i386.opt: Ditto.
5622 * config/i386/i386.c (ix86_target_string): Ditto.
5623 (ix86_option_override_internal): Ditto.
5624 (ix86_init_mpx_builtins): Move MPX to args2.
5625 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
5626 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
5627 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
5628 __builtin_ia32_vgf2p8affineinvqb_v32qi,
5629 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
5630 __builtin_ia32_vgf2p8affineinvqb_v16qi,
5631 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
5633 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
5636 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
5637 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
5638 (zero-extendsidi peephole2): Remove peephole.
5640 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
5643 * stor-layout.c: Include attribs.h.
5644 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
5645 explicit lookup of "aligned" attribute.
5647 2017-11-07 Andrew Waterman <andrew@sifive.com>
5649 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
5650 (riscv_expand_block_move): Likewise.
5651 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
5653 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
5654 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
5655 * config/riscv/riscv.c (riscv_block_move_straight): New function.
5656 (riscv_adjust_block_mem): Likewise.
5657 (riscv_block_move_loop): Likewise.
5658 (riscv_expand_block_move): Likewise.
5659 * config/riscv/riscv.md (movmemsi): New pattern.
5661 2017-11-07 Michael Clark <michaeljclark@mac.com>
5663 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
5664 (MUSL_DYNAMIC_LINKER): Likewise.
5666 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
5668 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
5669 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
5672 2017-11-07 Richard Biener <rguenther@suse.de>
5674 * match.pd: Fix build.
5676 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
5677 Jackson Woodruff <jackson.woodruff@arm.com>
5679 PR tree-optimization/71026
5680 * match.pd: Canonicalize negate in division.
5682 2017-11-07 Sudakshina Das <sudi.das@arm.com>
5685 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
5687 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5689 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
5690 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
5692 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5694 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
5695 non-scalar integral types.
5696 * match.pd (negate_expr_p): Handle MINUS_EXPR.
5697 (-(A-B), -(~A)): New transformations.
5699 2017-11-07 Tom de Vries <tom@codesourcery.com>
5701 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
5702 semicolon after "do {} while (0)".
5703 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5704 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5705 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5706 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5707 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5709 2017-11-07 Tom de Vries <tom@codesourcery.com>
5711 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
5712 after "do {} while (0)".
5713 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5714 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5715 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5716 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5717 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5719 2017-11-07 Tom de Vries <tom@codesourcery.com>
5722 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
5724 (arm_rtx_costs_internal): Add missing semicolon after
5725 HANDLE_NARROW_SHIFT_ARITH call.
5727 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5729 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
5730 disable isel if it was not set explicitly.
5732 2017-11-06 James Bowman <james.bowman@ftdichip.com>
5734 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
5735 (add_type_attribute) likewise.
5737 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
5739 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
5740 of struct ix86_frame.
5741 (ix86_initial_elimination_offset): Likewise.
5742 (ix86_expand_split_stack_prologue): Likewise.
5744 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
5746 * tree-vrp.h (enum value_range_type): Update stale comment.
5748 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5749 Alan Hayward <alan.hayward@arm.com>
5750 David Sherwood <david.sherwood@arm.com>
5752 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
5753 (aarch64_expand_vec_perm_const): Take the number of units too.
5754 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
5755 (aarch64_expand_vec_perm_const): Likewise.
5756 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
5757 (vec_perm<mode>): Update accordingly.
5759 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5760 Alan Hayward <alan.hayward@arm.com>
5761 David Sherwood <david.sherwood@arm.com>
5763 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
5764 Take the number of units too.
5765 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
5766 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
5767 but check for a vector mode before rather than after the call.
5768 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
5769 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
5770 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
5771 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
5772 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
5773 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
5774 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
5775 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
5776 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
5777 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
5778 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
5779 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
5780 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
5781 (aarch64_sqdmull2_n<mode>): Update accordingly.
5783 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5784 Alan Hayward <alan.hayward@arm.com>
5785 David Sherwood <david.sherwood@arm.com>
5787 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
5788 the number of units too.
5789 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
5790 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
5791 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
5792 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
5793 (vec_store_lanesxi<mode>): Update accordingly.
5795 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5796 Alan Hayward <alan.hayward@arm.com>
5797 David Sherwood <david.sherwood@arm.com>
5799 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
5800 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
5801 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
5802 of units rather than the mode.
5803 * config/aarch64/iterators.md (nunits): New mode attribute.
5804 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
5805 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
5806 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
5807 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
5808 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5809 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
5810 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
5811 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
5812 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
5813 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
5814 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5815 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
5816 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
5817 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
5818 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
5819 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
5820 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
5821 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
5822 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
5823 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
5824 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
5825 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
5826 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
5827 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
5828 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
5829 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
5830 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
5831 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
5832 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
5833 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
5834 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
5835 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
5836 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
5837 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
5838 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
5839 (aarch64_simd_vec_setv2di): Likewise.
5841 2017-11-06 Carl Love <cel@us.ibm.com>
5843 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
5845 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
5846 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
5848 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
5850 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
5851 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
5852 Add power 8 macro expansions.
5853 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
5854 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
5855 power 8 instructions. (VSX_XXBR): Add iterator.
5857 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
5859 * config/arm/arm.md (predicable_short_it): Change default to "no",
5860 improve documentation, remove uses that are identical to the default.
5861 (enabled_for_depr_it): Rename to enabled_for_short_it.
5862 * gcc/config/arm/arm-fixed.md (predicable_short_it):
5863 Remove default uses.
5864 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
5865 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
5866 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
5867 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
5869 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
5872 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
5873 float128 helper macros, which are no longer used after deleting
5874 the old 'q' built-in functions, and moving the round to odd
5875 built-in functions to being special built-in functions.
5876 (BU_FLOAT128_2): Likewise.
5877 (BU_FLOAT128_1_HW): Likewise.
5878 (BU_FLOAT128_2_HW): Likewise.
5879 (BU_FLOAT128_3_HW): Likewise.
5880 (FABSQ): Delete old 'q' built-in functions.
5881 (COPYSIGNQ): Likewise.
5882 (SQRTF128_ODD): Move round to odd built-in functions to be
5883 special built-in functions, so that we can handle
5884 -mabi=ieeelongdouble.
5885 (TRUNCF128_ODD): Likewise.
5886 (ADDF128_ODD): Likewise.
5887 (SUBF128_ODD): Likewise.
5888 (MULF128_ODD): Likewise.
5889 (DIVF128_ODD): Likewise.
5890 (FMAF128_ODD): Likewise.
5891 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
5892 built-in names to 'f128'.
5893 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
5894 old 'q' built-in functions, as the machine independent code for
5895 'f128' built-in functions handles this.
5896 (rs6000_expand_builtin): Add expansion for float128 round to odd
5897 functions, keying off on -mabi=ieeelongdouble of whether to use
5898 the KFmode or TFmode variant.
5899 (rs6000_init_builtins): Initialize the _Float128 round to odd
5901 * doc/extend.texi (PowerPC Built-in Functions): Document the old
5902 _Float128 'q' built-in functions are now mapped into the new
5903 'f128' built-in functions.
5905 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
5907 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
5908 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
5910 2017-11-06 David Malcolm <dmalcolm@redhat.com>
5913 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
5914 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
5915 * toplev.c: Include "ipa-fnsummary.h".
5916 (toplev::finalize): Call ipa_fnsummary_c_finalize.
5918 2017-11-06 Jakub Jelinek <jakub@redhat.com>
5920 PR tree-optimization/82838
5921 * gimple-ssa-store-merging.c
5922 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
5923 on a separate gimple_seq which is then appended to seq.
5925 2017-11-06 Jeff Law <law@redhat.com>
5928 * config/i386/i386.c (PROBE_INTERVAL): Remove.
5929 (get_probe_interval): New functions.
5930 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
5931 (ix86_adjust_stack_and_probe): Likewise.
5932 (output_adjust_stack_and_probe): Likewise.
5933 (ix86_emit_probe_stack_range): Likewise.
5934 (ix86_expand_prologue): Likewise.
5936 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5938 PR tree-optimization/82816
5939 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
5940 if the modes of the two types are the same.
5941 (convert_plusminus_to_widen): Likewise.
5943 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5945 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
5947 (usadv16qi): New define_expand.
5948 (usadv8hi): New define_expand.
5950 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
5953 * ipa-inline-transform.c (update_noncloned_frequencies): Always
5955 (inline_transform): Likewise.
5956 * predict.c (counts_to_freqs): Remove useless conditional.
5957 * profile-count.h (profile_count::apply_scale): Move sanity check.
5958 * tree-inline.c (copy_bb): Always scale.
5959 (copy_cfg_body): Likewise.
5961 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
5964 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
5966 (*cmp_ite0): Add enabled_for_depr_it attribute.
5967 (*cmp_ite1): Likewise.
5969 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5971 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
5974 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5976 * tree-vrp.c (vrp_int_const_binop): Return true on success and
5977 return the value by pointer.
5978 (extract_range_from_multiplicative_op_1): Update accordingly.
5979 Return as soon as an operation fails.
5981 2017-11-05 Tom de Vries <tom@codesourcery.com>
5984 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
5985 (DEF_SANITIZER_BUILTIN): ... here.
5986 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
5987 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
5989 2017-11-05 Tom de Vries <tom@codesourcery.com>
5992 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
5994 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
5995 ASM_OUTPUT_BEFORE_CASE_LABEL call.
5996 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
5998 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5999 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
6000 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
6002 2017-11-05 Tom de Vries <tom@codesourcery.com>
6005 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
6008 2017-11-04 Michael Clark <michaeljclark@mac.com>
6010 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
6011 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
6013 (*addsi3_extended): Likewise.
6014 (*addsi3_extended2): Likewise.
6015 (<optab>si3): Likewise.
6016 (<optab>di3): Likewise.
6017 (<optab><mode>3): Likewise.
6018 (<*optabe>si3_internal): Likewise.
6019 (zero_extendqi<SUPERQI:mode>2): Likewise.
6020 (*add<mode>hi3): Likewise.
6021 (*xor<mode>hi3): Likewise.
6022 (<optab>di3): Likewise.
6023 (*<optab>si3_extend): Likewise.
6024 (*sge<u>_<X:mode><GPR:mode>): Likewise.
6025 (*slt<u>_<X:mode><GPR:mode>): Likewise.
6026 (*sle<u>_<X:mode><GPR:mode>): Likewise.
6028 2017-11-04 Andrew Waterman <andrew@sifive.com>
6030 * config/riscv/riscv.c (riscv_option_override): Conditionally set
6031 TARGET_STRICT_ALIGN based upon -mtune argument.
6033 2017-11-04 Andrew Waterman <andrew@sifive.com>
6035 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
6037 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
6039 * config/i386/i386.c (choose_basereg): Use optional scratch
6040 register and add assertion.
6041 (x86_emit_outlined_ms2sysv_save): Use scratch register when
6042 needed, and don't allocate stack.
6043 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
6044 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
6045 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
6047 2017-11-03 Jeff Law <law@redhat.com>
6049 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
6050 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
6051 to probe at the start of a noreturn function.
6053 2017-11-03 Jakub Jelinek <jakub@redhat.com>
6055 PR tree-optimization/78821
6056 * gimple-ssa-store-merging.c: Update the file comment.
6057 (MAX_STORE_ALIAS_CHECKS): Define.
6058 (struct store_operand_info): New type.
6059 (store_operand_info::store_operand_info): New constructor.
6060 (struct store_immediate_info): Add rhs_code and ops data members.
6061 (store_immediate_info::store_immediate_info): Add rhscode, op0r
6062 and op1r arguments to the ctor, initialize corresponding data members.
6063 (struct merged_store_group): Add load_align_base and load_align
6065 (merged_store_group::merged_store_group): Initialize them.
6066 (merged_store_group::do_merge): Update them.
6067 (merged_store_group::apply_stores): Pick the constant for
6068 encode_tree_to_bitpos from one of the two operands, or skip
6069 encode_tree_to_bitpos if neither operand is a constant.
6070 (class pass_store_merging): Add process_store method decl. Remove
6071 bool argument from terminate_all_aliasing_chains method decl.
6072 (pass_store_merging::terminate_all_aliasing_chains): Remove
6073 var_offset_p argument and corresponding handling.
6074 (stmts_may_clobber_ref_p): New function.
6075 (compatible_load_p): New function.
6076 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
6077 if there is overlap and rhs_code is not INTEGER_CST. For
6078 non-overlapping stores terminate group if rhs is not mergeable.
6079 (get_alias_type_for_stmts): Change first argument from
6080 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
6081 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
6082 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
6084 (get_location_for_stmts): Change first argument from
6085 auto_vec<gimple *> & to vec<gimple *> &.
6086 (struct split_store): Remove orig_stmts data member, add orig_stores.
6087 (split_store::split_store): Create orig_stores rather than orig_stmts.
6088 (find_constituent_stmts): Renamed to ...
6089 (find_constituent_stores): ... this. Change second argument from
6090 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
6091 to info structures rather than the statements.
6092 (split_group): Rename ALLOW_UNALIGNED argument to
6093 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
6094 it. Adjust find_constituent_stores caller.
6095 (imm_store_chain_info::output_merged_store): Handle rhs_code other
6096 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
6097 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
6098 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
6099 (mem_valid_for_store_merging): New function.
6100 (handled_load): New function.
6101 (pass_store_merging::process_store): New method.
6102 (pass_store_merging::execute): Use process_store method. Adjust
6103 terminate_all_aliasing_chains caller.
6105 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6107 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
6108 Return true for more constants, symbols and label references.
6109 (aarch64_valid_floating_const): Remove unused function.
6111 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6114 * config/aarch64/aarch64.c (aarch64_layout_frame):
6115 Undo forcing of LR at bottom of frame.
6117 2017-11-03 Jeff Law <law@redhat.com>
6120 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
6121 for int_registers_saved.
6123 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
6124 extracted from tree-ssa-dom.c.
6125 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
6126 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
6127 (record_equivalences_from_incoming_edge): Add additional argument
6128 to single_pred_edge_ignoring_loop_edges call.
6129 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
6130 (uncprop_dom_walker::before_dom_children): Add additional argument
6131 to single_pred_edge_ignoring_loop_edges call.
6132 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
6133 single_pred_edge_ignoring_loop_edges rather than open coding.
6134 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
6136 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
6138 * match.pd (-(-A)): Rewrite.
6140 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
6142 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
6143 (rs6000_emit_int_cmove): New declaration.
6144 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
6145 (rs6000_emit_sISEL): Delete.
6146 (rs6000_emit_int_cmove): Make non-static.
6147 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
6148 instead of rs6000_emit_sISEL.
6150 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
6152 * asan.c (create_cond_insert_point): Maintain profile.
6153 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
6155 * basic-block.h (struct basic_block_def): Remove frequency.
6156 (EDGE_FREQUENCY): Use to_frequency
6157 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
6159 (find_traces): Update to use to_frequency.
6160 (find_traces_1_round): Likewise; use only IPA counts.
6161 (bb_to_key): Likewise.
6162 (connect_traces): Use IPA counts only.
6163 (copy_bb_p): Update to use to_frequency.
6164 (fix_up_crossing_landing_pad): Likewise.
6165 (sanitize_hot_paths): Likewise.
6166 * bt-load.c (basic_block_freq): Likewise.
6167 * cfg.c (init_flow): Set count_max to uninitialized.
6168 (check_bb_profile): Remove frequencies; check counts.
6169 (dump_bb_info): Do not dump frequencies.
6170 (update_bb_profile_for_threading): Update counts only.
6171 (scale_bbs_frequencies_int): Likewise.
6172 (MAX_SAFE_MULTIPLIER): Remove.
6173 (scale_bbs_frequencies_gcov_type): Update counts only.
6174 (scale_bbs_frequencies_profile_count): Update counts only.
6175 (scale_bbs_frequencies): Update counts only.
6176 * cfg.h (struct control_flow_graph): Add count-max.
6177 (update_bb_profile_for_threading): Update prototype.
6178 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
6179 (find_many_sub_basic_blocks): Likewise.
6180 * cfgcleanup.c (try_forward_edges): Likewise.
6181 (try_crossjump_to_edge): Likewise.
6182 * cfgexpand.c (expand_gimple_cond): Likewise.
6183 (expand_gimple_tailcall): Likewise.
6184 (construct_init_block): Likewise.
6185 (construct_exit_block): Likewise.
6186 * cfghooks.c (verify_flow_info): Check consistency of counts.
6187 (dump_bb_for_graph): Do not dump frequencies.
6188 (split_block_1): Do not update frequencies.
6189 (split_edge): Do not update frequencies.
6190 (make_forwarder_block): Do not update frequencies.
6191 (duplicate_block): Do not update frequencies.
6192 (account_profile_record): Do not update frequencies.
6193 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
6194 for global heuristics.
6195 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
6196 (expected_loop_iterations_unbounded): Use counts only.
6197 * cfgloopmanip.c (scale_loop_profile): Simplify.
6198 (create_empty_loop_on_edge): Simplify
6200 (duplicate_loop_to_header_edge): Simplify
6201 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
6202 (update_br_prob_note): Take care of removing note when profile
6204 (relink_block_chain): Do not dump frequency.
6205 (rtl_account_profile_record): Use to_frequency.
6206 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
6207 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
6208 (cgraph_update_edges_for_call_stmt_node): Likewise.
6209 (cgraph_edge::verify_count_and_frequency): Update.
6210 (cgraph_node::verify_node): Temporarily disable frequency verification.
6211 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
6212 to_cgraph_frequency.
6213 (cgraph_edge::rebuild_edges): Convert to ipa counts.
6214 * cgraphunit.c (init_lowered_empty_function): Do not initialize
6216 (cgraph_node::expand_thunk): Update profile.
6217 * except.c (dw2_build_landing_pads): Do not update frequency.
6218 * final.c (compute_alignments): Use to_frequency.
6219 (dump_basic_block_info): Do not dump frequency.
6220 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
6221 (dump_gimple_bb_header): Do not dump frequency.
6222 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
6224 * gimple-streamer-in.c (input_bb): Do not stream frequency.
6225 * gimple-streamer-out.c (output_bb): Do not stream frequency.
6226 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
6227 (init_before_recovery): Do not update frequency.
6228 (sched_create_recovery_edges): Do not update frequency.
6229 * hsa-gen.c (convert_switch_statements): Do not update frequency.
6230 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
6231 (ipa_cp_c_finalize): Set max_count to uninitialized.
6232 * ipa-fnsummary.c (get_minimal_bb): Use counts.
6233 (param_change_prob): Use counts.
6234 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
6236 * ipa-split.c (consider_split): Use to_frequency.
6237 (split_function): Use to_frequency.
6238 * ira-build.c (loop_compare_func): Likewise.
6239 (mark_loops_for_removal): Likewise.
6240 (mark_all_loops_for_removal): Likewise.
6241 * loop-doloop.c (doloop_modify): Do not update frequency.
6242 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
6244 * lto-streamer-in.c (input_function): Update count_max.
6245 * omp-expand.c (expand_omp_taskreg): Update count_max.
6246 * omp-simd-clone.c (simd_clone_adjust): Update profile.
6247 * predict.c (maybe_hot_frequency_p): Use to_frequency.
6248 (maybe_hot_count_p): Use ipa counts only.
6249 (maybe_hot_bb_p): Simplify.
6250 (maybe_hot_edge_p): Simplify.
6251 (probably_never_executed): Do not take frequency argument.
6252 (probably_never_executed_bb_p): Do not pass frequency.
6253 (probably_never_executed_edge_p): Likewise.
6254 (combine_predictions_for_bb): Check that profile is nonzero.
6255 (propagate_freq): Do not set frequency.
6256 (drop_profile): Simplify.
6257 (counts_to_freqs): Simplify.
6258 (expensive_function_p): Use to_frequency.
6259 (propagate_unlikely_bbs_forward): Simplify.
6260 (determine_unlikely_bbs): Simplify.
6261 (estimate_bb_frequencies): Add hack to silence graphite issues.
6262 (compute_function_frequency): Use ipa counts.
6263 (pass_profile::execute): Update.
6264 (rebuild_frequencies): Use counts only.
6265 (force_edge_cold): Use counts only.
6266 * profile-count.c (profile_count::dump): Dump new count types.
6267 (profile_count::differs_from_p): Check compatiblity.
6268 (profile_count::to_frequency): New function.
6269 (profile_count::to_cgraph_frequency): New function.
6270 * profile-count.h (struct function): Declare.
6271 (enum profile_quality): Add profile_guessed_local and
6272 profile_guessed_global0.
6273 (class profile_proability): Decrease number of bits to 29;
6274 update from_reg_br_prob_note and to_reg_br_prob_note.
6275 (class profile_count: Update comment; decrease number of bits
6276 to 61. Check compatibility.
6277 (profile_count::compatible_p): New private member function.
6278 (profile_count::ipa_p): New member function.
6279 (profile_count::operator<): Handle global zero correctly.
6280 (profile_count::operator>): Handle global zero correctly.
6281 (profile_count::operator<=): Handle global zero correctly.
6282 (profile_count::operator>=): Handle global zero correctly.
6283 (profile_count::nonzero_p): New member function.
6284 (profile_count::force_nonzero): New member function.
6285 (profile_count::max): New member function.
6286 (profile_count::apply_scale): Handle IPA scalling.
6287 (profile_count::guessed_local): New member function.
6288 (profile_count::global0): New member function.
6289 (profile_count::ipa): New member function.
6290 (profile_count::to_frequency): Declare.
6291 (profile_count::to_cgraph_frequency): Declare.
6292 * profile.c (OVERLAP_BASE): Delete.
6293 (compute_frequency_overlap): Delete.
6294 (compute_branch_probabilities): Do not use compute_frequency_overlap.
6295 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
6296 * sched-ebb.c (rank): Use counts only.
6297 * shrink-wrap.c (handle_simple_exit): Use counts only.
6298 (try_shrink_wrapping): Use counts only.
6299 (place_prologue_for_one_component): Use counts only.
6300 * tracer.c (find_best_predecessor): Use to_frequency.
6301 (find_trace): Use to_frequency.
6302 (tail_duplicate): Use to_frequency.
6303 * trans-mem.c (expand_transaction): Do not update frequency.
6304 * tree-call-cdce.c: Do not update frequency.
6305 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
6306 (gimple_merge_blocks): Likewise.
6307 (gimple_split_edge): Likewise.
6308 (gimple_duplicate_sese_region): Likewise.
6309 (gimple_duplicate_sese_tail): Likewise.
6310 (move_sese_region_to_fn): Likewise.
6311 (gimple_account_profile_record): Likewise.
6312 (insert_cond_bb): Likewise.
6313 * tree-complex.c (expand_complex_div_wide): Likewise.
6314 * tree-eh.c (lower_resx): Update profile.
6315 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
6317 (initialize_cfun): Do not initialize frequencies
6318 (freqs_to_counts): Delete.
6319 (copy_cfg_body): Ignore count parameter.
6320 (copy_body): Update.
6321 (expand_call_inline): Update count_max.
6322 (optimize_inline_calls): Update count_max.
6323 (tree_function_versioning): Update count_max.
6324 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
6325 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
6327 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
6328 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
6329 (try_peel_loop): Likewise.
6330 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
6332 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
6333 (tree_transform_and_unroll_loop): Do not use frequencies
6334 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
6335 Use reliable prediction only.
6336 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
6337 * tree-ssa-sink.c (select_best_block): Use to_frequency.
6338 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
6339 probability scaling.
6340 * tree-ssa-threadupdate.c (create_block_for_threading): Do
6341 not update frequency
6342 (any_remaining_duplicated_blocks): Likewise.
6343 (update_profile): Likewise.
6344 (estimated_freqs_path): Delete.
6345 (freqs_to_counts_path): Delete.
6346 (clear_counts_path): Delete.
6347 (ssa_fix_duplicate_block_edges): Likewise.
6348 (duplicate_thread_path): Likewise.
6349 * tree-switch-conversion.c (gen_inbound_check): Use counts.
6350 * tree-tailcall.c (decrease_profile): Do not update frequency.
6351 (eliminate_tail_call): Likewise.
6352 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
6353 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
6354 (optimize_mask_stores): Likewise.
6355 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
6356 * ubsan.c (ubsan_expand_null_ifn): Update profile.
6357 (ubsan_expand_ptr_ifn): Update profile.
6358 * value-prof.c (gimple_ic): Simplify.
6359 * value-prof.h (gimple_ic): Update prototype.
6360 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
6361 * ipa-inline.c (compute_uninlined_call_time): Be sure that
6363 (want_inline_self_recursive_call_p): Likewise.
6364 (resolve_noninline_speculation): Only cummulate defined counts.
6365 (inline_small_functions): Use nonzero_p.
6366 (ipa_inline): Do not access freed node.
6368 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6370 * config/aarch64/aarch64.c (aarch64_override_options_internal):
6371 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
6373 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
6375 * config/riscv/riscv.c (riscv_legitimize_move): Handle
6376 non-legitimate address.
6378 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
6380 * config/rs6000/rs6000.md (*lt0_disi): Delete.
6381 (*lt0_<mode>di, *lt0_<mode>si): New.
6383 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
6385 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
6386 TARGET_PAIRED_FLOAT.
6388 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
6389 Jim Wilson <jim.wilson@linaro.org>
6391 * config/aarch64/aarch64-cores.def (saphira): New CPU.
6392 * config/aarch64/aarch64-tune.md: Regenerated.
6393 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
6394 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
6396 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
6398 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
6399 (arc_expand_prologue): Restore blink for millicode.
6400 * config/arc/linux.h (LINK_EH_SPEC): Defined.
6402 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
6405 * config/i386/i386.c (ix86_vector_duplicate_value): Use
6406 gen_vec_duplicate after forcing the scalar into a register.
6408 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
6410 * combine (try_combine): Print the insns input to try_combine to the
6413 2017-11-02 Steve Ellcey <sellcey@cavium.com>
6416 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
6417 Remove second argument from aarch64_process_target_attr call.
6418 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
6420 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
6422 (aarch64_handle_attr_arch): Remove second argument.
6423 (aarch64_handle_attr_cpu): Ditto.
6424 (aarch64_handle_attr_tune): Ditto.
6425 (aarch64_handle_attr_isa_flags): Ditto.
6426 (aarch64_process_one_target_attr): Ditto.
6427 (aarch64_process_target_attr): Ditto.
6428 (aarch64_option_valid_attribute_p): Remove second argument.
6429 on aarch64_process_target_attr call.
6431 2017-11-02 David Malcolm <dmalcolm@redhat.com>
6433 * diagnostic.c: Include "selftest-diagnostic.h".
6434 (selftest::assert_location_text): New function.
6435 (selftest::test_diagnostic_get_location_text): New function.
6436 (selftest::diagnostic_c_tests): Call it.
6438 2017-11-02 David Malcolm <dmalcolm@redhat.com>
6440 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
6441 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
6442 (class selftest::test_diagnostic_context): Move to...
6443 * selftest-diagnostic.c: New file.
6444 * selftest-diagnostic.h: New file.
6446 2017-11-02 James Bowman <james.bowman@ftdichip.com>
6448 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
6449 offset range for FT32B.
6450 * config/ft32/ft32.h: option "mcompress" enables relaxation.
6451 * config/ft32/ft32.md: Add TARGET_NOPM.
6452 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
6453 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
6455 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
6457 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
6459 2017-11-02 Jeff Law <law@redhat.com>
6461 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
6462 virtual keyword on FINAL OVERRIDE members.
6464 * tree-ssa-propagate.h (ssa_propagation_engine): Group
6465 virtuals together. Add virtual destructor.
6466 (substitute_and_fold_engine): Similarly.
6468 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
6470 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
6472 2017-11-02 Richard Biener <rguenther@suse.de>
6474 PR tree-optimization/82795
6475 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
6477 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6479 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
6481 (gcc_GAS_CHECK_FEATURE): Remove.
6482 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
6483 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
6484 * configure: Regenerate.
6486 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
6488 * config/arc/arc.c (hwloop_optimize): Account for empty
6491 2017-11-02 Richard Biener <rguenther@suse.de>
6494 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
6495 Truncate ARRAY_REF index and element size.
6497 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
6499 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
6501 2017-11-01 Jeff Law <law@redhat.com>
6503 * tree-ssa-ccp.c (ccp_folder): New class derived from
6504 substitute_and_fold_engine.
6505 (ccp_folder::get_value): New member function.
6506 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
6507 (ccp_fold_stmt): Remove prototype.
6508 (ccp_finalize): Call substitute_and_fold from the ccp_class.
6509 * tree-ssa-copy.c (copy_folder): New class derived from
6510 substitute_and_fold_engine.
6511 (copy_folder::get_value): Renamed from get_value.
6512 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
6513 * tree-vrp.c (vrp_folder): New class derived from
6514 substitute_and_fold_engine.
6515 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
6516 (vrp_folder::get_value): New member function.
6517 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
6518 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
6519 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
6520 provide a class interface to folder/substitute routines.
6521 (ssa_prop_fold_stmt_fn): Remove typedef.
6522 (ssa_prop_get_value_fn): Likewise.
6523 (subsitute_and_fold): Remove prototype.
6524 (replace_uses_in): Likewise.
6525 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
6526 Renamed from replace_uses_in. Call the virtual member function
6527 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
6528 (substitute_and_fold_dom_walker): Remove initialization of
6529 data member entries for calbacks. Add substitute_and_fold_engine
6530 member and initialize it.
6531 (substitute_and_fold_dom_walker::before_dom_children0: Use the
6532 member functions for get_value, replace_phi_args_in c
6533 replace_uses_in, and fold_stmt calls.
6534 (substitute_and_fold_engine::substitute_and_fold): Renamed from
6535 substitute_and_fold. Remove assert. Update ctor call.
6537 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
6538 (ssa_prop_visit_phi_fn): Likewise.
6539 (class ssa_propagation_engine): New class to provide an interface
6541 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
6543 (ssa_prop_visit_phi): Likewise.
6544 (ssa_propagation_engine::simulate_stmt): Moved into class.
6545 Call visit_phi/visit_stmt from the class rather than via
6546 file scoped static variables.
6547 (ssa_propagation_engine::simulate_block): Moved into class.
6548 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
6549 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
6550 set file scoped statics for the visit_stmt/visit_phi callbacks.
6551 * tree-complex.c (complex_propagate): New class derived from
6552 ssa_propagation_engine.
6553 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
6554 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
6555 (tree_lower_complex): Call ssa_propagate via the complex_propagate
6557 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
6558 ssa_propagation_engine.
6559 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
6560 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
6561 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
6562 * tree-ssa-copy.c (copy_prop): New class derived from
6563 ssa_propagation_engine.
6564 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
6565 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
6566 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
6567 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
6568 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
6569 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
6570 (execute_vrp): Call ssa_propagate from the vrp_prop class.
6572 2017-11-01 Jakub Jelinek <jakub@redhat.com>
6574 PR rtl-optimization/82778
6575 PR rtl-optimization/82597
6576 * compare-elim.c (struct comparison): Add in_a_setter field.
6577 (find_comparison_dom_walker::before_dom_children): Remove killed
6578 bitmap and df_simulate_find_defs call, instead walk the defs.
6579 Compute last_setter and initialize in_a_setter. Merge definitions
6580 with first initialization for a few variables.
6581 (try_validate_parallel): Use insn_invalid_p instead of
6582 recog_memoized. Return insn rather than just the pattern.
6583 (try_merge_compare): Fix up comment. Don't uselessly test if
6584 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
6586 (execute_compare_elim_after_reload): Remove df_chain_add_problem
6589 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6590 Alan Hayward <alan.hayward@arm.com>
6591 David Sherwood <david.sherwood@arm.com>
6593 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
6594 aarch64_hard_regno_nregs to get the number of registers
6597 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6598 Alan Hayward <alan.hayward@arm.com>
6599 David Sherwood <david.sherwood@arm.com>
6601 * config/aarch64/constraints.md (Upl): Rename to...
6603 * config/aarch64/aarch64.md
6604 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
6607 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6608 Alan Hayward <alan.hayward@arm.com>
6609 David Sherwood <david.sherwood@arm.com>
6611 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
6612 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
6615 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6616 Alan Hayward <alan.hayward@arm.com>
6617 David Sherwood <david.sherwood@arm.com>
6619 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
6620 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
6621 (aarch64_evpc_dup): Generate rtl direcly, rather than using
6623 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
6624 of a single element.
6625 * config/aarch64/iterators.md: Add a comment above the permute
6626 unspecs to say that they are generated directly by
6627 aarch64_expand_vec_perm_const.
6628 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
6630 2017-11-01 Nathan Sidwell <nathan@acm.org>
6632 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
6634 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
6636 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
6637 and medany code models, and describe what they do.
6639 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6641 Revert accidental duplicate:
6643 * combine.c (can_change_dest_mode): Reject changes in
6644 REGMODE_NATURAL_SIZE.
6646 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
6648 PR rtl-optimization/64682
6649 PR rtl-optimization/69567
6650 PR rtl-optimization/69737
6651 PR rtl-optimization/82683
6652 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
6653 register mentioned in the note, drop the note, unless it came from I3,
6654 in which case it should go to I3 again.
6656 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6658 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
6659 and return false if not.
6660 (clear_bytes_written_by, live_bytes_read): Update accordingly.
6662 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6664 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
6665 range is known to be empty.
6667 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6668 Alan Hayward <alan.hayward@arm.com>
6669 David Sherwood <david.sherwood@arm.com>
6671 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
6672 and CONST_VECTOR_NUNITS instead of computing the number of units from
6673 the byte sizes of the vector and element.
6674 (simplify_binary_operation_1): Likewise.
6675 (simplify_const_binary_operation): Likewise.
6676 (simplify_ternary_operation): Likewise.
6678 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6679 Alan Hayward <alan.hayward@arm.com>
6680 David Sherwood <david.sherwood@arm.com>
6682 * var-tracking.c (INT_MEM_OFFSET): Replace with...
6683 (int_mem_offset): ...this new function.
6684 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
6685 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
6686 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
6689 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6690 Alan Hayward <alan.hayward@arm.com>
6691 David Sherwood <david.sherwood@arm.com>
6693 * lower-subreg.c (interesting_mode_p): New function.
6694 (compute_costs, find_decomposable_subregs, decompose_register)
6695 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
6696 (resolve_clobber, dump_choices): Use it.
6698 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6699 Alan Hayward <alan.hayward@arm.com>
6700 David Sherwood <david.sherwood@arm.com>
6702 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
6704 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6705 Alan Hayward <alan.hayward@arm.com>
6706 David Sherwood <david.sherwood@arm.com>
6708 * alias.c (find_base_value, find_base_term): Only process integer
6709 truncations. Check the precision rather than the size.
6711 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6712 Alan Hayward <alan.hayward@arm.com>
6713 David Sherwood <david.sherwood@arm.com>
6715 * machmode.h (is_narrower_int_mode): New function
6716 * optabs.c (expand_float, expand_fix): Use it.
6717 * dwarf2out.c (rotate_loc_descriptor): Likewise.
6719 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6720 Alan Hayward <alan.hayward@arm.com>
6721 David Sherwood <david.sherwood@arm.com>
6723 * rtl.h (narrower_subreg_mode): New function.
6724 * ira-color.c (update_costs_from_allocno): Use it.
6726 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6727 Alan Hayward <alan.hayward@arm.com>
6728 David Sherwood <david.sherwood@arm.com>
6730 * optabs-query.h (convert_optab_p): New function, split out from...
6731 (convert_optab_handler): ...here.
6732 (widening_optab_handler): Delete.
6733 (find_widening_optab_handler): Remove permit_non_widening parameter.
6734 (find_widening_optab_handler_and_mode): Likewise. Provide an
6735 override that operates on mode class wrappers.
6736 * optabs-query.c (widening_optab_handler): Delete.
6737 (find_widening_optab_handler_and_mode): Remove permit_non_widening
6738 parameter. Assert that the two modes are the same class and that
6739 the "from" mode is narrower than the "to" mode. Use
6740 convert_optab_handler instead of widening_optab_handler.
6741 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
6742 instead of widening_optab_handler.
6743 * expr.c (expand_expr_real_2): Update calls to
6744 find_widening_optab_handler.
6745 * optabs.c (expand_widen_pattern_expr): Likewise.
6746 (expand_binop_directly): Take the insn_code as a parameter.
6747 (expand_binop): Only call find_widening_optab_handler for
6748 conversion optabs; use optab_handler otherwise. Update calls
6749 to find_widening_optab_handler and expand_binop_directly.
6750 Use convert_optab_handler instead of widening_optab_handler.
6751 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
6752 find_widening_optab_handler and use scalar_mode rather than
6754 (convert_plusminus_to_widen): Likewise.
6756 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6757 Alan Hayward <alan.hayward@arm.com>
6758 David Sherwood <david.sherwood@arm.com>
6760 * machmode.h (fixed_size_mode): New class.
6761 * rtl.h (get_pool_mode): Return fixed_size_mode.
6762 * gengtype.c (main): Add fixed_size_mode.
6763 * target.def (get_raw_result_mode): Return a fixed_size_mode.
6764 (get_raw_arg_mode): Likewise.
6765 * doc/tm.texi: Regenerate.
6766 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
6767 * targhooks.c (default_get_reg_raw_mode): Likewise.
6768 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
6769 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
6770 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
6771 (msp430_get_raw_result_mode): Likewise.
6772 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
6773 * dbxout.c (dbxout_parms): Require fixed-size modes.
6774 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
6775 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
6776 * omp-low.c (lower_oacc_reductions): Likewise.
6777 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
6778 (simplify_subreg): Update accordingly.
6779 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
6780 (force_const_mem): Update accordingly. Return NULL_RTX for modes
6781 that aren't fixed-size.
6782 (get_pool_mode): Return a fixed_size_mode.
6783 (output_constant_pool_2): Take a fixed_size_mode.
6785 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6786 Alan Hayward <alan.hayward@arm.com>
6787 David Sherwood <david.sherwood@arm.com>
6789 * doc/rtl.texi (vec_series): Document.
6790 (const): Say that the operand can be a vec_series.
6791 * rtl.def (VEC_SERIES): New rtx code.
6792 * rtl.h (const_vec_series_p_1): Declare.
6793 (const_vec_series_p): New function.
6794 * emit-rtl.h (gen_const_vec_series): Declare.
6795 (gen_vec_series): Likewise.
6796 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
6797 (gen_vec_series): Likewise.
6798 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
6799 * simplify-rtx.c (simplify_unary_operation): Handle negations
6801 (simplify_binary_operation_series): New function.
6802 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
6803 (test_vector_ops_series): New function.
6804 (test_vector_ops): Call it.
6805 * config/powerpcspe/altivec.md (altivec_lvsl): Use
6806 gen_const_vec_series.
6807 (altivec_lvsr): Likewise.
6808 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
6810 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6811 Alan Hayward <alan.hayward@arm.com>
6812 David Sherwood <david.sherwood@arm.com>
6814 * doc/rtl.texi (const): Update description of address constants.
6815 Say that vector constants are allowed too.
6816 * common.md (E, F): Use CONSTANT_P instead of checking for
6818 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
6819 checking for CONST_VECTOR.
6820 * expmed.c (make_tree): Use build_vector_from_val for a CONST
6822 * expr.c (expand_expr_real_2): Check for vector modes instead
6823 of checking for CONST_VECTOR.
6824 * rtl.h (const_vec_p): New function.
6825 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
6826 (unwrap_const_vec_duplicate): Handle them here too.
6828 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6829 David Malcolm <dmalcolm@redhat.com>
6830 Alan Hayward <alan.hayward@arm.com>
6831 David Sherwood <david.sherwood@arm.com>
6833 * rtl.h (vec_duplicate_p): New function.
6834 * selftest-rtl.c (assert_rtx_eq_at): New function.
6835 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
6836 (assert_rtx_eq_at): Declare.
6837 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
6838 * selftest-run-tests.c (selftest::run_tests): Call it.
6839 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
6840 (simplify_unary_operation_1): Recursively handle vector duplicates.
6841 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
6843 (simplify_subreg): Handle subregs of vector duplicates.
6844 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
6845 (selftest::simplify_rtx_c_tests): New functions.
6847 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6848 Alan Hayward <alan.hayward@arm.com>
6849 David Sherwood <david.sherwood@arm.com>
6851 * emit-rtl.h (gen_const_vec_duplicate): Declare.
6852 (gen_vec_duplicate): Likewise.
6853 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
6855 (gen_const_vector): ...here.
6856 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
6857 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
6858 whose elements are all equal.
6859 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
6860 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
6861 (simplify_relational_operation): Likewise.
6862 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
6864 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
6865 (aarch64_expand_vector_init): Likewise.
6866 * config/arm/arm.c (neon_vdup_constant): Likewise.
6867 (neon_expand_vector_init): Likewise.
6868 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
6869 (arm_block_set_unaligned_vect): Likewise.
6870 (arm_block_set_aligned_vect): Likewise.
6871 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
6872 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
6873 (expand_vec_perm_even_odd_pack): Likewise.
6874 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
6875 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
6876 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
6877 gen_const_vec_duplicate.
6878 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
6879 * config/mips/mips.c (mips_gen_const_int_vector): Use
6880 gen_const_vec_duplicate.
6881 (mips_expand_vector_init): Use CONST0_RTX.
6882 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
6883 (define_split): Use gen_const_vec_duplicate.
6884 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
6885 (define_split): Use gen_const_vec_duplicate.
6886 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
6887 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
6888 * config/spu/spu.c (spu_const): Likewise.
6890 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6891 Alan Hayward <alan.hayward@arm.com>
6892 David Sherwood <david.sherwood@arm.com>
6894 * combine.c (can_change_dest_mode): Reject changes in
6895 REGMODE_NATURAL_SIZE.
6897 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
6899 * configure.ac (--enable-libssp): New.
6900 (gcc_cv_libc_provides_ssp): Check for explicit setting before
6901 trying to determine target-specific default. Adjust indentation.
6902 * configure: Regenerated.
6903 * doc/install.texi (Configuration): Expand --disable-libssp
6906 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
6908 config/i386/i386.c (ix86_expand_epilogue): Correct stack
6911 2017-10-31 Martin Jambor <mjambor@suse.cz>
6914 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
6916 2017-10-31 David Malcolm <dmalcolm@redhat.com>
6918 * auto-profile.c (autofdo_source_profile::read): Use
6919 UNKNOWN_LOCATION rather than 0.
6920 * diagnostic-core.h (warning_at_rich_loc): Rename to...
6921 (warning_at): ...this overload.
6922 (warning_at_rich_loc_n): Rename to...
6923 (warning_n): ...this overload.
6924 (error_at_rich_loc): Rename to...
6925 (error_at): ...this overload.
6926 (pedwarn_at_rich_loc): Rename to...
6927 (pedwarn): ...this overload.
6928 (permerror_at_rich_loc): Rename to...
6929 (permerror): ...this overload.
6930 (inform_at_rich_loc): Rename to...
6931 (inform): ...this overload.
6932 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
6933 (diagnostic_n_impl_richloc): Rename to...
6934 (diagnostic_n_impl): ...this rich_location *-based decl.
6935 (inform_at_rich_loc): Rename to...
6936 (inform): ...this, and add an assertion.
6937 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
6938 (warning_at_rich_loc): Rename to...
6939 (warning_at): ...this, and add an assertion.
6940 (warning_at_rich_loc_n): Rename to...
6941 (warning_n): ...this, and add an assertion.
6942 (warning_n): Update location_t-based implementation for removal of
6943 location_t-based diagnostic_n_impl.
6944 (pedwarn_at_rich_loc): Rename to...
6945 (pedwarn): ...this, and add an assertion.
6946 (permerror_at_rich_loc): Rename to...
6947 (permerror): ...this, and add an assertion.
6948 (error_n): Update for removal of location_t-based diagnostic_n_impl.
6949 (error_at_rich_loc): Rename to...
6950 (error_at): ...this, and add an assertion.
6951 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
6952 (driver::do_spec_on_infiles): Likewise.
6953 * substring-locations.c (format_warning_va): Update for renaming
6954 of inform_at_rich_loc.
6956 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
6958 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
6959 _Float<N>X built-in functions so that the variant without the
6960 "__builtin_" prefix is only enabled for the GNU C and Objective C
6961 languages when they are in non-strict ANSI/ISO mode.
6962 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
6963 * target.def (floatn_builtin_p): Add a target hook to control
6964 whether _Float<N> and _Float<N>X built-in functions without the
6965 "__builtin_" prefix are enabled, and return true for C and
6966 Objective C in the default hook. Include langhooks.h in
6968 * targhooks.h (default_floatn_builtin_p): Likewise.
6969 * targhooks.c (default_floatn_builtin_p): Likewise.
6970 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
6971 floatn_builtin_p target hook.
6972 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
6974 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
6975 Eric Botcazou <ebotcazou@adacore.com>
6977 PR rtl-optimization/81803
6978 * lra-constraints.c (curr_insn_transform): Also reload the whole
6979 register for a strict subreg no wider than a word if this is for
6980 a WORD_REGISTER_OPERATIONS target.
6982 2017-10-31 Jason Merrill <jason@redhat.com>
6984 * gdbinit.in: Skip over inlines from timevar.h.
6986 2017-10-31 Martin Liska <mliska@suse.cz>
6988 * doc/gcov.texi: Document new option.
6989 * gcov.c (print_usage): Likewise print it.
6990 (process_args): Support the argument.
6991 (format_count): New function.
6992 (format_gcov): Use the function.
6994 2017-10-31 Martin Liska <mliska@suse.cz>
6996 * gcov.c (struct name_map): do not use typedef.
6997 Define operator== and operator<.
6998 (name_search): Remove.
6999 (name_sort): Remove.
7000 (main): Do not allocate names.
7001 (process_file): Add vertical space.
7002 (generate_results): Use std::find.
7003 (release_structures): Do not release memory.
7004 (find_source): Use std::find.
7006 2017-10-31 Martin Liska <mliska@suse.cz>
7008 * gcov.c (struct line_info): Remove it's typedef.
7009 (line_info::line_info): Add proper ctor.
7010 (line_info::has_block): Do not use a typedef.
7011 (struct source_info): Do not use typedef.
7012 (circuit): Likewise.
7013 (get_cycles_count): Likewise.
7014 (output_intermediate_file): Iterate via vector iterator.
7015 (add_line_counts): Use std::vector methods.
7016 (accumulate_line_counts): Likewise.
7017 (output_lines): Likewise.
7019 2017-10-31 Martin Liska <mliska@suse.cz>
7021 * gcov.c (struct source_info): Remove typedef.
7022 (source_info::source_info): Add proper ctor.
7023 (accumulate_line_counts): Use struct, not it's typedef.
7024 (output_gcov_file): Likewise.
7025 (output_lines): Likewise.
7026 (main): Do not allocate an array.
7027 (output_intermediate_file): Use size of vector container.
7028 (process_file): Resize the vector.
7029 (generate_results): Do not preallocate, use newly added vector
7031 (release_structures): Do not release sources.
7032 (find_source): Use vector methods.
7033 (add_line_counts): Do not use typedef.
7035 2017-10-31 Martin Liska <mliska@suse.cz>
7037 * doc/gcov.texi: Document that.
7038 * gcov.c (add_line_counts): Mark lines with a non-executed
7040 (output_line_beginning): Handle such lines.
7041 (output_lines): Pass new argument.
7042 (output_intermediate_file): Print it in intermediate format.
7044 2017-10-31 Martin Liska <mliska@suse.cz>
7046 * color-macros.h: New file.
7047 * diagnostic-color.c: Factor out color related to macros to
7049 * doc/gcov.texi: Document -k option.
7050 * gcov.c (INCLUDE_STRING): Include string.h.
7051 (print_usage): Add -k option.
7052 (process_args): Parse it.
7053 (pad_count_string): New function.
7054 (output_line_beginning): Likewise.
7055 (DEFAULT_LINE_START): New macro.
7056 (output_lines): Support color output.
7058 2017-10-31 Martin Liska <mliska@suse.cz>
7060 PR gcov-profile/82633
7061 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
7062 their interaction with GCOV infrastructure.
7063 * configure.ac: Add -fkeep-{inline,static}-functions to
7065 * configure: Regenerate.
7067 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
7070 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
7072 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
7075 * config/rs6000/rs6000.md (allocate_stack): Force update interval
7076 into a register if it does not fit into an immediate offset field.
7078 2017-10-31 Olivier Hainque <hainque@adacore.com>
7080 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
7082 2017-10-31 Julia Koval <julia.koval@intel.com>
7084 * config.gcc: Add gfniintrin.h.
7085 * config/i386/gfniintrin.h: New.
7086 * config/i386/i386-builtin-types.def
7087 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7088 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7089 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7090 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7091 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7092 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
7093 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
7094 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
7095 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
7096 V64QI_FTYPE_V64QI_V64QI_INT): New types.
7097 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
7098 * config/i386/immintrin.h: Include gfniintrin.h.
7099 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
7101 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
7103 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
7105 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
7107 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
7108 (arm_ashldi3_1bit): Remove pattern.
7109 (ashrdi3): Remove shift by 1 expansion.
7110 (arm_ashrdi3_1bit): Remove pattern.
7111 (lshrdi3): Remove shift by 1 expansion.
7112 (arm_lshrdi3_1bit): Remove pattern.
7113 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
7114 cost of ashldi3 by 1.
7115 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
7116 (<shift>di3_neon): Likewise.
7118 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7120 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
7121 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
7122 and (*aarch64_simd_mov<VQ:mode>).
7123 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
7124 pattern alternative.
7125 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
7126 attributes to match pattern alternative.
7128 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
7130 * config.gcc (powerpc*-*-*): Add emmintrin.h.
7131 * config/rs6000/emmintrin.h: New file.
7132 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
7134 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
7136 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
7137 * (movdi_vfp_cortexa8): Remove pattern.
7139 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7141 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
7143 (Specific, alpha*-dec-osf5.1): Remove.
7144 (Specific, mips-sgi-irix5): Remove.
7145 (Specific, mips-sgi-irix6): Remove.
7147 2017-10-30 Jakub Jelinek <jakub@redhat.com>
7150 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
7151 arguments to clear_bit_region_be.
7153 2017-10-30 Jim Wilson <wilson@tuliptree.org>
7155 * gimplify.c: Include memmodel.h.
7157 2017-10-30 Martin Jambor <mjambor@suse.cz>
7159 * omp-grid.c (grid_attempt_target_gridification): Also insert a
7160 condition whether loop should be executed at all.
7162 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
7164 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7165 gimple folding of vec_madd() intrinsics.
7166 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
7167 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
7168 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
7170 2017-10-30 Richard Biener <rguenther@suse.de>
7172 PR tree-optimization/82762
7174 2017-10-23 Richard Biener <rguenther@suse.de>
7176 PR tree-optimization/82129
7178 2017-08-01 Richard Biener <rguenther@suse.de>
7180 PR tree-optimization/81181
7181 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7182 (compute_antic): ... end of iteration here.
7184 2017-10-30 Joseph Myers <joseph@codesourcery.com>
7186 * doc/invoke.texi (C Dialect Options): Document -std=c17,
7187 -std=iso9899:2017 and -std=gnu17.
7188 * doc/standards.texi (C Language): Document C17 support.
7189 * doc/cpp.texi (Overview): Mention -std=c17.
7190 (Standard Predefined Macros): Document C11 and C17 values of
7191 __STDC_VERSION__. Do not refer to C99 support as incomplete.
7192 * doc/extend.texi (Inline): Do not list individual options for
7193 standards newer than C99.
7194 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
7196 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
7199 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
7201 * asan.c (asan_finish_file): Align asan globals array by shadow
7204 2017-10-30 Jakub Jelinek <jakub@redhat.com>
7207 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
7208 (struct store_immediate_info): Add bitregion_start and bitregion_end
7210 (store_immediate_info::store_immediate_info): Add brs and bre
7211 arguments and initialize bitregion_{start,end} from those.
7212 (struct merged_store_group): Add bitregion_start, bitregion_end,
7213 align_base and mask fields. Drop unnecessary struct keyword from
7214 struct store_immediate_info. Add do_merge method.
7215 (clear_bit_region_be): Use memset instead of loop storing zeros.
7216 (merged_store_group::do_merge): New method.
7217 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
7218 stores as long as the surrounding bitregions have no gaps.
7219 (merged_store_group::merge_overlapping): Use do_merge.
7220 (merged_store_group::apply_stores): Test that bitregion_{start,end}
7221 is byte aligned, rather than requiring that start and width are
7222 byte aligned. Drop unnecessary struct keyword from
7223 struct store_immediate_info. Allocate and populate also mask array.
7224 Make start of the arrays relative to bitregion_start rather than
7225 start and size them according to bitregion_{end,start} difference.
7226 (struct imm_store_chain_info): Drop unnecessary struct keyword from
7227 struct store_immediate_info.
7228 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
7229 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
7230 struct keyword from struct store_immediate_info.
7231 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
7232 between stores as long as the surrounding bitregions have no gaps.
7234 (struct split_store): Add orig non-static data member.
7235 (split_store::split_store): Initialize orig to false.
7236 (find_constituent_stmts): Return store_immediate_info *, non-NULL
7237 if there is exactly a single original stmt. Change stmts argument
7238 to pointer from reference, if NULL, don't push anything to it. Add
7239 first argument, use it to optimize skipping over orig stmts that
7240 are known to be before bitpos already. Simplify.
7241 (split_group): Return unsigned int count how many stores are or
7242 would be needed rather than a bool. Add allow_unaligned argument.
7243 Change split_stores argument from reference to pointer, if NULL,
7244 only do a dry run computing how many stores would be produced.
7245 Rewritten algorithm to use both alignment and misalign if
7246 !allow_unaligned and handle bitfield stores with gaps.
7247 (imm_store_chain_info::output_merged_store): Set start_byte_pos
7248 from bitregion_start instead of start. Compute allow_unaligned
7249 here, if true, do 2 split_group dry runs to compute which one
7250 produces fewer stores and prefer aligned if equal. Punt if
7251 new count is bigger or equal than original before emitting any
7252 statements, rather than during that. Remove no longer needed
7253 new_ssa_names tracking. Replace num_stmts with
7254 split_stores.length (). Use 32-bit stack allocated entries
7255 in split_stores auto_vec. Try to reuse original store lhs/rhs1
7256 if possible. Handle bitfields with gaps.
7257 (pass_store_merging::execute): Ignore bitsize == 0 stores.
7258 Compute bitregion_{start,end} for the stores and construct
7259 store_immediate_info with that. Formatting fixes.
7261 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
7264 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
7265 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
7267 2017-10-29 Jim Wilson <wilson@tuliptree.org>
7269 * gimplify.c: Include tm_p.h.
7271 * common.opt (gcoff): Re-add as ignored option.
7272 (gcoff1, gcoff2, gcoff3): Likewise.
7274 * Makefile.in (OBJS): Delete sdbout.o.
7275 (GTFILES): Delete $(srcdir)/sdbout.c.
7276 * debug.h: Delete sdb_debug_hooks.
7277 * final.c: Delete sdbout.h include.
7278 (final_scan_insn): Delete SDB_DEBUG check.
7279 (rest_of_clean_state): Likewise.
7280 * output.h: Delete sdb_begin_function_line.
7283 * toplev.c: Delete sdbout.h include.
7284 (process_options): Delete SDB_DEBUG check.
7285 * tree-core.h (tree_type_common): Delete pointer field of
7287 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
7288 TYPE_SYMTAB_POINTER.
7289 * tree.h (TYPE_SYMTAB_POINTER): Delete.
7290 (TYPE_SYMTAB_IS_POINTER): Delete.
7291 (TYPE_SYMTAB_IS_DIE): Renumber.
7292 * xcoffout.c: Refer to former sdbout.c file.
7293 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
7295 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
7296 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
7297 (Debugging Options): Delete -gcoff.
7298 (-gstabs): Delete SDB reference.
7300 (-gcoff@var{level}): Delete.
7301 * doc/passes.texi (Debugging information output): Delete SDB and
7302 sdbout.c references.
7303 * doc/tm.texi: Regenerate.
7304 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
7305 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
7307 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
7308 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
7309 (SDB_DEBUGGING_INFO): Delete.
7310 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
7311 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
7312 * target.def (output_source_filename): Delete COFF reference.
7314 * common.opt (gcoff): Delete.
7315 (gxcoff+): Update Negative chain.
7316 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
7318 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
7319 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
7320 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
7322 (expand_function_start): Change sdb reference to past tense.
7323 (expand_function_end): Change sdb reference to past tense.
7324 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
7325 * opts.c (debug_type_names): Delete coff entry.
7326 (common_handle_option): Delete OPT_gcoff case.
7327 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
7329 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
7330 * config/cris/cris.h: Delete SDB reference in comment.
7331 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
7332 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
7333 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
7334 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
7336 (ix86_expand_prologue): Likewise.
7337 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
7338 * config/ia64/ia64.h: Likewise.
7339 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
7340 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
7342 * config/mmix/mmix.h: Likewise.
7343 * config/nds32/nds32.c: Likewise.
7344 * config/stormy/storym16.h: Likewise.
7345 * config/visium/visium.h: Likewise.
7346 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
7348 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
7350 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
7351 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
7352 FRAME_POINTER_REGNUM point at high end of local var area.
7354 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
7356 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
7357 Move comment around. Do not reset best_edge for a copiable
7358 destination if the copy would cause a partition change.
7359 (better_edge_p): Remove redundant check.
7361 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
7363 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
7365 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
7367 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
7368 for math functions that have _Float<N> and _Float<N>X variants.
7369 (mathfn_built_in_2): Add support for math functions that have
7370 _Float<N> and _Float<N>X variants.
7371 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
7372 (expand_builtin_mathfn_ternary): Add support for fma with
7373 _Float<N> and _Float<N>X variants.
7374 (expand_builtin): Likewise.
7375 (fold_builtin_3): Likewise.
7376 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
7377 create math function _Float<N> and _Float<N>X variants as external
7379 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
7380 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
7381 the __builtin_ prefix and if not strict ansi, without the prefix.
7382 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
7383 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
7384 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
7385 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
7386 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
7387 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
7388 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
7389 function signatures for fma _Float<N> and _Float<N>X variants.
7390 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
7391 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
7392 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
7393 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
7394 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
7395 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
7396 * gencfn-macros.c (print_case_cfn): Add support for math functions
7397 that have _Float<N> and _Float<N>X variants.
7398 (print_define_operator_list): Likewise.
7399 (fltfn_suffixes): Likewise.
7401 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
7402 for math functions that have _Float<N> and _Float<N>X variants.
7403 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
7404 and _Float<N>X variants.
7405 (COPYSIGN): Likewise.
7408 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
7409 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
7411 (integer_valued_read_call_p): Likewise.
7412 * fold-const-call.c (fold_const_call_ss): Likewise.
7413 (fold_const_call_sss): Add support for copysign, fmin, and fmax
7414 _Float<N> and _Float<N>X variants.
7415 (fold_const_call_ssss): Add support for fma _Float<N> and
7416 _Float<N>X variants.
7417 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
7418 support for copysign and fma _Float<N> and _Float<N>X variants.
7419 (backprop::process_builtin_call_use): Likewise.
7420 * tree-call-cdce.c (can_test_argument_range); Add support for
7421 sqrt _Float<N> and _Float<N>X variants.
7422 (edom_only_function): Likewise.
7423 (get_no_error_domain): Likewise.
7424 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
7425 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
7426 copysign _Float<N> and _Float<N>X variants.
7427 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
7428 handled by machine independent code.
7429 (FMAF128): Likewise.
7430 * doc/cpp.texi (Common Predefined Macros): Document defining
7431 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
7432 fma _Float<N> and _Float<N>X variants.
7434 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
7437 * config/i386/i386-modes.def (CCFPU): Remove definition.
7438 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
7439 (ix86_cc_modes_compatible): Ditto.
7440 (ix86_expand_carry_flag_compare): Ditto.
7441 (ix86_expand_int_movcc): Ditto.
7442 (ix86_expand_int_addcc): Ditto.
7443 (ix86_reverse_condition): Ditto.
7444 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
7445 Return true/false for unordered/ordered fp comparisons.
7446 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
7447 (ix86_prepare_fp_compare_args): Update for rename.
7448 (ix86_expand_fp_compare): Update for rename. Generate unordered
7449 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
7450 (ix86_expand_sse_compare_and_jump): Ditto.
7451 * config/i386/predicates.md (fcmov_comparison_operator):
7452 Remove CCFPU mode handling.
7453 (ix86_comparison_operator): Ditto.
7454 (ix86_carry_flag_operator): Ditto.
7455 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
7456 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
7457 (*cmpu<mode>_cc_i387): Ditto.
7458 (FPCMP): Remove mode iterator.
7459 (unord): Remove mode attribute.
7460 (unord_subst): New define_subst transformation
7461 (unord): New define_subst attribute.
7463 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
7464 (*cmpi<unord>xf_i387): Ditto.
7465 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
7466 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
7467 using unord_subst transformation.
7468 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
7469 (round_saeonly): Also handle CCFP mode.
7470 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
7471 Remove UNSPEC_SAHF unspec handling.
7473 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
7475 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
7477 2017-10-27 Jeff Law <law@redhat.com>
7479 * gimple-ssa-sprintf.c: Include domwalk.h.
7480 (class sprintf_dom_walker): New class, derived from dom_walker.
7481 (sprintf_dom_walker::before_dom_children): New function.
7482 (struct call_info): Moved into sprintf_dom_walker class
7483 (compute_formath_length, handle_gimple_call): Likewise.
7484 (sprintf_length::execute): Call the dominator walker rather
7485 than walking the statements.
7487 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
7488 gimple statement locations.
7489 (check_array_bounds): Corresponding changes. Get the statement's
7490 location directly from wi->stmt.
7492 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
7495 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
7497 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
7499 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
7500 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
7502 2017-10-27 Jakub Jelinek <jakub@redhat.com>
7505 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
7506 * config/i386/i386.c (maybe_get_pool_constant): Removed.
7507 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
7508 maybe_get_pool_constant.
7509 * config/i386/predicates.md (zero_extended_scalar_load_operand):
7512 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7514 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
7515 2.26 caveat. Update gas and gld versions.
7516 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
7519 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7521 * cgraph.h (set_malloc_flag): Declare.
7522 * cgraph.c (set_malloc_flag_1): New function.
7523 (set_malloc_flag): Likewise.
7524 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
7525 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
7527 (read_ipa_call_summary): Add support for reading is_return_callee.
7528 (write_ipa_call_summary): Stream is_return_callee.
7529 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
7530 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
7531 ipa-prop.h, ipa-fnsummary.h.
7532 (pure_const_names): Change to static.
7533 (malloc_state_e): Define.
7534 (malloc_state_names): Define.
7535 (funct_state_d): Add field malloc_state.
7536 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
7537 (check_retval_uses): New function.
7538 (malloc_candidate_p): Likewise.
7539 (analyze_function): Add support for malloc attribute.
7540 (pure_const_write_summary): Stream malloc_state.
7541 (pure_const_read_summary): Add support for reading malloc_state.
7542 (dump_malloc_lattice): New function.
7543 (propagate_malloc): New function.
7544 (warn_function_malloc): New function.
7545 (ipa_pure_const::execute): Call propagate_malloc and
7546 ipa_free_fn_summary.
7547 (pass_local_pure_const::execute): Add support for malloc attribute.
7548 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
7549 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
7551 2017-10-27 Martin Liska <mliska@suse.cz>
7553 PR gcov-profile/82457
7554 * doc/invoke.texi: Document that one needs a non-strict ISO mode
7555 for fork-like functions to be properly instrumented.
7557 2017-10-27 Richard Biener <rguenther@suse.de>
7560 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
7561 info when we redirected EH.
7563 2017-10-26 Michael Collison <michael.collison@arm.com>
7565 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
7567 (<optab>_trunchf<GPI:mode>2: New pattern.
7568 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
7569 * config/aarch64/iterators.md (wv): New mode attribute.
7570 (vf, VF): New mode attributes.
7571 (vgp, VGP): New mode attributes.
7572 (s): Update attribute with SImode and DImode prefixes.
7574 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7576 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
7577 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
7578 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
7579 (nios2_option_overide): Initialize it. Don't allow R0-relative
7580 addressing with PIC.
7581 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
7582 (nios2_symbolic_constant_p): Likewise.
7583 (nios2_legitimate_address_p): Likewise.
7584 (nios2_r0rel_section_name_p): New.
7585 (nios2_symbol_ref_in_r0rel_data_p): New.
7586 (nios2_emit_move_sequence): Handle r0rel_constant_p.
7587 (r0rel_constant_p): New.
7588 (nios2_print_operand_address): Handle r0rel_constant_p.
7589 (nios2_cdx_narrow_form_p): Likewise.
7590 * config/nios2/nios2.opt (mr0rel-sec=): New option.
7591 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
7592 (Nios II Options): Document -mr0rel-sec.
7594 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7596 * config/nios2/nios2.c: Include xregex.h.
7597 (nios2_gprel_sec_regex): New.
7598 (nios2_option_overide): Initialize it. Don't allow GP-relative
7599 addressing with PIC.
7600 (nios2_small_section_name_p): Check for regex match.
7601 * config/nios2/nios2.opt (mgprel-sec=): New option.
7602 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
7603 (Nios II Options): Document -mgprel-sec.
7605 2017-10-26 Jim Wilson <wilson@tuliptree.org>
7607 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
7609 2017-10-26 Tom de Vries <tom@codesourcery.com>
7611 PR tree-optimization/82707
7612 * gimple.c (gimple_copy): Fix unsharing of
7613 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
7615 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
7617 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
7618 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
7619 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
7620 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
7621 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
7622 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
7623 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
7624 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
7625 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
7626 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
7627 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
7628 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
7629 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
7630 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
7631 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
7632 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
7633 _mm512_mask_cmpunord_ps_mask): New intrinsics.
7635 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7637 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
7639 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
7640 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
7641 warning to rs6000.c. Remove the Undocumented flag, since it has
7643 (-mabi=ibmlongdouble): Likewise.
7644 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
7645 already set, set the default format for long double.
7646 (rs6000_debug_reg_global): Print whether long double is IBM or
7648 (rs6000_option_override_internal): Rework setting long double
7649 format. Only warn if the user is changing the long double default
7650 and they did not use -Wno-psabi.
7651 * doc/invoke.texi (PowerPC options): Update the documentation for
7652 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
7654 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7655 Alan Hayward <alan.hayward@arm.com>
7656 David Sherwood <david.sherwood@arm.com>
7658 * rtl.h (wider_subreg_mode): New function.
7659 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
7660 rather than an unsigned int *.
7661 * ira-color.c (regno_max_ref_width): Replace with...
7662 (regno_max_ref_mode): ...this new variable.
7663 (coalesced_pseudo_reg_slot_compare): Update accordingly.
7664 Use wider_subreg_mode.
7665 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
7666 rather than an unsigned int *.
7667 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
7668 (process_alt_operands): Likewise.
7669 (invariant_p): Likewise.
7670 * lra-spills.c (assign_mem_slot): Likewise.
7671 (add_pseudo_to_slot): Likewise.
7672 * lra.c (collect_non_operand_hard_regs): Likewise.
7673 (add_regs_to_insn_regno_info): Likewise.
7674 * reload1.c (regno_max_ref_width): Replace with...
7675 (regno_max_ref_mode): ...this new variable.
7676 (reload): Update accordingly. Update call to
7677 ira_sort_regnos_for_alter_reg.
7678 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
7679 (init_eliminable_invariants): Update to use regno_max_ref_mode.
7680 (scan_paradoxical_subregs): Likewise.
7682 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7684 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
7685 (aarch64_frame): Add emit_frame_chain boolean.
7686 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7687 Move eh_return case to aarch64_layout_frame.
7688 (aarch64_layout_frame): Initialize emit_frame_chain.
7689 (aarch64_expand_prologue): Use emit_frame_chain.
7691 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7693 * config/aarch64/aarch64.c (aarch64_layout_frame):
7694 Ensure LR is always stored at the bottom of the callee-saves.
7695 Remove rarely used frame layout which saves callee-saves at top of
7696 frame, so the store of LR can be used as a valid probe in all cases.
7698 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7700 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
7701 Improve unaligned TImode/TFmode base/offset split.
7703 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7704 Alan Hayward <alan.hayward@arm.com>
7705 David Sherwood <david.sherwood@arm.com>
7707 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
7708 * combine.c (find_single_use_1): Likewise.
7709 (expand_field_assignment): Likewise.
7710 (move_deaths): Likewise.
7711 * lra-constraints.c (simplify_operand_subreg): Likewise.
7712 (curr_insn_transform): Likewise.
7713 * lra.c (collect_non_operand_hard_regs): Likewise.
7714 (add_regs_to_insn_regno_info): Likewise.
7715 * rtlanal.c (reg_referenced_p): Likewise.
7716 (covers_regno_no_parallel_p): Likewise.
7718 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7720 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
7721 Don't print any bits outside the precision of the value.
7722 * wide-int.cc (test_printing): Add some new tests.
7724 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7726 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
7727 supports -xbrace_comment option.
7728 * configure: Regenerate.
7729 * config.in: Regenerate.
7730 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
7731 (ASM_CPU_SPEC): Use it.
7733 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7735 * target.def (static_rtx_alignment): New hook.
7736 * targhooks.h (default_static_rtx_alignment): Declare.
7737 * targhooks.c (default_static_rtx_alignment): New function.
7738 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
7739 * doc/tm.texi: Regenerate.
7740 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
7741 instead of targetm.constant_alignment. Remove call to
7743 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7744 (cris_preferred_mininum_alignment): New function, split out from...
7745 (cris_constant_alignment): ...here.
7746 (cris_static_rtx_alignment): New function.
7747 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
7749 (ix86_constant_alignment): ...here.
7750 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7751 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7752 (mmix_static_rtx_alignment): New function.
7753 * config/spu/spu.c (spu_static_rtx_alignment): New function.
7754 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7756 2017-10-26 Tamar Christina <tamar.christina@arm.com>
7759 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
7760 Add flag_trapping_math and flag_fp_int_builtin_inexact.
7762 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
7764 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
7765 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
7766 mark as a sign-extending load.
7767 (local_pic_load_u): Define.
7769 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
7772 * fold-const.c (operand_equal_for_comparison_p): Also return true
7773 if ARG0 is a simple variant of ARG1 with narrower precision.
7774 (fold_ternary_loc): Always pass unstripped operands to the predicate.
7776 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
7778 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
7780 * i386.h (processor_costs): Add gather_static, gather_per_elt,
7781 scatter_static, scatter_per_elt.
7782 * x86-tune-costs.h: Add new cost entries.
7784 2017-10-25 Richard Biener <rguenther@suse.de>
7786 * tree-ssa-sccvn.h (vn_eliminate): Declare.
7787 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
7788 class pass_fre): Move to ...
7789 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
7790 class pass_fre): ... here and adjust for statistics.
7792 2017-10-25 Jakub Jelinek <jakub@redhat.com>
7795 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
7796 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
7797 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
7798 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
7801 2017-10-25 Richard Biener <rguenther@suse.de>
7803 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
7804 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
7805 eliminate_push_avail, eliminate_insert): Move inside...
7806 (class eliminate_dom_walker): ... this class in preparation
7808 (fini_eliminate): Remove by merging with ...
7809 (eliminate): ... this function. Adjust for class changes.
7810 (pass_pre::execute): Remove fini_eliminate call.
7811 (pass_fre::execute): Likewise.
7813 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7816 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
7817 (VPERMI2, VPERMI2I): New mode iterators.
7818 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
7819 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
7821 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
7822 mode iterator. Remove 3 old define_insn patterns.
7823 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
7824 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
7825 VPERMI2 mode iterator, remove the other two expanders.
7826 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
7827 to use VPERMI2 mode iterator, add another alternative for vpermi2*
7828 instructions, remove the other two patterns.
7829 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
7830 mode iterator, remove the other two patterns.
7831 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
7832 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
7833 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
7834 and adjust argument order accordingly.
7835 (ix86_expand_vec_perm): Adjust caller.
7836 (expand_vec_perm_1): Likewise.
7837 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
7838 (expand_vec_perm_vpermt2_vpshub2): ... this.
7839 (ix86_expand_vec_perm_const_1): Adjust caller.
7840 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
7843 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
7844 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
7845 (vec_shl_<mode>): Remove unused expander.
7846 (avx512bw_<shift_insn><mode>3): New define_insn.
7847 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
7848 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
7850 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
7853 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
7856 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7858 PR rtl-optimization/82396
7859 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
7860 (autopref_multipass_init): Simplify initialization.
7861 (autopref_rank_data): Simplify sort order.
7862 * gcc/sched-int.h (autopref_multipass_data_): Remove
7863 multi_mem_insn_p, min_offset and max_offset.
7865 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7868 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7869 Check special value of flag_omit_frame_pointer.
7870 (aarch64_can_eliminate): Likewise.
7871 (aarch64_override_options_after_change_1): Simplify handling of
7872 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
7874 2017-10-24 Richard Biener <rguenther@suse.de>
7876 PR tree-optimization/82697
7877 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
7878 zero for conditional load and unconditional store.
7880 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7882 * doc/install.texi: Document bootstrap-cet.
7884 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7887 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
7888 ENDBR instruction at function entrance if function is only
7891 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7894 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
7895 patterns to better describe from which operation the CF is computed.
7896 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
7897 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
7898 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
7899 is 0, use _0 suffixed expanders instead of emitting a comparison
7902 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
7904 * config/i386/i386.md(*movsf_internal, *movdf_internal):
7905 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
7907 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
7910 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
7911 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
7912 * loop-iv.c (iv_get_reaching_def): Likewise.
7913 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
7914 variable is promoted and the partition contains undefined values.
7916 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7918 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
7920 (nios2_address_cost): Define.
7921 (nios2_legitimize_address): Recognize (exp + constant) directly.
7922 (TARGET_ADDRESS_COST): Define.
7924 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7926 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
7927 (nios2_symbolic_memory_operand_p): Declare.
7928 (nios2_split_large_constant): Declare.
7929 (nios2_split_symbolic_memory_operand): Declare.
7930 * config/nios2/nios2.c: Adjust includes.
7931 (nios2_symbolic_constant_allowed): New.
7932 (nios2_symbolic_constant_p): New.
7933 (nios2_plus_symbolic_constant_p): New.
7934 (nios2_valid_addr_expr_p): Recognize addresses involving
7936 (nios2_legitimate_address_p): Likewise, also LO_SUM.
7937 (nios2_symbolic_memory_operand_p): New.
7938 (nios2_large_constant_p): New.
7939 (nios2_split_large_constant): New.
7940 (nios2_split_plus_large_constant): New.
7941 (nios2_split_symbolic_memory_operand): New.
7942 (nios2_legitimize_address): Code refactoring. Handle addresses
7943 involving symbolic constants.
7944 (nios2_emit_move_sequence): Likewise.
7945 (nios2_print_operand): Improve error output.
7946 (nios2_print_operand_address): Handle LO_SUM.
7947 (nios2_cdx_narrow_form_p): Likewise.
7948 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
7949 operands involving symbolic constants.
7950 (movhi_internal, movsi_internal): Likewise.
7951 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
7952 (extendhisi2, extendqi<mode>2): Likewise.
7954 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7956 * tree-pass.h (PROP_rtl_split_insns): Define.
7957 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
7959 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7961 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
7963 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7966 * target.def (const_not_ok_for_debug_p): Default to
7967 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
7968 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
7969 * targhooks.c (default_const_not_ok_for_debug_p): New function.
7970 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
7971 which targetm.const_not_ok_for_debug_p returned true.
7972 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
7974 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
7976 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
7977 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
7978 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
7980 (ix86_const_not_ok_for_debug_p): New function.
7981 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
7982 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
7984 2017-10-23 David Malcolm <dmalcolm@redhat.com>
7987 * system.h: Conditionally include "unique-ptr.h" if
7988 INCLUDE_UNIQUE_PTR is defined.
7989 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
7990 of defining INCLUDE_UNIQUE_PTR before including "system.h".
7992 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
7994 * config/rl78/rl78.md: New define_expand "subdi3".
7996 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
7999 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
8000 DF_REF_INSN if DF_REF_INSN_INFO is false.
8002 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
8004 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
8005 xmm_move instead of sse_move.
8006 (sse_store_index): New function.
8007 (ix86_register_move_cost): Be more sensible about mismatch stall;
8008 model AVX moves correctly; make difference between sse->integer and
8010 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
8011 moves; make difference between SSE and AVX.
8012 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
8013 and zmm_move. Increase size of sse load and store tables;
8014 add unaligned load and store tables; add ssemmx_to_integer.
8015 * x86-tune-costs.h: Update all entries according to real
8016 move latencies from Agner Fog's manual and chip documentation.
8018 2017-10-23 Jakub Jelinek <jakub@redhat.com>
8021 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
8022 * config/i386/constraints.md (Wf): New constraint.
8023 * config/i386/i386.md (UNSPEC_SBB): New unspec.
8024 (cmp<dwi>_doubleword): Removed.
8025 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
8026 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
8027 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
8028 expand with cmp<dwi>_doubleword. For LTU and GEU use
8029 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
8031 * common.opt (gcolumn-info): Enable by default.
8032 * doc/invoke.texi (gcolumn-info): Document new default.
8034 2017-10-23 Richard Biener <rguenther@suse.de>
8036 PR tree-optimization/82672
8037 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
8038 Fold the stmt if we propagated into it.
8040 2017-10-23 Richard Biener <rguenther@suse.de>
8042 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
8043 (bitmap_remove_expr_from_set): ... this. All callers call this
8044 for non-constant values.
8045 (bitmap_set_subtract): Rename to...
8046 (bitmap_set_subtract_expressions): ... this. Adjust and
8048 (bitmap_set_contains_value): Remove superfluous check.
8049 (bitmap_set_replace_value): Inline into single caller ...
8050 (bitmap_value_replace_in_set): ... here and simplify.
8051 (dependent_clean): Merge into ...
8052 (clean): ... this using an overload. Adjust.
8053 (prune_clobbered_mems): Adjust.
8054 (compute_antic_aux): Likewise.
8055 (compute_partial_antic_aux): Likewise.
8057 2017-10-23 Richard Biener <rguenther@suse.de>
8059 PR tree-optimization/82129
8061 2017-08-01 Richard Biener <rguenther@suse.de>
8063 PR tree-optimization/81181
8064 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8065 (compute_antic): ... end of iteration here.
8067 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8069 * target.def (starting_frame_offset): New hook.
8070 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
8071 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
8072 * doc/tm.texi.in: Regenerate.
8073 * hooks.h (hook_hwi_void_0): Declare.
8074 * hooks.c (hook_hwi_void_0): New function.
8075 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
8076 STARTING_FRAME_OFFSET.
8077 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
8078 * reload1.c (reload): Likewise.
8079 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
8080 instead of STARTING_FRAME_OFFSET.
8081 * function.c (try_fit_stack_local): Likewise.
8082 (assign_stack_local_1): Likewise
8083 (instantiate_virtual_regs): Likewise.
8084 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
8085 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
8086 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
8087 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
8088 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
8089 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
8090 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
8091 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
8092 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
8093 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
8094 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
8095 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
8096 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
8097 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
8098 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
8099 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
8100 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
8101 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
8102 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
8103 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
8104 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
8105 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
8106 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
8107 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
8108 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
8109 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
8110 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
8111 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
8112 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
8113 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
8114 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
8115 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
8116 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
8117 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
8118 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
8119 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
8120 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
8121 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
8122 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
8123 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
8124 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
8125 * config/avr/avr.c (avr_starting_frame_offset): Make static and
8126 return a HOST_WIDE_INT.
8127 (avr_builtin_setjmp_frame_value): Use it instead of
8128 STARTING_FRAME_OFFSET.
8129 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8130 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
8131 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
8133 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8134 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
8135 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
8136 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8137 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
8138 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
8139 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8140 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
8141 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
8142 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8143 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
8144 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
8146 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8147 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
8148 * config/mips/mips.c (mips_compute_frame_info): Refer to
8149 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
8150 (mips_starting_frame_offset): New function.
8151 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8152 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
8153 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
8154 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
8155 and return a HOST_WIDE_INT.
8156 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8157 (mmix_initial_elimination_offset): Refer to
8158 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
8159 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
8160 * config/pa/pa.c (pa_starting_frame_offset): New function.
8161 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
8162 (pa_expand_prologue): Likewise.
8163 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8164 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
8165 !FRAME_GROWS_DOWNWARD handling to...
8166 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8167 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
8168 !FRAME_GROWS_DOWNWARD handling to...
8169 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8170 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
8171 !FRAME_GROWS_DOWNWARD handling to...
8172 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8173 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
8175 (rs6000_starting_frame_offset): New function.
8176 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
8177 !FRAME_GROWS_DOWNWARD handling to...
8178 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8179 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
8180 !FRAME_GROWS_DOWNWARD handling to...
8181 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8182 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
8183 !FRAME_GROWS_DOWNWARD handling to...
8184 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8185 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
8186 (rs6000_starting_frame_offset): New function.
8187 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
8188 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
8189 * config/vax/vax.c (vax_starting_frame_offset): New function.
8190 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
8191 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8192 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
8193 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
8194 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8195 * system.h (STARTING_FRAME_OFFSET): Poison.
8197 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8199 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
8200 SCALAR_TYPE_MODE instead of TYPE_MODE.
8202 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8203 Alan Hayward <alan.hayward@arm.com>
8204 David Sherwood <david.sherwood@arm.com>
8206 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
8208 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8209 Alan Hayward <alan.hayward@arm.com>
8210 David Sherwood <david.sherwood@arm.com>
8212 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
8214 2017-10-23 Richard Biener <rguenther@suse.de>
8216 PR tree-optimization/82129
8217 * tree-ssa-pre.c (bitmap_set_and): Remove.
8218 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
8219 canonicalizing expressions in the set to those with lowest
8220 ID rather than taking that from the first edge.
8222 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8224 * combine.c (rtx_equal_for_field_assignment_p): Use
8225 byte_lowpart_offset.
8227 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8228 Alan Hayward <alan.hayward@arm.com>
8229 David Sherwood <david.sherwood@arm.com>
8231 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
8232 to a SUBREG_PROMOTED_VAR.
8234 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8235 Alan Hayward <alan.hayward@arm.com>
8236 David Sherwood <david.sherwood@arm.com>
8238 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
8239 (expand_debug_source_expr): Likewise.
8240 * combine.c (combine_simplify_rtx): Likewise.
8241 * cse.c (fold_rtx): Likewise.
8242 * optabs.c (expand_float): Likewise.
8243 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8244 (simplify_binary_operation_1): Likewise.
8246 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8247 Alan Hayward <alan.hayward@arm.com>
8248 David Sherwood <david.sherwood@arm.com>
8250 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
8251 (record_promoted_value): Likewise.
8252 * expr.c (expand_expr_real_2): Likewise.
8253 * ree.c (update_reg_equal_equiv_notes): Likewise.
8254 (combine_set_extension): Likewise.
8255 * rtlanal.c (low_bitmask_len): Likewise.
8256 * simplify-rtx.c (neg_const_int): Likewise.
8257 (simplify_binary_operation_1): Likewise.
8259 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8260 Alan Hayward <alan.hayward@arm.com>
8261 David Sherwood <david.sherwood@arm.com>
8263 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
8264 * regcprop.c (maybe_mode_change): Likewise.
8265 * reload1.c (alter_reg): Likewise.
8267 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8269 * inchash.h (inchash::hash::add_wide_int): New function.
8270 * lto-streamer-out.c (hash_tree): Use it.
8272 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8274 * inchash.h (inchash::hash::add_wide_int): Rename to...
8275 (inchash::hash::add_hwi): ...this.
8276 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
8277 (polymorphic_call_target_hasher::hash): Likewise.
8278 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
8279 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
8280 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
8281 * lto-streamer-out.c (hash_tree): Likewise.
8282 * optc-save-gen.awk: Likewise.
8283 * tree.c (add_expr): Likewise.
8285 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
8288 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
8289 for ordered inequality comparisons even with TARGET_IEEE_FP.
8291 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
8294 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
8295 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
8296 Expand with cmp<dwi>_doubleword.
8298 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8300 * extend.texi: Add x86 specific to 'nocf_check' attribute.
8301 List CET intrinsics.
8302 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
8303 specific to -fcf-protection option.
8305 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8307 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
8308 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
8309 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
8310 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
8311 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
8312 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
8313 (extra_objs): Add cet.o for Linux/x86 targets.
8314 (tmake_file): Add i386/t-cet for Linux/x86 targets.
8315 * config/i386/cet.c: New file.
8316 * config/i386/cetintrin.h: Likewise.
8317 * config/i386/t-cet: Likewise.
8318 * config/i386/cpuid.h (bit_SHSTK): New.
8319 (bit_IBT): Likewise.
8320 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
8321 pass IBT and SHSTK bits.
8322 * config/i386/i386-builtin-types.def
8323 (VOID_FTYPE_UNSIGNED_PVOID): New.
8324 (VOID_FTYPE_UINT64_PVOID): Likewise.
8325 * config/i386/i386-builtin.def: Add CET intrinsics.
8326 * config/i386/i386-c.c (ix86_target_macros_internal): Add
8327 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
8328 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
8329 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
8331 * config/i386/i386.c (rest_of_insert_endbranch): New.
8332 (pass_data_insert_endbranch): Likewise.
8333 (pass_insert_endbranch): Likewise.
8334 (make_pass_insert_endbranch): Likewise.
8335 (ix86_notrack_prefixed_insn_p): Likewise.
8336 (ix86_target_string): Add -mibt, -mshstk flags.
8337 (ix86_option_override_internal): Add flag_cf_protection
8339 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
8340 (ix86_print_operand): Add 'notrack' prefix output.
8341 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
8342 (ix86_expand_builtin): Expand CET intrinsics.
8343 (x86_output_mi_thunk): Add 'endbranch' instruction.
8344 * config/i386/i386.h (TARGET_IBT): New.
8345 (TARGET_IBT_P): Likewise.
8346 (TARGET_SHSTK): Likewise.
8347 (TARGET_SHSTK_P): Likewise.
8348 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
8349 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
8350 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
8351 (builtin_setjmp_setup): New pattern.
8352 (builtin_longjmp): Likewise.
8353 (rdssp<mode>): Likewise.
8354 (incssp<mode>): Likewise.
8355 (saveprevssp): Likewise.
8356 (rstorssp): Likewise.
8357 (wrss<mode>): Likewise.
8358 (wruss<mode>): Likewise.
8359 (setssbsy): Likewise.
8360 (clrssbsy): Likewise.
8361 (nop_endbr): Likewise.
8362 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
8364 * config/i386/immintrin.h: Include <cetintrin.h>.
8365 * config/i386/linux-common.h
8366 (file_end_indicate_exec_stack_and_cet): New prototype.
8367 (TARGET_ASM_FILE_END): New.
8369 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
8371 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
8372 latencies instead of having separate table; make difference between
8373 integer and float costs.
8374 * i386.h (processor_costs): Remove scalar_stmt_cost,
8375 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
8376 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
8378 * x86-tune-costs.h: Remove entries which has been removed in
8379 procesor_costs from all tables; make cond_taken_branch_cost
8380 and cond_not_taken_branch_cost COST_N_INSNS based.
8382 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
8384 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
8386 2017-10-20 Jakub Jelinek <jakub@redhat.com>
8388 * config/i386/i386.md (isa): Remove fma_avx512f.
8389 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
8390 <avx512>_fmadd_<mode>_mask3<round_name>,
8391 <avx512>_fmsub_<mode>_mask<round_name>,
8392 <avx512>_fmsub_<mode>_mask3<round_name>,
8393 <avx512>_fnmadd_<mode>_mask<round_name>,
8394 <avx512>_fnmadd_<mode>_mask3<round_name>,
8395 <avx512>_fnmsub_<mode>_mask<round_name>,
8396 <avx512>_fnmsub_<mode>_mask3<round_name>,
8397 <avx512>_fmaddsub_<mode>_mask<round_name>,
8398 <avx512>_fmaddsub_<mode>_mask3<round_name>,
8399 <avx512>_fmsubadd_<mode>_mask<round_name>,
8400 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
8401 (*vec_widen_umult_even_v16si<mask_name>,
8402 *vec_widen_smult_even_v16si<mask_name>): Likewise.
8403 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
8405 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8407 * extend.texi: Add 'nocf_check' documentation.
8408 * gimple.texi: Add second parameter to
8409 gimple_build_call_from_tree.
8410 * invoke.texi: Add -fcf-protection documentation.
8411 * rtl.texi: Add REG_CALL_NOTRACK documenation.
8413 2017-10-20 Richard Biener <rguenther@suse.de>
8415 PR tree-optimization/82473
8416 * tree-vect-loop.c (vectorizable_reduction): Properly get at
8417 the largest input type.
8419 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8421 * c-attribs.c (handle_nocf_check_attribute): New function.
8422 (c_common_attribute_table): Add 'nocf_check' handling.
8423 * gimple-parser.c: Add second argument NULL to
8424 gimple_build_call_from_tree.
8425 * attrib.c (comp_type_attributes): Check nocf_check attribute.
8426 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
8428 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
8429 * common.opt: Add fcf-protection flag.
8430 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
8431 * flag-types.h: Add enum cf_protection_level.
8432 * gimple.c (gimple_build_call_from_tree): Add second parameter.
8433 Add 'nocf_check' attribute propagation to gimple call.
8434 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
8435 (gimple_build_call_from_tree): Update prototype.
8436 (gimple_call_nocf_check_p): New function.
8437 (gimple_call_set_nocf_check): Likewise.
8438 * gimplify.c: Add second argument to gimple_build_call_from_tree.
8439 * ipa-icf.c: Add nocf_check attribute in statement hash.
8440 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
8441 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
8442 * toplev.c (process_options): Add flag_cf_protection handling.
8444 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8446 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
8448 2017-10-20 Richard Biener <rguenther@suse.de>
8450 PR tree-optimization/82603
8451 * tree-if-conv.c (predicate_mem_writes): Make sure to only
8452 remove false predicated stores.
8454 2017-10-20 Richard Biener <rguenther@suse.de>
8456 * graphite-isl-ast-to-gimple.c
8457 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
8458 Remove return value and simplify, dump copied stmt after lhs
8460 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
8461 Reduce dump verbosity.
8462 (gsi_insert_earliest): Likewise.
8463 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
8464 * graphite.c (print_global_statistics): Adjust dumping.
8465 (print_graphite_scop_statistics): Likewise.
8466 (print_graphite_statistics): Do not dump loops here.
8467 (graphite_transform_loops): But here.
8469 2017-10-20 Nicolas Roche <roche@adacore.com>
8471 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
8472 * configure: Regenerate.
8474 2017-10-20 Jakub Jelinek <jakub@redhat.com>
8477 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
8478 functions when optimizing replace GIMPLE_RETURN stmts with
8479 calls to __builtin_unreachable ().
8482 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
8483 for -fsanitize=thread link of executables.
8484 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
8485 link of executables.
8488 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
8490 (<shift_insn><mode>3<mask_name>): Change the last of the 3
8491 define_insns for logical vector shifts to use VI248_AVX512BW
8492 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
8493 condition, useless isa and prefix attributes. Change the first
8494 2 of these define_insns to ...
8495 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
8496 define_insn for avx512vl.
8497 (<shift_insn><mode>3): ... and this, new define_insn without
8498 masking for non-avx512vl.
8501 * config/i386/sse.md (*andnot<mode>3,
8502 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
8503 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
8504 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
8505 not applied use empty suffix even for TARGET_AVX512VL.
8506 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
8507 is applied, supply evex,evex or evex,evex,evex instead of just
8510 2017-10-20 Julia Koval <julia.koval@intel.com>
8512 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
8513 (OPTION_MASK_ISA_GFNI_UNSET): New.
8514 (ix86_handle_option): Handle OPT_mgfni.
8515 * config/i386/cpuid.h (bit_GFNI): New.
8516 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
8517 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
8518 * config/i386/i386.c (ix86_target_string): Add -mgfni.
8519 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
8520 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
8521 * config/i386/i386.opt: Add mgfni.
8523 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
8525 * config/msp430/msp430.c (msp430_option_override): Disable
8526 -fdelete-null-pointer-checks.
8527 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
8529 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8531 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
8532 of x87 and SSE instructions.
8534 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8536 * asan.c (create_cond_insert_point): Do not update edge count.
8537 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
8538 (afdo_propagate_circuit): Likewise.
8539 (afdo_calculate_branch_prob): Likewise.
8540 (afdo_annotate_cfg): Likewise.
8541 * basic-block.h (struct edge_def): Remove count.
8542 (edge_def::count): New accessor.
8543 * bb-reorder.c (rotate_loop): Update.
8544 (find_traces_1_round): Update.
8545 (connect_traces): Update.
8546 (sanitize_hot_paths): Update.
8547 * cfg.c (unchecked_make_edge): Update.
8548 (make_single_succ_edge): Update.
8549 (check_bb_profile): Update.
8550 (dump_edge_info): Update.
8551 (update_bb_profile_for_threading): Update.
8552 (scale_bbs_frequencies_int): Update.
8553 (scale_bbs_frequencies_gcov_type): Update.
8554 (scale_bbs_frequencies_profile_count): Update.
8555 (scale_bbs_frequencies): Update.
8556 * cfganal.c (connect_infinite_loops_to_exit): Update.
8557 * cfgbuild.c (compute_outgoing_frequencies): Update.
8558 (find_many_sub_basic_blocks): Update.
8559 * cfgcleanup.c (try_forward_edges): Update.
8560 (try_crossjump_to_edge): Update
8561 * cfgexpand.c (expand_gimple_cond): Update
8562 (expand_gimple_tailcall): Update
8563 (construct_exit_block): Update
8564 * cfghooks.c (verify_flow_info): Update
8565 (redirect_edge_succ_nodup): Update
8566 (split_edge): Update
8567 (make_forwarder_block): Update
8568 (duplicate_block): Update
8569 (account_profile_record): Update
8570 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
8571 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
8572 * cfgloopmanip.c (scale_loop_profile): Update.
8574 (lv_adjust_loop_entry_edge): Update.
8575 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
8576 (force_nonfallthru_and_redirect): Update.
8577 (purge_dead_edges): Update.
8578 (rtl_flow_call_edges_add): Update.
8579 * cgraphunit.c (init_lowered_empty_function): Update.
8580 (cgraph_node::expand_thunk): Update.
8581 * gimple-pretty-print.c (dump_probability): Update.
8582 (dump_edge_probability): Update.
8583 * gimple-ssa-isolate-paths.c (isolate_path): Update.
8584 * haifa-sched.c (sched_create_recovery_edges): Update.
8585 * hsa-gen.c (convert_switch_statements): Update.
8586 * ifcvt.c (dead_or_predicable): Update.
8587 * ipa-inline-transform.c (inline_transform): Update.
8588 * ipa-split.c (split_function): Update.
8589 * ipa-utils.c (ipa_merge_profiles): Update.
8590 * loop-doloop.c (add_test): Update.
8591 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
8592 * lto-streamer-in.c (input_cfg): Update.
8593 (input_function): Update.
8594 * lto-streamer-out.c (output_cfg): Update.
8595 * modulo-sched.c (sms_schedule): Update.
8596 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
8597 * predict.c (maybe_hot_edge_p): Update.
8598 (unlikely_executed_edge_p): Update.
8599 (probably_never_executed_edge_p): Update.
8600 (dump_prediction): Update.
8601 (drop_profile): Update.
8602 (propagate_unlikely_bbs_forward): Update.
8603 (determine_unlikely_bbs): Update.
8604 (force_edge_cold): Update.
8605 * profile.c (compute_branch_probabilities): Update.
8606 * reg-stack.c (better_edge): Update.
8607 * shrink-wrap.c (handle_simple_exit): Update.
8608 * tracer.c (better_p): Update.
8609 * trans-mem.c (expand_transaction): Update.
8610 (split_bb_make_tm_edge): Update.
8611 * tree-call-cdce.c: Update.
8612 * tree-cfg.c (gimple_find_sub_bbs): Update.
8613 (gimple_split_edge): Update.
8614 (gimple_duplicate_sese_region): Update.
8615 (gimple_duplicate_sese_tail): Update.
8616 (gimple_flow_call_edges_add): Update.
8617 (insert_cond_bb): Update.
8618 (execute_fixup_cfg): Update.
8619 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
8620 * tree-complex.c (expand_complex_div_wide): Update.
8621 * tree-eh.c (lower_resx): Update.
8622 (unsplit_eh): Update.
8623 (cleanup_empty_eh_move_lp): Update.
8624 * tree-inline.c (copy_edges_for_bb): Update.
8625 (freqs_to_counts): Update.
8626 (copy_cfg_body): Update.
8627 * tree-ssa-dce.c (remove_dead_stmt): Update.
8628 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
8629 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
8630 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
8631 (unloop_loops): Update.
8632 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8633 * tree-ssa-loop-split.c (connect_loops): Update.
8634 (split_loop): Update.
8635 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
8636 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
8637 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
8638 * tree-ssa-reassoc.c (branch_fixup): Update.
8639 * tree-ssa-tail-merge.c (replace_block_by): Update.
8640 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
8641 (compute_path_counts): Update.
8642 (update_profile): Update.
8643 (recompute_probabilities): Update.
8644 (update_joiner_offpath_counts): Update.
8645 (estimated_freqs_path): Update.
8646 (freqs_to_counts_path): Update.
8647 (clear_counts_path): Update.
8648 (ssa_fix_duplicate_block_edges): Update.
8649 (duplicate_thread_path): Update.
8650 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
8651 (case_bit_test_cmp): Update.
8652 (collect_switch_conv_info): Update.
8653 (gen_inbound_check): Update.
8654 (do_jump_if_equal): Update.
8655 (emit_cmp_and_jump_insns): Update.
8656 * tree-tailcall.c (decrease_profile): Update.
8657 (eliminate_tail_call): Update.
8658 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
8659 (vect_do_peeling): Update.
8660 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8661 * ubsan.c (ubsan_expand_null_ifn): Update.
8662 (ubsan_expand_ptr_ifn): Update.
8663 * value-prof.c (gimple_divmod_fixed_value): Update.
8664 (gimple_mod_pow2): Update.
8665 (gimple_mod_subtract): Update.
8666 (gimple_ic): Update.
8667 (gimple_stringop_fixed_value): Update.
8669 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8672 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
8674 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
8676 PR rtl-optimization/82395
8677 * ira-color.c (allocno_priority_compare_func): Fix comparison step
8678 based on non_spilled_static_chain_regno_p.
8680 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8682 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
8683 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
8684 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
8686 2017-10-19 Martin Sebor <msebor@redhat.com>
8688 PR tree-optimization/82596
8689 * tree.c (array_at_struct_end_p): Handle STRING_CST.
8691 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8693 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
8694 (get_mem_refs_of_builtin_call): Likewise.
8695 * builtins.c (expand_builtin_apply): Adjust call to
8696 allocate_dynamic_stack_space.
8697 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
8698 the third argument to allocate_dynamic_stack_space, otherwise -1.
8699 (expand_builtin): Deal with all alloca variants.
8700 (is_inexpensive_builtin): Likewise.
8701 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
8702 * calls.c (special_function_p): Deal with all alloca variants.
8703 (initialize_argument_information): Adjust call to
8704 allocate_dynamic_stack_space.
8705 (expand_call): Likewise.
8706 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
8707 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
8708 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
8709 use it for the stack usage computation.
8710 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
8711 * function.c (gimplify_parameters): Call build_alloca_call_expr.
8712 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
8713 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
8714 (in_loop_p): Remove first argument and useless check.
8715 (pass_walloca::execute): Remove useless test and adjust call to above.
8716 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
8717 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
8718 (gimplify_call_expr): Deal with all alloca variants.
8719 * hsa-gen.c (gen_hsa_alloca): Likewise.
8720 (gen_hsa_insns_for_call): Likewise.
8721 * ipa-pure-const.c (special_builtin_state): Likewise.
8722 * tree-chkp.c (chkp_build_returned_bound): Likewise.
8723 * tree-object-size.c (alloc_object_size): Likewise.
8724 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
8725 (call_may_clobber_ref_p_1): Likewise.
8726 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
8727 (ccp_fold_stmt): Likewise.
8728 (optimize_stack_restore): Likewise.
8729 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
8730 (mark_all_reaching_defs_necessary_1): Likewise.
8731 (propagate_necessity): Likewise.
8732 (eliminate_unnecessary_stmts): Likewise.
8733 * tree.c (build_common_builtin_nodes): Build
8734 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
8735 (build_alloca_call_expr): New function.
8736 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
8737 (CASE_BUILT_IN_ALLOCA): Likewise.
8738 (build_alloca_call_expr): Declare.
8739 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
8741 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8744 * dwarf2out.c (new_die_raw): New static inline function.
8745 (new_die): Use it to create the DIE.
8746 (add_AT_external_die_ref): Likewise.
8747 (clone_die): Likewise.
8748 (clone_as_declaration): Likewise.
8749 (dwarf2out_vms_debug_main_pointer): Likewise.
8750 (base_type_die): Likewise. Remove early return for corner cases.
8751 Do not call add_pubtype on the DIE here.
8752 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
8753 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
8754 typedefs for base types with DW_AT_endianity. Make sure a DIE with
8755 native order exists for base types, attach the DIE manually and call
8756 add_pubtype on it. Do not equate a reverse order DIE to the type.
8758 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
8760 * config/arm/arm.c (align_ok_ldrd_strd): New function.
8761 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
8763 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
8765 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8767 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
8768 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
8769 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
8770 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
8771 * opts.c (sanitizer_opts): Add builtin.
8772 * ubsan.c (instrument_builtin): New function.
8773 (pass_ubsan::execute): Call it.
8774 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
8775 * doc/invoke.texi: Document -fsanitize=builtin.
8777 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
8778 builtins, store max (log2 (align), 0) into uchar field instead of
8779 align into uptr field.
8780 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
8781 store uchar 0 field instead of uptr 0 field.
8782 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
8783 instead of passing one address of struct with 2 locations pass
8784 two addresses of structs with 1 location each.
8785 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
8786 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
8787 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
8788 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
8789 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
8790 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
8791 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
8792 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
8794 2017-10-19 Martin Liska <mliska@suse.cz>
8797 * file-find.c (remove_prefix): Remove.
8798 * file-find.h (remove_prefix): Likewise.
8799 * gcc-ar.c: Remove smartness of lookup.
8801 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
8803 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
8804 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
8805 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
8807 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8810 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
8811 (setcc + and to xor + setcc): New peephole2.
8813 2017-10-19 Tom de Vries <tom@codesourcery.com>
8815 * doc/sourcebuild.texi (Test Directives, Variants of
8816 dg-require-support): Add dg-require-stack-size.
8818 2017-10-19 Martin Liska <mliska@suse.cz>
8821 * gimplify.c (gimplify_decl_expr): Do not instrument variables
8822 that have a large alignment.
8823 (gimplify_target_expr): Likewise.
8825 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
8827 PR rtl-optimization/82602
8828 * ira.c (rtx_moveable_p): Return false for volatile asm.
8830 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
8833 * config/i386/i386-modes.def (CCGZ): New CC mode.
8834 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
8835 * config/i386/predicates.md (ix86_comparison_operator):
8837 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
8838 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
8839 with double-word subtraction.
8840 (put_condition_code): Handle CCGZmode.
8842 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
8844 * wide-int.cc (debug (const wide_int &)): New.
8845 (debug (const wide_int *)): New.
8846 (debug (const widest_int &)): New.
8847 (debug (const widest_int *)): New.
8849 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
8852 * lra-constraints.c (curr_insn_transform): Use non-input operand
8853 instead of output one for matched reload.
8855 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8857 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
8858 (tree-ssa-loop-ivopts.h): New header file.
8859 (struct builtin_info): New fields.
8860 (classify_builtin_1): Compute and record base and offset parts for
8861 memset builtin partition by calling strip_offset.
8862 (offset_cmp, fuse_memset_builtins): New functions.
8863 (finalize_partitions): Fuse adjacent memset partitions by calling
8865 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
8866 Expose the interface.
8867 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
8869 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8871 PR tree-optimization/82574
8872 * tree-loop-distribution.c (find_single_drs): New parameter. Check
8873 that data reference must be executed exactly once per iteration
8874 against the outermost loop in nest.
8875 (classify_partition): Update call to above function.
8877 2017-10-18 Richard Biener <rguenther@suse.de>
8879 PR tree-optimization/82591
8880 * graphite.c (graphite_transform_loops): Move code gen message
8882 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8883 Here. Handle scop_to_isl_ast failing.
8884 (scop_to_isl_ast): Limit the number of ISL operations.
8886 2017-10-18 Richard Biener <rguenther@suse.de>
8888 * graphite-isl-ast-to-gimple.c
8889 (translate_isl_ast_to_gimple::set_rename): Simplify.
8890 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
8891 (graphite_copy_stmts_from_block): ... here.
8892 (copy_bb_and_scalar_dependences): Simplify.
8893 (add_parameters_to_ivs_params): Canonicalize.
8894 (generate_entry_out_of_ssa_copies): Simplify.
8895 * graphite-sese-to-poly.c (extract_affine_name): Simplify
8896 by passing in ISL dimension.
8897 (parameter_index_in_region_1): Rename to ...
8898 (parameter_index_in_region): ... this.
8899 (extract_affine): Adjust assert, pass down parameter index.
8900 (add_param_constraints): Use range-info when available.
8901 (build_scop_context): Adjust.
8902 * sese.c (new_sese_info): Adjust.
8903 (free_sese_info): Likewise.
8904 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
8905 Remove unused typedefs.
8906 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
8908 2017-10-18 Martin Liska <mliska@suse.cz>
8910 * combine.c (simplify_compare_const): Add gcc_fallthrough.
8912 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8914 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
8915 (s390_sched_init): Do not reset s390_sched_state if we entered the
8916 current basic block via a fallthru edge and all others are unlikely.
8918 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8920 * config/s390/s390.c (NUM_SIDES): New variable.
8921 (LONGRUNNING_THRESHOLD): New variable.
8922 (LATENCY_FACTOR): New variable.
8923 (s390_sched_score): Decrease score for long-running instructions on
8925 (s390_sched_variable_issue): Perform bookkeeping for long-running
8928 2017-10-18 Richard Biener <rguenther@suse.de>
8930 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
8931 Simplify with removal of the parameter rename map.
8932 (set_rename): Likewise.
8933 (should_copy_to_new_region): Likewise.
8934 (graphite_copy_stmts_from_block): Likewise.
8935 (copy_bb_and_scalar_dependences): Remove initialization of
8936 unused copied_bb_map.
8938 (copy_internal_parameters): Likewise.
8939 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
8940 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
8941 Use INTEGRAL_TYPE_P.
8942 (parameter_index_in_region_1): Rename to ...
8943 (assign_parameter_index_in_region): ... this. Assert we have
8944 a parameter we handle.
8945 (scan_tree_for_params): Adjust.
8946 * sese.h (parameter_rename_map_t): Remove.
8947 (struct sese_info_t): Remove unused parameter_rename_map and
8948 copied_bb_map members.
8949 * sese.c (new_sese_info): Adjust.
8950 (free_sese_info): Likewise.
8952 2017-10-18 Martin Liska <mliska@suse.cz>
8955 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
8956 on an abnormal edge.
8958 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
8960 * doc/invoke.texi (ffunction-sections and fdata-sections):
8963 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8965 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
8966 the use statement can throw internally.
8968 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8970 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
8971 any RTX present on the RHS of a SET.
8972 * compare-elim.c (try_eliminate_compare): Restore comment.
8974 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8976 * langhooks.h (struct lang_hooks): Document that tree_size langhook
8977 may be also called on tcc_type nodes.
8978 * langhooks.c (lhd_tree_size): Likewise.
8980 2017-10-17 David Malcolm <dmalcolm@redhat.com>
8982 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
8983 format_warning_at_substring.
8984 (maybe_warn): Convert source_range * param to a location_t. Pass
8985 UNKNOWN_LOCATION rather than NULL to fmtwarn.
8986 (format_directive): Remove code to extract source_ranges and
8987 source_range * in favor of just a location_t.
8988 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
8990 * substring-locations.c (format_warning_va): Convert
8991 source_range * param to a location_t.
8992 (format_warning_at_substring): Likewise.
8993 * substring-locations.h (format_warning_va): Likewise.
8994 (format_warning_at_substring): Likewise.
8996 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
8998 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
9000 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
9001 and scatter/gather ops.
9003 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
9004 vec_gather_load and vec_scatter_store.
9005 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
9006 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
9007 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
9008 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
9009 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
9010 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
9012 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
9014 * reg-stack.c (compare_for_stack_reg): Add bool argument.
9015 Detect FTST instruction and handle its register pops. Only pop
9016 second operand if can_pop_second_op is true.
9017 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
9018 set can_pop_second_op to false in the compare_for_stack_reg call.
9020 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
9021 output_fp_compare for stack register operands.
9022 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
9023 instructions here. Do not emit stack register pops here. Assert
9024 that FCOMPP pops next to top stack register. Rewrite function.
9026 2017-10-17 Nathan Sidwell <nathan@acm.org>
9029 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
9030 use DECL_ASSEMBLER_NAME_RAW.
9033 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
9036 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
9037 Wilco Dijkstra <wilco.dijkstra@arm.com>
9039 * builtins.c (expand_builtin_update_setjmp_buf): Add a
9040 converstion to Pmode from the buf_addr.
9042 2017-10-17 Richard Biener <rguenther@suse.de>
9044 * graphite-dependences.c (scop_get_reads_and_writes): Change
9045 output parameters to references.
9047 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
9049 PR 71026/tree-optimization
9050 * fold-const.c (distribute_real_division): Removed.
9051 (fold_binary_loc): Remove calls to distribute_real_divison.
9053 2017-10-17 Richard Biener <rguenther@suse.de>
9055 * graphite-scop-detection.c
9056 (scop_detection::stmt_has_simple_data_refs_p): Always use
9057 the full nest as region.
9058 (try_generate_gimple_bb): Likewise.
9059 * sese.c (scalar_evolution_in_region): Simplify now that
9060 SCEV can handle instantiation in regions.
9061 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
9062 in the non-loop part of a function if requested.
9064 2017-10-17 Richard Biener <rguenther@suse.de>
9066 PR tree-optimization/82563
9067 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
9069 (graphite_regenerate_ast_isl): Call it.
9070 * graphite-scop-detection.c (build_scops): Remove entry edge split.
9072 2017-10-17 Jakub Jelinek <jakub@redhat.com>
9074 PR tree-optimization/82549
9075 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
9076 Formatting fixes. Instead of calling make_bit_field_ref with negative
9079 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
9081 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
9082 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
9083 _mm_maskz_reduce_ss): New.
9084 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
9085 __builtin_ia32_reducess_mask): Ditto..
9086 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
9087 * config/i386/sse.md (reduces<mode>): Renamed to ...
9088 (reduces<mode><mask_scalar_name>): ... this.
9089 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
9091 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
9092 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
9094 2017-10-16 David Malcolm <dmalcolm@redhat.com>
9096 * Makefile.in (OBJS): Add unique-ptr-tests.o.
9097 * selftest-run-tests.c (selftest::run_tests): Call
9098 selftest::unique_ptr_tests_cc_tests.
9099 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
9100 * unique-ptr-tests.cc: New file.
9102 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
9105 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9107 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9108 (make_hard_regno_born, make_hard_regno_dead): Update
9109 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
9111 2017-10-16 Jeff Law <law@redhat.com>
9113 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
9115 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
9117 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
9119 2017-10-16 Olivier Hainque <hainque@adacore.com>
9121 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
9122 with_cpu if we were configured for an e500v2 target cpu name.
9124 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
9126 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
9127 * doc/invoke.texi: Document +nodsp as a valid extension for
9130 2017-10-16 Martin Liska <mliska@suse.cz>
9132 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
9133 (test_set_range): Likewise.
9134 (test_range_functions): Rename to ...
9135 (test_bit_in_range): ... this.
9136 (sbitmap_c_tests): Add new test.
9138 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9140 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
9142 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
9143 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
9145 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9147 * config/aarch64/aarch64-builtins.c
9148 (aarch64_types_quadopu_lane_qualifiers): New.
9149 (TYPES_QUADOPU_LANE): New.
9150 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
9151 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
9152 (aarch64_<sur>dot_laneq<vsi2qi>): New.
9153 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
9154 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
9155 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
9156 (Vdottype, DOTPROD): New.
9157 (sur): Add SDOT and UDOT.
9159 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9161 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
9162 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
9163 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
9165 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
9166 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
9167 Enable TARGET_DOTPROD.
9168 (cortex-a75.cortex-a55): Likewise.
9169 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
9171 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9173 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
9174 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
9176 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
9178 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
9179 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
9180 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
9181 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
9182 * config/arm/types.md (neon_dot, neon_dot_q): New.
9183 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
9185 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9187 * config/arm/arm.h (TARGET_DOTPROD): New.
9188 * config/arm/arm.c (arm_arch_dotprod): New.
9189 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
9190 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
9191 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
9192 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
9193 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
9194 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
9195 * doc/invoke.texi (armv8.2-a): Document dotprod
9197 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
9199 * i386.c (ix86_vec_cost): New function.
9200 (ix86_rtx_costs): Handle vector operations better.
9201 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
9202 * x86-tune-costs.h: Add new costs to all tables.
9204 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
9206 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
9208 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
9209 divsd, sqrtss and sqrtsd
9210 * x86-tune-costs.h: Add new entries to all costs.
9211 (znver1_cost): Fix to match real instruction latencies.
9213 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9214 Michael Collison <michael.collison@arm.com>
9216 * compare-elim.c: Include emit-rtl.h.
9217 (can_merge_compare_into_arith): New function.
9218 (try_validate_parallel): Likewise.
9219 (try_merge_compare): Likewise.
9220 (try_eliminate_compare): Call the above when no previous clobber
9222 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
9225 2017-10-14 Jakub Jelinek <jakub@redhat.com>
9229 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
9230 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
9234 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
9235 to be any operand_equal_p operands. For & (B - 1) require
9236 B to be power of 2. Recognize
9237 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
9239 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
9242 * optabs.c (expand_memory_blockage): Fix call of
9243 targetm.have_memory_blockage.
9245 2017-10-14 Jakub Jelinek <jakub@redhat.com>
9248 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
9249 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
9250 objects to extra_objs instead of overwriting it.
9252 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
9254 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
9255 Use any_fp_register_operand as operand[3] predicate. Simplify
9256 equality test for operands[2] and operands[4] memory location.
9257 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
9258 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
9259 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
9260 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
9261 any_fp_register_operand as operand[1] predicate. Simplify
9262 equality test for operands[0] and operands[3] memory location.
9263 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
9264 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
9265 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
9267 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
9269 * target-insns.def: Add memory_blockage.
9270 * optabs.c (expand_memory_blockage): New function.
9271 (expand_asm_memory_barrier): Rename ...
9272 (expand_asm_memory_blockage): ... to this.
9273 (expand_mem_thread_fence): Call expand_memory_blockage
9274 instead of expand_asm_memory_barrier.
9275 (expand_mem_singnal_fence): Ditto.
9276 (expand_atomic_load): Ditto.
9277 (expand_atomic_store): Ditto.
9278 * doc/md.texi (Standard Pattern Names For Generation):
9279 Document memory_blockage instruction pattern.
9281 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
9283 * config/rl78/rl78.c (rl78_emit_libcall): New function.
9284 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
9285 * config/rl78/rl78.md: New define_expand "adddi3".
9287 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
9289 * cfghooks.c (verify_flow_info): Disable check that all probabilities
9292 2017-10-13 Jeff Law <law@redhat.com>
9294 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
9296 2017-10-13 Jakub Jelinek <jakub@redhat.com>
9299 * internal-fn.c (expand_mul_overflow): If both operands have
9300 the same highpart of -1 or 0 and the topmost bit of lowpart
9301 is different, overflow is if res <= 0 rather than res < 0.
9303 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
9305 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
9306 TARGET_P9_VECTOR code for unaligned_load case.
9308 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
9310 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
9312 2017-10-13 Nathan Sidwell <nathan@acm.org>
9314 * tree-core.h (tree_contains_struct): Make bool.
9315 * tree.c (tree_contains_struct): Likewise.
9316 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
9317 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
9318 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
9319 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
9320 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
9322 2017-10-13 Richard Biener <rguenther@suse.de>
9324 * graphite-isl-ast-to-gimple.c
9325 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
9326 parameters and dominance check.
9327 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
9328 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
9329 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
9330 Do not update SSA form here or do intermediate IL verification.
9331 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
9332 (graphite_initialize): Remove check on the number of loops in
9333 the function and inline into graphite_transform_loops.
9334 (graphite_finalize): Inline into graphite_transform_loops.
9335 (graphite_transform_loops): Perform SSA update and IL verification
9337 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
9339 2017-10-13 Richard Biener <rguenther@suse.de>
9341 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
9342 graphite_expression_type_precision): Avoid global constructor
9344 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
9345 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
9346 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
9347 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
9348 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
9349 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
9351 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
9354 * config/i386/i386.h (ix86_red_zone_size): New.
9355 * config/i386/i386.md (push peephole2s): Replace
9356 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
9358 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9359 Alan Hayward <alan.hayward@arm.com>
9360 David Sherwood <david.sherwood@arm.com>
9362 * combine.c (can_change_dest_mode): Reject changes in
9363 REGMODE_NATURAL_SIZE.
9365 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9366 Alan Hayward <alan.hayward@arm.com>
9367 David Sherwood <david.sherwood@arm.com>
9369 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
9370 (expand_debug_source_expr): Likewise.
9371 * combine.c (combine_simplify_rtx): Likewise.
9372 * cse.c (fold_rtx): Likewise.
9373 * fwprop.c (canonicalize_address): Likewise.
9374 * targhooks.c (default_shift_truncation_mask): Likewise.
9376 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9377 Alan Hayward <alan.hayward@arm.com>
9378 David Sherwood <david.sherwood@arm.com>
9380 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
9381 (widened_mode): Likewise.
9382 (expand_unop): Likewise.
9383 * ree.c (transform_ifelse): Likewise.
9384 (merge_def_and_ext): Likewise.
9385 (combine_reaching_defs): Likewise.
9386 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9388 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9389 Alan Hayward <alan.hayward@arm.com>
9390 David Sherwood <david.sherwood@arm.com>
9392 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
9393 * combine.c (gen_lowpart_for_combine): Likewise.
9394 * dwarf2out.c (rtl_for_decl_location): Likewise.
9395 * final.c (alter_subreg): Likewise.
9396 * rtlhooks.c (gen_lowpart_general): Likewise.
9397 (gen_lowpart_if_possible): Likewise.
9399 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9400 Alan Hayward <alan.hayward@arm.com>
9401 David Sherwood <david.sherwood@arm.com>
9403 * calls.c (expand_call): Use subreg_lowpart_offset.
9404 * cse.c (cse_insn): Likewise.
9405 * regcprop.c (copy_value): Likewise.
9406 (copyprop_hardreg_forward_1): Likewise.
9408 2017-10-13 Jakub Jelinek <jakub@redhat.com>
9411 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
9412 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
9413 =Q constraints to +Q and into insn condition add check
9414 that operands[0] and operands[1] are equal.
9415 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
9416 =Q constraints to +Q and into insn condition add check
9417 that operands[0] is equal to either operands[1] or operands[2].
9420 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
9421 instead of handling MINUS_EXPR twice (once for each argument),
9422 canonicalize operand order and handle just once, use rtype where
9423 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
9426 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
9427 any values of __C while still being pattern recognizable as a simple
9430 2017-10-13 Richard Biener <rguenther@suse.de>
9432 PR tree-optimization/82451
9434 2017-10-02 Richard Biener <rguenther@suse.de>
9436 PR tree-optimization/82355
9437 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
9438 a mapping for the enclosing loop but avoid generating one for
9440 (copy_bb_and_scalar_dependences): Remove premature codegen
9441 error on PHIs in blocks duplicated into multiple places.
9442 * graphite-scop-detection.c
9443 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
9444 in the region use it as loop and nest to analyze the DR in.
9445 (try_generate_gimple_bb): Likewise.
9446 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
9447 (add_loop_constraints): For blocks in a loop not in the region
9448 create a dimension with a single iteration.
9449 * sese.h (gbb_loop_at_index): Remove assert.
9451 * cfgloop.c (loop_preheader_edge): For the loop tree root
9452 return the single successor of the entry block.
9453 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
9454 Reset the SCEV hashtable and niters.
9455 * graphite-scop-detection.c
9456 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
9457 assert that we only have POLYNOMIAL_CHREC that vary in loops
9458 contained in the region.
9459 (scop_detection::graphite_can_represent_expr): Adjust.
9460 (scop_detection::stmt_has_simple_data_refs_p): For loops
9461 not in the region set loop to NULL. The nest is now the
9462 entry edge to the region.
9463 (try_generate_gimple_bb): Likewise.
9464 * sese.c (scalar_evolution_in_region): Adjust for
9465 instantiate_scev change.
9466 * tree-data-ref.h (graphite_find_data_references_in_stmt):
9467 Make nest parameter the edge into the region.
9468 (create_data_ref): Likewise.
9469 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
9470 entry edge into a region and adjust instantiate_scev calls.
9471 (create_data_ref): Likewise.
9472 (graphite_find_data_references_in_stmt): Likewise.
9473 (find_data_references_in_stmt): Pass the loop preheader edge
9474 from the nest argument.
9475 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
9476 parameter the edge into the region.
9477 (instantiate_parameters): Use the loop preheader edge as entry.
9478 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
9480 (get_instantiated_value_entry): Make instantiate_below parameter
9481 the edge into the region.
9482 (instantiate_scev_name): Likewise. Adjust dominance checks,
9483 when we cannot use loop-based instantiation instantiate by
9484 walking use-def chains.
9485 (instantiate_scev_poly): Adjust.
9486 (instantiate_scev_binary): Likewise.
9487 (instantiate_scev_convert): Likewise.
9488 (instantiate_scev_not): Likewise.
9489 (instantiate_array_ref): Remove.
9490 (instantiate_scev_3): Likewise.
9491 (instantiate_scev_2): Likewise.
9492 (instantiate_scev_1): Likewise.
9493 (instantiate_scev_r): Do not blindly handle N-operand trees.
9494 Do not instantiate array-refs. Handle all constants and invariants.
9495 (instantiate_scev): Make instantiate_below parameter
9496 the edge into the region.
9497 (resolve_mixers): Use the loop preheader edge for the region
9498 parameter to instantiate_scev_r.
9499 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
9501 2017-10-13 Richard Biener <rguenther@suse.de>
9503 PR tree-optimization/82525
9504 * graphite-isl-ast-to-gimple.c
9505 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
9507 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
9508 Fail code generation when we cannot represent the isl integer.
9509 (binary_op_to_tree): Elide modulo operations that are no-ops
9510 in the type we code generate. Remove now superfluous code
9513 2017-10-13 Richard Biener <rguenther@suse.de>
9515 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
9516 (scop_detection::harmful_loop_in_region): Remove premature
9517 IV type restriction.
9518 (scop_detection::graphite_can_represent_scev): We can handle
9519 pointer IVs just fine.
9521 2017-10-13 Alan Modra <amodra@gmail.com>
9523 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
9524 "Clobbers and Scratch Registers". Add paragraph on
9525 alternative to clobbers for scratch registers and OpenBLAS
9528 2017-10-13 Alan Modra <amodra@gmail.com>
9530 * doc/extend.texi (Clobbers): Correct vax example. Delete old
9531 example of a memory input for a string of known length. Move
9532 commentary out of table. Add a number of new examples
9533 covering array memory inputs.
9535 2017-10-12 Martin Liska <mliska@suse.cz>
9537 PR tree-optimization/82493
9538 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
9539 (test_range_functions): New function.
9540 (sbitmap_c_tests): Likewise.
9541 * selftest-run-tests.c (selftest::run_tests): Run new tests.
9542 * selftest.h (sbitmap_c_tests): New function.
9544 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
9546 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
9548 * config/rs6000/amo.h: Fix spacing issue.
9550 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9553 * config/i386/i386.md (*ashl<mode>3_mask_1,
9554 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
9555 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
9558 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
9560 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
9561 (profile_probability): Set max_probability
9562 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
9564 * profile-count.c (profile_probability::differs_from_p): Do not
9565 rely on max_probaiblity == 10000
9567 2017-10-12 Jeff Law <law@redhat.com>
9569 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
9572 2017-10-12 Martin Sebor <msebor@redhat.com>
9576 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
9577 (handle_alias_pairs): Call it.
9578 * common.opt (-Wattribute-alias): New option.
9579 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
9580 * doc/invoke.texi (-Wattribute-alias): Document.
9582 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
9585 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9587 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9589 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9590 (make_hard_regno_born, make_hard_regno_dead): Update
9591 bb_killed_pseudos and bb_gen_pseudos.
9593 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
9595 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
9597 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
9599 * config/alpha/alpha.c (alpha_split_conditional_move):
9600 Use std::swap instead of manually swapping.
9601 (alpha_stdarg_optimize_hook): Ditto.
9602 (alpha_canonicalize_comparison): Ditto.
9604 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9606 * tree-loop-distribution.c (struct builtin_info): New struct.
9607 (struct partition): Refactor fields into struct builtin_info.
9608 (partition_free): Free struct builtin_info.
9609 (build_size_arg_loc, build_addr_arg_loc): Delete.
9610 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
9611 information from struct builtin_info.
9612 (find_single_drs): New function refactored from classify_partition.
9613 Also moved builtin validity checks to this function.
9614 (compute_access_range, alloc_builtin): New functions.
9615 (classify_builtin_st, classify_builtin_ldst): New functions.
9616 (classify_partition): Refactor code into functions find_single_drs,
9617 classify_builtin_st and classify_builtin_ldst.
9618 (distribute_loop): Don't do runtime alias check when distributing
9620 (find_seed_stmts_for_distribution): New function.
9621 (pass_loop_distribution::execute): Refactor code finding seed
9622 stmts into above function. Support distribution for the innermost
9623 two-level loop nest. Adjust dump information.
9625 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9627 * tree-loop-distribution.c: Adjust the general comment.
9628 (NUM_PARTITION_THRESHOLD): New macro.
9629 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
9630 (classify_partition): Skip builtin pattern of loop nest's inner loop.
9631 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
9632 in call to build_partition_graph.
9633 (finalize_partitions): New parameter. Make loop distribution more
9634 conservative by fusing more partitions.
9635 (distribute_loop): Don't do runtime alias check in case of loop nest
9637 (find_seed_stmts_for_distribution): New function.
9638 (prepare_perfect_loop_nest): New function.
9639 (pass_loop_distribution::execute): Refactor code finding seed stmts
9640 and loop nest into above functions. Support loop nest distribution.
9641 Adjust dump information accordingly.
9643 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9645 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
9646 and set PTYPE_SEQUENTIAL for merged partition.
9648 2017-10-12 Richard Biener <rguenther@suse.de>
9650 PR tree-optimization/69728
9652 2017-09-19 Richard Biener <rguenther@suse.de>
9654 PR tree-optimization/69728
9655 * graphite-sese-to-poly.c (schedule_error): New global.
9656 (add_loop_schedule): Handle empty domain by failing the
9658 (build_original_schedule): Handle schedule_error.
9660 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
9661 domain by returning an unchanged schedule.
9663 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9665 * genrecog.c (validate_pattern): For VEC_SELECT verify that
9666 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
9668 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
9670 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
9673 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9676 * expr.c (store_field): Don't optimize away bitsize == 0 store
9677 from CALL_EXPR with addressable return type.
9679 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9681 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
9682 * config/rs6000/rs6000.md (sel): Delete mode attribute.
9683 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
9684 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
9685 TARGET_ISEL instead of TARGET_ISEL<sel>.
9687 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
9689 * config/rs6000/rs6000.c
9690 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
9692 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9694 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
9696 (reg_or_cint_operand): Fix comment.
9697 (reg_or_zero_operand): New predicate.
9698 * config/rs6000/rs6000-protos.h (output_isel): Delete.
9699 * config/rs6000/rs6000.c (output_isel): Delete.
9700 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
9701 instead of reg_or_cint_operand. Output instruction directly (not via
9703 (isel_unsigned_<mode>): Ditto.
9704 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
9705 gpc_reg_operand. Add an instruction alternative for this. Output
9706 instruction directly.
9707 (*isel_reversed_unsigned_<mode>): Ditto.
9709 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9711 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
9712 (TARGET_CANONICALIZE_COMPARISON): Define.
9714 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
9717 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
9718 Check whether the dest is REG before adding REG_EQUIV note.
9720 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9723 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9725 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9726 (make_hard_regno_born, make_hard_regno_dead): Update
9727 bb_killed_pseudos and bb_gen_pseudos.
9729 2017-10-11 Nathan Sidwell <nathan@acm.org>
9731 * incpath.h (enum incpath_kind): Name enum, prefix values.
9732 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
9733 * incpath.c (heads, tails): Use INC_MAX.
9734 (add_env_var_paths, add_standard_paths): Use incpath_kind.
9735 (merge_include_chains, split_quote_chain,
9736 register_include_chains): Update incpath_kind names.
9737 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
9738 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
9740 (add_framework_path, darwin_register_objc_includes): Likewise.
9741 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
9743 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9745 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
9746 Do not use float_operator operator predicate.
9747 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
9748 * config/i386/predicates.md (float_operator): Remove predicate.
9750 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9752 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
9753 (*jccxf_i387): Ditto.
9754 (*jcc<mode>_i387): Ditto.
9755 (*jccu<mode>_i387): Ditto.
9756 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
9757 (*jcc_*_i387 splitters): Remove.
9758 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
9759 * config/i386/i386.c (ix86_split_fp_branch): Remove.
9760 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
9763 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9765 * profile-count.h (slow_safe_scale_64bit): New function.
9766 (safe_scale_64bit): New inline.
9767 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
9768 * profile-count.c: Include wide-int.h
9769 (slow_safe_scale_64bit): New.
9771 2017-10-11 Nathan Sidwell <nathan@acm.org>
9773 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
9774 HAS_DECL_ASSEMBLER_NAME_P.
9775 * gimple-expr.c (gimple_decl_printable_name: Check
9776 HAS_DECL_ASSEMBLER_NAME_P too.
9777 * ipa-utils.h (type_in_anonymous_namespace_p): Check
9778 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
9779 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
9780 * passes.c (rest_of_decl_compilation): Check
9781 HAS_DECL_ASSEMBLER_NAME_P too.
9782 * recog.c (verify_changes): Likewise.
9783 * tree-pretty-print.c (dump_decl_name): Likewise.
9784 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
9786 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
9787 (SET_DECL_ASSEMBLER_NAME): Use it.
9788 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
9789 (COPY_DECL_ASSEMBLER_NAME): Likewise.
9790 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
9792 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9794 * config.gcc (i386, x86_64): Add extra objects.
9795 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
9796 (ix86_min_insn_size): Declare.
9797 (ix86_issue_rate): Declare.
9798 (ix86_adjust_cost): Declare.
9799 (ia32_multipass_dfa_lookahead): Declare.
9800 (ix86_macro_fusion_p): Declare.
9801 (ix86_macro_fusion_pair_p): Declare.
9802 (ix86_bd_has_dispatch): Declare.
9803 (ix86_bd_do_dispatch): Declare.
9804 (ix86_core2i7_init_hooks): Declare.
9805 (ix86_atom_sched_reorder): Declare.
9806 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
9807 (COSTS_N_BYTES): Move to x86-tune-costs.h.
9808 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
9809 (rip_relative_addr_p): Rename to ...
9810 (ix86_rip_relative_addr_p): ... this one; export.
9811 (memory_address_length): Update.
9812 (ix86_issue_rate): Move to x86-tune-sched.c.
9813 (ix86_flags_dependent): Move to x86-tune-sched.c.
9814 (ix86_agi_dependent): Move to x86-tune-sched.c.
9815 (exact_dependency_1): Move to x86-tune-sched.c.
9816 (exact_store_load_dependency): Move to x86-tune-sched.c.
9817 (ix86_adjust_cost): Move to x86-tune-sched.c.
9818 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
9819 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
9820 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
9821 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
9822 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
9823 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
9824 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
9825 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
9826 (min_insn_size): Rename to ...
9827 (ix86_min_insn_size): ... this one; export.
9828 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
9829 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
9830 (core2i7_first_cycle_multipass_backtrack): Move to
9831 x86-tune-sched-core.c.
9832 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
9833 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
9834 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
9835 (ix86_avoid_jump_mispredicts): Update.
9836 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
9837 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
9838 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
9839 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
9840 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
9841 (MAX_INSN): Move to ix86-tune-sched-bd.c.
9842 (MAX_IMM): Move to ix86-tune-sched-bd.c.
9843 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
9844 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
9845 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
9846 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
9847 (MAX_STORE): Move to ix86-tune-sched-bd.c.
9848 (BIG): Move to ix86-tune-sched-bd.c.
9849 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
9850 (enum insn_path): Move to ix86-tune-sched-bd.c.
9851 (get_mem_group): Move to ix86-tune-sched-bd.c.
9852 (is_cmp): Move to ix86-tune-sched-bd.c.
9853 (dispatch_violation): Move to ix86-tune-sched-bd.c.
9854 (is_branch): Move to ix86-tune-sched-bd.c.
9855 (is_prefetch): Move to ix86-tune-sched-bd.c.
9856 (init_window): Move to ix86-tune-sched-bd.c.
9857 (allocate_window): Move to ix86-tune-sched-bd.c.
9858 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
9859 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
9860 (process_end_window): Move to ix86-tune-sched-bd.c.
9861 (allocate_next_window): Move to ix86-tune-sched-bd.c.
9862 (find_constant): Move to ix86-tune-sched-bd.c.
9863 (get_num_immediates): Move to ix86-tune-sched-bd.c.
9864 (has_immediate): Move to ix86-tune-sched-bd.c.
9865 (get_insn_path): Move to ix86-tune-sched-bd.c.
9866 (get_insn_group): Move to ix86-tune-sched-bd.c.
9867 (count_num_restricted): Move to ix86-tune-sched-bd.c.
9868 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
9869 (add_insn_window): Move to ix86-tune-sched-bd.c.
9870 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
9871 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
9872 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
9873 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
9874 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
9875 (do_dispatch): Move to ix86-tune-sched-bd.c.
9876 (has_dispatch): Move to ix86-tune-sched-bd.c.
9877 * config/i386/t-i386: Add new object files.
9878 * config/i386/x86-tune-costs.h: New file.
9879 * config/i386/x86-tune-sched-atom.c: New file.
9880 * config/i386/x86-tune-sched-bd.c: New file.
9881 * config/i386/x86-tune-sched-core.c: New file.
9882 * config/i386/x86-tune-sched.c: New file.
9884 2017-10-11 Liu Hao <lh_mouse@126.com>
9886 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
9887 the generic version below instead.
9888 (should_colorize): Recognize Windows consoles as terminals
9890 * pretty-print.c [__MINGW32__] (write_all): New function.
9891 [__MINGW32__] (find_esc_head): Likewise.
9892 [__MINGW32__] (find_esc_terminator): Likewise.
9893 [__MINGW32__] (eat_esc_sequence): Likewise.
9894 [__MINGW32__] (mingw_ansi_fputs): New function that handles
9896 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
9899 2017-10-11 Richard Biener <rguenther@suse.de>
9901 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
9902 Properly call analyze_scalar_evolution with the loop of the stmt.
9904 2017-10-11 Richard Biener <rguenther@suse.de>
9906 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
9907 * tree-core.h (tree_base): Add chrec_var union member.
9908 * tree.h (CHREC_VAR): Remove.
9909 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
9910 * tree-chrec.h (build_polynomial_chrec): Adjust.
9911 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
9912 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
9914 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
9916 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
9917 * match.pd: ... here.
9918 ((T) X == (T) Y): Relax condition.
9920 2017-10-11 Bin Cheng <bin.cheng@arm.com>
9922 PR tree-optimization/82472
9923 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
9925 (break_alias_scc_partitions): Update postorder number.
9927 2017-10-11 Martin Liska <mliska@suse.cz>
9930 * opts.c (parse_no_sanitize_attribute): Do not use error_value
9932 * opts.h (parse_no_sanitize_attribute): Remove last argument.
9934 2017-10-11 Martin Liska <mliska@suse.cz>
9936 * print-rtl.c (print_insn): Move declaration of idbuf
9937 to same scope as name.
9939 2017-10-11 Martin Liska <mliska@suse.cz>
9944 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9947 2017-10-11 Martin Liska <mliska@suse.cz>
9950 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9953 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
9955 * config.gcc (default_gnu_indirect_function): Default to yes for
9956 arm*-*-linux* with glibc.
9958 2017-10-11 Richard Biener <rguenther@suse.de>
9960 * tree-scalar-evolution.c (get_scalar_evolution): Handle
9961 default-defs and types we do not want to analyze.
9962 (interpret_loop_phi): Replace unreachable code with an assert.
9963 (compute_scalar_evolution_in_loop): Remove and inline ...
9964 (analyze_scalar_evolution_1): ... here, replacing condition with
9965 what makes the intent clearer. Remove handling of cases
9966 get_scalar_evolution now handles.
9968 2017-10-10 Jim Wilson <wilson@tuliptree.org>
9970 PR rtl-optimization/81434
9971 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
9972 comment for main loop. In sched_group_found if, also add checks for
9973 pass and min_cost_group.
9975 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
9977 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
9978 (rs6000_insn_cost): New function.
9979 * config/rs6000/rs6000.md (cost): New attribute.
9981 2017-10-10 Jakub Jelinek <jakub@redhat.com>
9982 H.J. Lu <hongjiu.lu@intel.com>
9986 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
9987 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
9988 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
9989 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
9990 to OPTION_MASK_ISA_AVX512VL - builtins that have both
9991 OPTION_MASK_ISA_MMX and some other bit set require both
9992 mmx and the ISAs without the mmx bit.
9993 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
9994 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
9995 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
9996 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
9997 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
9998 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
9999 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
10000 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
10001 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
10002 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
10003 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
10004 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
10005 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
10006 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
10007 Add OPTION_MASK_ISA_MMX.
10009 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
10011 * config.gcc (armv7*-*-freebsd*): New target.
10012 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
10014 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
10016 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
10017 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
10020 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
10022 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
10023 a template parameter.
10024 (WIDE_INT_REF_FOR): Update accordingly.
10025 * tree.h (wi::int_traits <const_tree>): Delete.
10026 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
10027 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
10028 (wi::tree_to_wide_ref): New typedef.
10029 (wi::to_wide): New function.
10030 * calls.c (get_size_range): Use wi::to_wide when operating on
10031 trees as wide_ints.
10032 * cgraph.c (cgraph_node::create_thunk): Likewise.
10033 * config/i386/i386.c (ix86_data_alignment): Likewise.
10034 (ix86_local_alignment): Likewise.
10035 * dbxout.c (stabstr_O): Likewise.
10036 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
10037 * expr.c (const_vector_from_tree): Likewise.
10038 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
10039 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
10040 (fold_negate_expr_1, int_const_binop_1, const_binop)
10041 (fold_convert_const_int_from_real, optimize_bit_field_compare)
10042 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
10043 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
10044 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
10045 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
10046 (fold_not_const, round_up_loc): Likewise.
10047 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
10048 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
10049 (alloca_call_type): Likewise.
10050 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
10051 * godump.c (go_output_typedef): Likewise.
10052 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
10053 * internal-fn.c (get_min_precision): Likewise.
10054 * ipa-cp.c (ipcp_store_vr_results): Likewise.
10055 * ipa-polymorphic-call.c
10056 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
10057 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
10058 (ipa_modify_call_arguments): Likewise.
10059 * match.pd: Likewise.
10060 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
10061 * print-tree.c (print_node_brief, print_node): Likewise.
10062 * stmt.c (expand_case): Likewise.
10063 * stor-layout.c (layout_type): Likewise.
10064 * tree-affine.c (tree_to_aff_combination): Likewise.
10065 * tree-cfg.c (group_case_labels_stmt): Likewise.
10066 * tree-data-ref.c (dr_analyze_indices): Likewise.
10067 (prune_runtime_alias_test_list): Likewise.
10068 * tree-dump.c (dequeue_and_dump): Likewise.
10069 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
10070 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
10071 * tree-pretty-print.c (dump_generic_node): Likewise.
10072 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
10073 (simple_iv_with_niters): Likewise.
10074 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
10075 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
10076 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
10077 * tree-ssa-loop-niter.c (split_to_var_and_offset)
10078 (refine_value_range_using_guard, number_of_iterations_ne_max)
10079 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
10080 (get_cst_init_from_scev, record_nonwrapping_iv)
10081 (scev_var_range_cant_overflow): Likewise.
10082 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
10083 * tree-ssa-pre.c (compute_avail): Likewise.
10084 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
10085 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
10086 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
10087 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
10088 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
10089 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
10090 (dump_case_nodes, try_switch_expansion): Likewise.
10091 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
10092 (vect_do_peeling): Likewise.
10093 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
10094 * tree-vect-stmts.c (vectorizable_load): Likewise.
10095 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
10096 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
10097 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
10098 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
10099 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
10100 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
10101 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
10102 (evrp_dom_walker::before_dom_children): Likewise.
10103 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
10104 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
10105 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
10106 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
10107 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
10108 (get_range_pos_neg): Likewise.
10109 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
10110 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
10111 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
10112 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
10113 * config/avr/avr.c (avr_fold_builtin): Likewise.
10114 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
10115 * config/msp430/msp430.c (msp430_attr): Likewise.
10116 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
10117 * config/powerpcspe/powerpcspe-c.c
10118 (altivec_resolve_overloaded_builtin): Likewise.
10119 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
10120 (rs6000_expand_ternop_builtin): Likewise.
10121 * config/rs6000/rs6000-c.c
10122 (altivec_resolve_overloaded_builtin): Likewise.
10123 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
10124 (rs6000_expand_ternop_builtin): Likewise.
10125 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
10127 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10129 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
10130 when copying loop nest with only one inner loop.
10132 2017-10-10 Richard Biener <rguenther@suse.de>
10134 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
10135 blocks if SCEV is active.
10136 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
10138 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
10139 (scev_initialize): Assert we are not yet initialized.
10141 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10143 * tree-loop-distribution.c (generate_loops_for_partition): Remove
10144 inner loop's exit stmt by making it always exit the loop, otherwise
10145 we would generate an infinite empty loop.
10147 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10149 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
10150 renaming variables in new preheader if it's deleted.
10152 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10154 * tree-loop-distribution.c (struct partition): Remove unused field
10155 loops of the structure.
10156 (partition_alloc, partition_free): Ditto.
10157 (build_rdg_partition_for_vertex): Ditto.
10159 2017-10-09 Jeff Law <law@redhat.com>
10161 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
10162 return type to match prototype and documentation.
10164 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10166 * config/rs6000/rs6000.c (processor_costs): Move to ...
10167 * config/rs6000/rs6000.h: ... here.
10168 (rs6000_cost): Declare.
10170 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
10172 * except.c (setjmp_fn): New global variable.
10173 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
10174 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
10175 if DONT_USE_BUILTIN_SETJMP is defined.
10177 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10179 * target.def (insn_cost): New hook.
10180 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
10181 * doc/tm.texi: Regenerate.
10182 * rtlanal.c (insn_cost): Use the new hook.
10184 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10186 * combine.c (combine_validate_cost): Compute the new insn_cost,
10187 not just pattern_cost.
10188 (try_combine): Adjust comment.
10190 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10192 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
10194 * combine.c (uid_insn_cost): Adjust comment.
10195 (combine_validate_cost): Adjust comment. Use pattern_cost instead
10197 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
10198 * dse.c (find_shift_sequence): Ditto.
10199 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
10200 (bb_valid_for_noce_process_p): Use pattern_cost.
10201 * rtl.h (insn_rtx_cost): Delete.
10202 (pattern_cost): New prototype.
10203 (insn_cost): New prototype.
10204 * rtlanal.c (insn_rtx_cost): Rename to...
10205 (pattern_cost): ... this.
10208 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
10210 * config/i386/i386.md (*jcc_2): Remove insn pattern.
10211 (*jcc<mode>_0_r_i387): Ditto.
10212 (*jccxf_r_i387): Ditto.
10213 (*jcc<mode>_r_i387): Ditto.
10214 (*jccu<mode>_r_i387): Ditto.
10215 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
10216 (*jcc): Rename from *jcc_1.
10218 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10220 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
10221 deferred rescans after the lvx/stvx recombination pre-pass.
10223 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
10225 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
10226 memory operation instruction support.
10227 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
10228 (rs6000-ibm-aix[789]*): Likewise.
10229 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
10230 Document new functions.
10232 2017-10-09 Richard Biener <rguenther@suse.de>
10234 PR tree-optimization/82397
10235 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
10236 equality only for semantically equal trees.
10238 2017-10-09 Richard Biener <rguenther@suse.de>
10240 PR tree-optimization/82449
10241 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
10242 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
10243 allow constant addresses.
10244 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
10247 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10249 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
10252 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10255 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
10258 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10261 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
10263 2017-10-09 Jakub Jelinek <jakub@redhat.com>
10266 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
10267 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
10269 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
10271 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
10272 (WI_BINARY_PREDICATE_RESULT): Likewise.
10273 (wi::binary_traits::operator_result): New type.
10274 (wi::binary_traits::predicate_result): Likewise.
10275 (generic_wide_int::operator~, unary generic_wide_int::operator-)
10276 (generic_wide_int::operator==, generic_wide_int::operator!=)
10277 (generic_wide_int::operator&, generic_wide_int::and_not)
10278 (generic_wide_int::operator|, generic_wide_int::or_not)
10279 (generic_wide_int::operator^, generic_wide_int::operator+
10280 (binary generic_wide_int::operator-, generic_wide_int::operator*):
10282 (operator~, unary operator-, operator==, operator!=, operator&)
10283 (operator|, operator^, operator+, binary operator-, operator*): New
10285 * expr.c (get_inner_reference): Use wi::bit_and_not.
10286 * fold-const.c (fold_binary_loc): Likewise.
10287 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
10288 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
10289 (bit_value_binop): Likewise.
10290 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
10291 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
10292 (extract_range_from_binary_expr_1): Likewise.
10293 (masked_increment): Likewise.
10294 (simplify_bit_ops_using_ranges): Likewise.
10296 2017-10-09 Martin Jambor <mjambor@suse.cz>
10299 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
10300 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
10301 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
10303 (hsa_fixup_mov_insn_type): New function.
10304 (hsa_op_with_type::get_in_type): Use it.
10305 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
10306 immediates in an assert.
10307 (hsa_op_with_type::extend_int_to_32bit): New method.
10308 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
10309 types. Convert to dest type if necessary.
10310 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
10311 (reg_for_gimple_ssa): Pass false as min32int to
10312 hsa_type_for_scalar_tree_type.
10313 (gen_hsa_addr): Fixup type when creating addresable temporary.
10314 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
10315 (gen_hsa_unary_operation): Extend operands and convert to dest type if
10316 necessary. Call hsa_fixup_mov_insn_type.
10317 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
10318 extend operands and convert to dest type if necessary.
10319 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
10320 to dest type if necessary.
10321 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
10322 if conversion nt necessary and size matches.
10323 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
10324 to dest type if necessary.
10325 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
10326 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
10328 (gen_hsa_clrsb): Likewise.
10329 (gen_hsa_ffs): Likewise.
10330 (gen_hsa_divmod): Extend operands and convert to dest type if
10332 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
10334 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
10336 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
10337 Remove empty default arguments. Use a brace block as output
10339 (conditional return): Ditto.
10341 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
10342 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
10344 (group_ending_nop): Ditto.
10345 (doloop_end): Ditto.
10346 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
10347 (splitters for those): Ditto.
10349 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
10351 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
10352 a conditional jump (and the compare for it) so that pc_rtx is the
10354 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
10355 for the deleted and renamed ctr<mode>_internal[234] patterns.
10356 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
10357 Delete second conditional return pattern.
10358 (ctr<mode>_internal2): Delete this second bdnz pattern.
10359 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
10360 (ctr<mode>_internal4): Delete this second bdz pattern.
10362 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
10364 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
10365 (always_initialized_rtx_for_ssa_name_p): New predicate.
10366 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
10367 (finish_out_of_ssa): Free new field of SA.
10368 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
10369 * tree-ssa-coalesce.c: Include tree-ssa.h.
10370 (get_parm_default_def_partitions): Remove extern keyword.
10371 (get_undefined_value_partitions): New function.
10372 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
10373 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
10374 uninitialized bits.
10375 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
10377 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
10379 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
10381 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
10383 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
10384 for targets that preffer 128bit.
10386 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
10388 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
10390 2017-10-08 Olivier Hainque <hainque@adacore.com>
10392 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
10393 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
10395 (thumb_set_return_address): Likewise.
10397 2017-10-08 Olivier Hainque <hainque@adacore.com>
10399 * common/config/arm/arm-common.c (arm_except_unwind_info):
10400 Handle DWARF2_UNWIND_INFO.
10402 2017-10-07 Michael Collison <michael.collison@arm.com>
10404 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
10407 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
10409 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
10410 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
10411 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
10412 defined, force the creation of a new block for a dispatch label.
10414 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
10416 * invoke.texi (Wsuggest-attribute=cold): Document.
10417 * common.opt (Wsuggest-attribute=cold): New
10418 * ipa-pure-const.c (warn_function_cold): New function.
10419 * predict.c (compute_function_frequency): Use it.
10420 * predict.h (warn_function_cold): Declare.
10422 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
10424 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
10427 2017-10-06 Martin Liska <mliska@suse.cz>
10429 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
10430 keyword for member functions.
10431 (struct sanopt_tree_couple): New struct.
10432 (struct sanopt_tree_couple_hash): New function.
10433 (struct sanopt_ctx): Add new hash_map.
10434 (has_dominating_ubsan_ptr_check): New function.
10435 (record_ubsan_ptr_check_stmt): Likewise.
10436 (maybe_optimize_ubsan_ptr_ifn): Likewise.
10437 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
10438 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
10440 2017-10-06 Sudakshina Das <sudi.das@arm.com>
10443 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
10444 aarch64_simd_valid_immediate on CONST_VECTORs.
10445 (aarch64_reg_or_bic_imm): Likewise.
10447 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
10449 PR rtl-optimization/82396
10450 * haifa-sched.c (ready_sort_real): Disable qsort checking.
10452 2017-10-06 Sebastian Pop <sebpop@gmail.com>
10454 * graphite-dependences.c (scop_get_reads): Move code to...
10455 (scop_get_must_writes): Move code to...
10456 (scop_get_may_writes): Move code to...
10457 (scop_get_reads_and_writes): ... here.
10458 (scop_get_dependences): Call scop_get_reads_and_writes.
10460 2017-10-06 Jakub Jelinek <jakub@redhat.com>
10462 PR tree-optimization/82434
10463 * fold-const.h (can_native_encode_type_p,
10464 can_native_encode_string_p): Remove.
10465 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
10466 don't encode anything, just return what would be otherwise returned.
10467 (native_encode_fixed, native_encode_complex, native_encode_vector):
10469 (native_encode_string): Likewise. Inline by hand
10470 can_native_encode_string_p.
10471 (can_native_encode_type_p): Remove.
10472 (can_native_encode_string_p): Remove.
10473 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
10474 STRING_CSTs using can_native_encode_string_p, test all
10475 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
10476 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
10477 argument from native_encode_expr.
10478 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
10479 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
10482 2017-10-06 Richard Biener <rguenther@suse.de>
10484 PR tree-optimization/82397
10485 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
10486 operand_equal_p but rely on data_ref_compare_tree for detecting
10488 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
10489 to match up with dr_group_sort_cmp.
10491 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10494 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
10496 * config/s390/s390-builtin-types.def: Regenerate.
10498 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10501 * config/s390/s390-builtin-types.def: Regenerate.
10502 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
10503 Change flag from B_VXE to B_VX.
10504 (s390_vec_min_dbl): Remove B_VXE flag.
10506 2017-10-06 Richard Biener <rguenther@suse.de>
10508 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
10509 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
10510 translate_isl_ast_to_gimple::is_valid_rename,
10511 translate_isl_ast_to_gimple::get_rename,
10512 translate_isl_ast_to_gimple::get_def_bb_for_const,
10513 translate_isl_ast_to_gimple::get_new_name,
10514 translate_isl_ast_to_gimple::collect_all_ssa_names,
10515 translate_isl_ast_to_gimple::copy_loop_phi_args,
10516 translate_isl_ast_to_gimple::collect_all_ssa_names,
10517 translate_isl_ast_to_gimple::copy_loop_phi_args,
10518 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
10519 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
10520 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
10521 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
10522 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
10523 translate_isl_ast_to_gimple::copy_cond_phi_args,
10524 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
10525 translate_isl_ast_to_gimple::edge_for_new_close_phis,
10526 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
10527 translate_isl_ast_to_gimple::rename_uses,
10528 translate_isl_ast_to_gimple::rename_all_uses): Remove.
10529 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
10530 (set_rename_for_each_def): Likewise.
10531 (graphite_copy_stmts_from_block): Handle debug stmt resetting
10532 here. Handle rewriting SCEV analyzable uses here.
10533 (copy_bb_and_scalar_dependences): Generate code for PHI
10535 (graphite_regenerate_ast_isl): Adjust.
10536 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
10537 (add_write, add_read): New functions.
10538 (build_cross_bb_scalars_def): Use it and simplify.
10539 (build_cross_bb_scalars_use): Likewise.
10540 (graphite_find_cross_bb_scalar_vars): Inline into...
10541 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
10542 simulating out-of-SSA. Compute liveout and add dependencies.
10543 (build_scops): Force an empty entry block.
10544 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
10546 (sese_build_liveouts): Declare.
10547 (sese_trivially_empty_bb_p): Likewise.
10548 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
10549 compute liveout and debug_liveout.
10550 (sese_bad_liveouts_use): Remove.
10551 (sese_reset_debug_liveouts_bb): Likewise.
10552 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
10553 (sese_build_liveouts): Build liveout and debug_liveout and store
10555 (new_sese_info): Adjust.
10556 (free_sese_info): Likewise.
10557 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
10558 do not build liveout here.
10559 (move_sese_in_condition): Adjust region entry.
10560 (scev_analyzable_p): Match up with chrec_apply requirements.
10561 (sese_trivially_empty_bb_p): New.
10562 * tree-into-ssa.c (get_reaching_def): Properly support generating
10563 default-defs for incremental rewrite of anonymous names.
10565 2017-10-06 Richard Biener <rguenther@suse.de>
10567 * graphite-sese-to-poly.c (extract_affine): For casts increasing
10568 precision do not perform modulo reduction.
10570 2017-10-06 Richard Biener <rguenther@suse.de>
10572 PR tree-optimization/82436
10573 * tree-vect-slp.c (vect_supported_load_permutation_p): More
10574 conservatively choose the vectorization factor when checking
10575 whether we can perform the required load permutation.
10576 (vect_transform_slp_perm_load): Assert when we may not fail.
10578 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
10580 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
10581 message for incompatible -msdata=* and -mcall-* options.
10583 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10585 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
10586 rate for post-reload scheduling.
10588 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10590 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
10592 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10594 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
10595 to improve monte carlo in scimark.
10597 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10599 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
10600 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
10601 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
10602 pentium4_cost, nocona_cost): Set reassociation width to 1.
10603 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
10604 width to 2 for fp operations and 1 otherwise.
10605 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
10607 (atom_cost): Set reassociation width to 2.
10608 (slm_cost, generic_cost): Set fp reassociation width
10609 to 2 and 1 otherwise.
10610 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
10611 (core_cost): Set fp reassociation width to 4 and vector to 2.
10612 (ix86_reassociation_width): Rewrite using cost table; special case
10613 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
10614 and TARGET_AVX128_OPTIMAL.
10615 * config/i386/i386.h (processor_costs): Add
10616 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
10617 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
10618 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
10619 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
10620 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
10621 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
10623 2017-10-05 Nathan Sidwell <nathan@acm.org>
10625 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
10627 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10629 * config/arm/arm.c (arm_test_fpu_data): New.
10630 (arm_run_selftests): Call arm_test_fpu_data.
10632 2017-10-04 Nathan Sidwell <nathan@acm.org>
10634 * toplev.c (toplev::main): Remove excess parens on pretty_printer
10636 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
10638 2017-10-04 Sudakshina Das <sudi.das@arm.com>
10640 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
10641 check type for aarch64_simd_valid_immediate.
10642 (aarch64_output_simd_mov_immediate): Update prototype.
10643 (aarch64_simd_valid_immediate): Update prototype.
10644 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
10645 support for ORR-immediate.
10646 (and<mode>3): modified pattern to add support for BIC-immediate.
10647 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
10648 now checks for valid immediate for BIC and ORR based on new enum
10650 (aarch64_output_simd_mov_immediate): Function now used to output
10651 BIC/ORR imm as well based on new enum argument.
10652 * config/aarch64/constraints.md (Do): New vector immediate constraint.
10654 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
10655 (aarch64_reg_or_bic_imm): Likewise.
10657 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10659 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
10660 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
10662 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10666 PR rtl-optimization/82396
10667 * haifa-sched.c (autopref_multipass_init): Simplify
10669 (autopref_rank_data): Simplify sort order.
10670 * sched-int.h (autopref_multipass_data_): Remove
10671 multi_mem_insn_p, min_offset and max_offset.
10673 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10675 * doc/sourcebuild.texi: Document vect_peeling_profitable.
10677 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10679 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
10680 vect_doubleint_cvt.
10682 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10684 * doc/sourcebuild.texi: Document vect_long_mult.
10686 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
10688 PR tree-optimization/82413
10689 * fold-const.c (build_range_check): Use widest_int when comparing
10690 the maximum ETYPE value with HIGH.
10692 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10694 PR rtl-optimization/82396
10695 * haifa-sched.c (autopref_multipass_init): Simplify
10697 (autopref_rank_data): Simplify sort order.
10698 * sched-int.h (autopref_multipass_data_): Remove
10699 multi_mem_insn_p, min_offset and max_offset.
10701 2017-10-04 Jakub Jelinek <jakub@redhat.com>
10703 PR tree-optimization/82381
10704 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
10705 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
10708 PR tree-optimization/82374
10709 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
10710 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
10711 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
10712 current_function_decl to the new decl.
10714 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
10716 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
10717 helper macro for IEEE float128 hardware built-in functions.
10718 (SQRTF128_ODD): Add built-in functions with the round-to-odd
10720 (TRUNCF128_ODD): Likewise.
10721 (ADDF128_ODD): Likewise.
10722 (SUBF128_ODD): Likewise.
10723 (MULF128_ODD): Likewise.
10724 (DIVF128_ODD): Likewise.
10725 (FMAF128_ODD): Likewise.
10726 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
10727 UNSPEC_TRUNC_ROUND_TO_ODD.
10728 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
10729 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
10730 floating point round to odd instructions.
10731 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
10732 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
10733 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
10734 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
10735 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
10736 (trunc<mode>sf2_hw): Change the truncate with round to odd
10737 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
10738 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
10739 to odd hardware instructions.
10740 (sub<mode>3_odd): Likewise.
10741 (mul<mode>3_odd): Likewise.
10742 (div<mode>3_odd): Likewise.
10743 (sqrt<mode>2_odd): Likewise.
10744 (fma<mode>4_odd): Likewise.
10745 (fms<mode>4_odd): Likewise.
10746 (nfma<mode>4_odd): Likewise.
10747 (nfms<mode>4_odd): Likewise.
10748 (trunc<mode>df2_odd): Change the truncate with round to odd
10749 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
10751 * doc/extend.texi (PowerPC built-in functions): Update documentation
10752 for existing IEEE float128-bit built-in functions. Add built-in
10753 functions that generate the IEEE 128-bit floating point round to
10756 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
10758 PR rtl-optimization/77729
10759 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
10760 to (X&(C1&~C2))|C2 transformations.
10762 2017-10-03 Martin Jambor <mjambor@suse.cz>
10764 PR tree-optimization/82363
10765 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
10766 mismatch, mark lacc written regardless of racc.
10768 2017-10-03 Jakub Jelinek <jakub@redhat.com>
10770 PR tree-optimization/82381
10771 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
10772 stmt_to_insert nor wheather SSA_NAMEs are default defs.
10773 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
10774 fallthrough into reassoc_stmt_dominates_stmt_p.
10777 * combine.c (combine_instructions): Don't combine in unreachable
10780 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
10783 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
10784 function to not use the have_cpu variable. Do not set cpu_index,
10785 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
10786 or the default cpu.
10787 (rs6000_valid_attribute_p): Remove duplicate initializations of
10788 old_optimize and func_optimize.
10789 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
10790 (rs6000_activate_target_options): Make global.
10791 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
10794 2017-10-02 Jakub Jelinek <jakub@redhat.com>
10796 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
10797 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
10798 Set *poffset to 0 and *psize and *pmax_size to -1 if
10799 *poffset + *psize overflows in HOST_WIDE_INT.
10801 PR tree-optimization/82387
10802 PR tree-optimization/82388
10803 PR tree-optimization/82389
10804 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
10805 instead of live_bytes non-NULL.
10807 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
10810 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
10813 2017-10-02 Richard Biener <rguenther@suse.de>
10815 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
10816 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
10817 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
10819 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10821 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
10822 requested precision matches the type's.
10823 * calls.c (alloc_max_size): Calculate the new candidate size as
10824 a widest_int and use wi::to_widest when comparing it with the
10825 current candidate size.
10826 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
10827 zero rather than integer_zero_node.
10828 * match.pd: Check for a no-op conversion before using wi::add
10829 rather than after. Use tree_to_uhwi when summing small shift
10830 counts into an unsigned int.
10832 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10833 Alan Hayward <alan.hayward@arm.com>
10834 David Sherwood <david.sherwood@arm.com>
10837 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
10838 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
10839 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
10840 POINTER_AND_FP_REGS.
10842 2017-10-02 Richard Biener <rguenther@suse.de>
10844 PR tree-optimization/82355
10845 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
10846 a mapping for the enclosing loop but avoid generating one for
10847 the loop tree root.
10848 (copy_bb_and_scalar_dependences): Remove premature codegen
10849 error on PHIs in blocks duplicated into multiple places.
10850 * graphite-scop-detection.c
10851 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
10852 in the region use it as loop and nest to analyze the DR in.
10853 (try_generate_gimple_bb): Likewise.
10854 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
10855 (add_loop_constraints): For blocks in a loop not in the region
10856 create a dimension with a single iteration.
10857 * sese.h (gbb_loop_at_index): Remove assert.
10859 2017-10-01 Kevin Buettner <kevinb@redhat.com>
10861 * omp-expand.c (adjust_context_scope): New function.
10862 (expand_parallel_call): Call adjust_context_scope.
10864 2017-10-01 Jeff Law <law@redhat.com>
10866 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
10867 dom_opt_dom_walker class with direct access to private members.
10868 Add comments. Call test_for_singularity.
10869 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
10870 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
10871 m_dummy_cond anymore.
10872 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
10874 (pass_dominator:execute): Build the dummy_cond here and pass it
10875 to the dom_opt_dom_walker ctor.
10876 (test_for_singularity): New function.
10878 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
10879 Maya Rashish <coypu@sdf.org>
10881 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
10882 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
10883 (alpha*-*-netbsd*) Use nbsd_tm_file.
10884 (arm*-*-netbsdelf*) Likewise.
10885 (i[34567]86-*-netbsdelf*) Likewise.
10886 (x86_64-*-netbsd*) Likewise.
10887 (mips*-*-netbsd*) Likewise.
10888 (powerpc-*-netbsd*) Likewise.
10889 (sh*-*-netbsd*) Likewise.
10890 (sparc-*-netbsdelf*) Likewise.
10891 (sparc64-*-netbsd*) Likewise.
10892 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
10894 (vax-*-netbsdelf*) Likewise.
10895 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
10896 (UINT_FAST8_TYPE) Likewise.
10897 (INT_FAST16_TYPE) Check CHAR_FAST16.
10898 (UINT_FAST16_TYPE) Likewise.
10900 2017-09-30 Jakub Jelinek <jakub@redhat.com>
10903 * config/i386/i386.md
10904 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
10905 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
10906 *divmodsi4_zext_2): New define_insn_and_split.
10907 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
10908 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
10909 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
10910 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
10911 New define_insn_and_split.
10912 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
10913 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
10914 operands[1] having DImode when mode is SImode.
10916 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
10917 always SImode for DIV and MOD in REG_EQUAL notes.
10919 2017-09-29 Yury Gribov <tetra2005@gmail.com>
10921 PR middle-end/82319
10922 * match.pd: Fix handling of NaNs in pattern.
10924 2017-09-29 Jeff Law <law@redhat.com>
10926 * sbitmap.c (bitmap_bit_in_range_p): New function.
10927 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
10928 * tree-ssa-dse.c (live_bytes_read): New function.
10929 (dse_classify_store): Ignore reads of dead bytes.
10931 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
10932 typos and whitespace errors.
10933 * config/i386/predicates.md (address_no_seg_operand): Likewise.
10934 * config/s390/s390.c (s390_emit_prologue): Likewise.
10936 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10939 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
10940 with a symbol for LRA.
10942 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10944 PR rtl-optimization/82338
10945 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
10947 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
10949 * genmodes.c (calc_wider_mode): Suppress qsort macro.
10950 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
10951 (qsort_chk): Declare.
10952 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
10953 (qsort_chk): New function.
10955 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10957 PR tree-optimization/82337
10958 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
10959 phi definition if the PHI result appears in an abnormal PHI.
10960 (find_basis_for_base_expr): Don't record a basis if the LHS of the
10961 basis appears in an abnormal PHI.
10963 2017-09-29 Richard Biener <rguenther@suse.de>
10965 * graphite-isl-ast-to-gimple.c
10966 (translate_isl_ast_to_gimple::set_codegen_error): New function.
10967 (binary_op_to_tree): Use it.
10968 (get_rename_from_scev): Likewise.
10969 (copy_loop_phi_nodes): Likewise.
10970 (copy_bb_and_scalar_dependences): Likewise.
10971 (translate_pending_phi_nodes): Likewise.
10973 2017-09-29 Jakub Jelinek <jakub@redhat.com>
10976 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
10977 for movabsq $(i32 << shift), r64.
10979 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
10981 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
10982 index when encoding %esp as %rsp to avoid 0x67 prefix.
10984 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
10986 * config/i386/i386.md (*movsf_internal, *movdf_internal):
10987 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
10989 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
10991 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
10992 Extensions with more than 16 double VFP registers.
10993 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
10994 to_clear_mask and all code related to it. Replace the remaining
10995 entry by a sbitmap and adapt code accordingly.
10997 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
10999 * brig-builtins.def: Change pure attributes to const.
11001 2017-09-28 Joseph Myers <joseph@codesourcery.com>
11003 * config.gcc (default_gnu_indirect_function): Default to yes for
11004 sparc*-*-linux* with glibc.
11006 2017-09-28 Joseph Myers <joseph@codesourcery.com>
11008 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
11009 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
11010 when creating .init_array and .fini_array sections with priority
11013 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
11016 * config/aarch64/aarch64.c
11017 (aarch64_builtin_support_vector_misalignment): Always return false
11018 when misalignment is unknown.
11020 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
11022 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
11023 this function to return false if the definition used by the swap
11024 instruction is artificial, or if the memory address from which the
11025 constant value is loaded is not represented by a base address held
11026 in a register or if the base address register is a frame or stack
11027 pointer. Additionally, return false if the base address of the
11028 loaded constant is a SYMBOL_REF but is not considered to be a
11030 (replace_swapped_load_constant): New function.
11031 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
11032 loaded constant vector with a load of a swapped constant vector.
11034 2017-09-27 Carl Love <cel@us.ibm.com>
11036 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
11037 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
11038 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
11041 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
11043 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
11044 first, always call autopref_rank_data otherwise.
11046 2017-09-27 Richard Biener <rguenther@suse.de>
11048 * graphite-scop-detection.c (find_scop_parameters): Move
11049 loop bound handling ...
11050 (gather_bbs::before_dom_children): ... here, avoiding the need
11051 to build scop_info->loop_nest.
11052 (record_loop_in_sese): Remove.
11053 * sese.h (sese_info_t::loop_nest): Remove.
11054 * sese.c (new_sese_info): Do not allocate loop_nest.
11055 (free_sese_info): Do not free loop_nest.
11057 2017-09-27 Jakub Jelinek <jakub@redhat.com>
11060 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
11061 lhs from calls if the lhs has addressable type.
11063 2017-09-27 Richard Biener <rguenther@suse.de>
11065 * graphite.h (scop::max_alias_set): New member.
11066 * graphite-scop-detection.c: Remove references to non-existing
11067 --param in comments.
11068 (build_alias_sets): Record the maximum alias set used for drs.
11069 (build_scops): Support zero as unlimited for
11070 --param graphite-max-arrays-per-scop.
11071 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
11072 and inline into ...
11073 (build_poly_sr_1): ... here. Compute alias set based on the
11074 maximum alias set used for drs rather than
11075 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
11077 2017-09-27 Richard Biener <rguenther@suse.de>
11079 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
11080 --param loop-block-tile-size=0 to disable tiling.
11082 2017-09-27 Richard Biener <rguenther@suse.de>
11084 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
11085 (graphite-max-nb-scop-params): Document special value zero.
11086 * domwalk.h (dom_walker::STOP): New symbolical constant.
11087 (dom_walker::dom_walker): Add optional parameter for bb to
11089 (dom_walker::~dom_walker): Declare.
11090 (dom_walker::before_dom_children): Document STOP return value.
11091 (dom_walker::m_user_bb_to_rpo): New member.
11092 (dom_walker::m_bb_to_rpo): Likewise.
11093 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
11094 mapping here if not provided by the user.
11095 (dom_walker::~dom_walker): Free bb to RPO mapping if not
11096 provided by the user.
11097 (dom_walker::STOP): Define.
11098 (dom_walker::walk): Do not compute bb to RPO mapping here.
11099 Support STOP return value from before_dom_children to stop
11101 * graphite-optimize-isl.c (optimize_isl): If the schedule
11102 is the same still generate code if -fgraphite-identity
11103 or -floop-parallelize-all are given.
11104 * graphite-scop-detection.c: Include cfganal.h.
11105 (gather_bbs::gather_bbs): Get and pass through bb to RPO
11107 (gather_bbs::before_dom_children): Return STOP for BBs
11109 (build_scops): Compute bb to RPO mapping and pass it to
11110 the domwalk. Treat --param graphite-max-nb-scop-params=0
11111 as not limiting the number of params.
11112 * graphite.c (graphite_initialize): Remove limit on the
11113 number of basic-blocks in a function.
11114 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
11115 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
11116 default value of 10.
11118 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11120 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
11121 Adjust code to eliminate needing to do the shift right 32-bits
11122 operation after XSCVDPSPN.
11124 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11126 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
11127 ((X / Y) != 0 -> X >= Y): Likewise.
11129 2017-09-26 Carl Love <cel@us.ibm.com>
11131 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
11132 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
11133 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
11134 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
11135 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
11136 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
11137 definitions and overloading.
11138 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
11139 statement for P9V_BUILTIN_XST_LEN_R.
11140 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
11141 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
11142 define_expand and define_insn for the instructions and builtins.
11143 * doc/extend.texi: Update the built-in documentation file for the new
11144 built-in functions.
11145 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
11146 define_insn for the instructions
11148 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
11151 * gcc/config/netbsd-protos.h: New file.
11152 * gcc/config/netbsd.c: New file.
11153 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
11154 * gcc/config/t-netbsd: New file.
11155 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
11156 (tmake_file) Add t-netbsd.
11157 (extra_objs) Add netbsd.o.
11159 2017-09-26 Janus Weil <janus@gcc.gnu.org>
11163 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
11165 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11167 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
11168 sign extension from a vector register to a GPR by doing a 32-bit
11169 direct move and then an EXTSW.
11170 (extendsi<mode>2 splitter): Likewise.
11171 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
11172 right or vector extract after doing XSCVDPSPN. Use
11173 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
11175 (movdi_from_sf_zero_ext): Likewise.
11176 (reload_gpr_from_vsxsf): Likewise.
11177 (p8_mfvsrd_4_disf): Delete, no longer used.
11178 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
11179 then needing to move the SFmode to a GPR to use the XSCVDPSP
11180 instruction instead of FRSP and XSCVDPSPN.
11181 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
11182 it is adjacent to the other XSCVSPDP insns.
11183 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
11184 SFmode to be in traditional Altivec registers.
11185 (vsx_xscvdpspn): Eliminate useless alternative constraint.
11186 (vsx_xscvspdpn): Likewise.
11187 (vsx_xscvspdpn_scalar): Likewise.
11189 2017-09-26 Martin Jambor <mjambor@suse.cz>
11191 * tree-sra.c (compare_access_positions): Put integral types first,
11192 stabilize sorting of integral types, remove conditions putting
11193 non-full-precision integers last.
11194 (sort_and_splice_var_accesses): Disable scalarization if a
11195 non-integert would be represented by a non-full-precision integer.
11197 2017-09-26 Joseph Myers <joseph@codesourcery.com>
11199 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
11200 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
11201 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
11202 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
11203 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
11204 conditionals inside the function instead of around it. Call
11205 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
11206 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
11208 2017-09-26 Richard Biener <rguenther@suse.de>
11210 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
11212 (scop_detection::build_scop_breadth): ... this. Removed.
11213 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
11214 (scop_detection::harmful_stmt_in_bb): Likewise.
11215 (scop_detection::graphite_can_represent_stmt): Likewise.
11216 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
11217 (scop_detection::can_represent_loop): Remove recursion, fold in ...
11218 (scop_detection::can_represent_loop_1): ... this. Removed.
11219 (scop_detection::harmful_loop_in_region): Simplify after inlining
11220 the above and remove more quadraticness.
11221 (build_scops): Adjust.
11222 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
11225 2017-09-26 Jakub Jelinek <jakub@redhat.com>
11228 * config/i386/i386.c (ix86_print_operand_address_as): Only test
11229 REGNO (base) == SP_REG if base is a REG.
11231 PR middle-end/35691
11232 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
11233 if it is different SSA_NAME.
11234 (optimize_range_tests_cmp_bitwise): New function.
11235 (optimize_range_tests): Call it.
11237 2017-09-26 Richard Biener <rguenther@suse.de>
11239 PR tree-optimization/82321
11240 * graphite.c (canonicalize_loop_closed_ssa): Properly check
11241 for the def being inside the loop.
11243 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11245 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
11247 * config/s390/s390-builtins.def: Fix constraint on op4.
11249 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11251 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
11252 independent expanders.
11253 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
11254 ("vec_ordered", "vec_unordered"): New expanders.
11256 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11258 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
11261 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11263 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
11264 vec_unpacks_lo_v16qi.
11265 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
11267 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11269 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
11270 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
11271 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
11273 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11275 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
11277 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
11279 ("*vec_slb<mode>"): New insn pattern.
11280 ("vec_shr_<mode>"): New expander.
11281 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
11282 and force the shift count operand to V16QImode.
11283 ("vec_srb<mode>"): Set shift count mode to V16QI.
11285 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11287 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
11288 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
11289 ("vec_widen_smult_hi_<mode>"): New expander definitions.
11291 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
11294 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
11296 2017-09-26 Richard Biener <rguenther@suse.de>
11298 PR tree-optimization/82320
11299 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
11302 2017-09-25 Jeff Law <law@redhat.com>
11304 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
11305 prototype for new argument.
11306 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
11307 mostly extracted from rs6000_emit_allocate_stack.
11308 (rs6000_emit_probe_stack_range_stack_clash): New function.
11309 (rs6000_emit_allocate_stack): Call
11310 rs6000_emit_probe_stack_range_stack_clash as needed.
11311 (rs6000_emit_probe_stack_range): Add additional argument
11312 to call to gen_probe_stack_range{si,di}.
11313 (output_probe_stack_range): New.
11314 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
11315 (output_probe_stack_range_stack_clash): New.
11316 (rs6000_emit_prologue): Emit notes into dump file as requested.
11317 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
11318 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
11319 Add additional operand and pass it to output_probe_stack_range.
11321 2017-09-25 Bin Cheng <bin.cheng@arm.com>
11323 PR tree-optimization/82163
11324 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
11325 (checking_verify_loop_closed_ssa): New parameter.
11326 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
11327 (check_loop_closed_ssa_stmt): Delete.
11328 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
11329 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
11330 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
11333 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
11335 * brig-builtins.def: Treat HSAIL barrier builtins as
11336 setjmp/longjump style functions.
11338 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
11340 * target.def (constant_alignment): New hook.
11341 * defaults.h (CONSTANT_ALIGNMENT): Delete.
11342 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
11343 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
11344 * doc/tm.texi: Regenerate.
11345 * targhooks.h (default_constant_alignment): Declare.
11346 (constant_alignment_word_strings): Likewise.
11347 * targhooks.c (default_constant_alignment): New function.
11348 (constant_alignment_word_strings): Likewise.
11349 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
11350 instead of CONSTANT_ALIGNMENT.
11351 * varasm.c (align_variable, get_variable_align, build_constant_desc)
11352 (force_const_mem): Likewise.
11353 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
11354 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
11355 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
11356 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11357 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
11359 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
11360 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11361 constant_alignment_word_strings.
11362 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
11363 (CONSTANT_ALIGNMENT): Likewise.
11364 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11365 (arm_constant_alignment): New function.
11366 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
11367 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11368 constant_alignment_word_strings.
11369 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
11370 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11371 constant_alignment_word_strings.
11372 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
11373 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11374 (cris_constant_alignment): New function.
11375 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
11376 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11377 (epiphany_constant_alignment): New function.
11378 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
11379 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11380 constant_alignment_word_strings.
11381 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
11382 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11383 constant_alignment_word_strings.
11384 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
11385 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11386 constant_alignment_word_strings.
11387 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
11388 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
11389 * config/i386/i386.c (ix86_constant_alignment): Make static.
11390 Use the same interface as the target hook.
11391 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11392 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
11393 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11394 constant_alignment_word_strings.
11395 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
11396 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
11397 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11398 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
11399 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11400 constant_alignment_word_strings.
11401 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
11402 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11403 constant_alignment_word_strings.
11404 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
11405 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11406 constant_alignment_word_strings.
11407 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
11408 * config/microblaze/microblaze.c (microblaze_constant_alignment):
11410 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11411 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
11412 * config/mips/mips.c (mips_constant_alignment): New function.
11413 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11414 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
11415 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
11416 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11417 (mmix_constant_alignment): Make static. Use the same interface
11418 as the target hook.
11419 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
11420 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11421 constant_alignment_word_strings.
11422 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
11423 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11424 constant_alignment_word_strings.
11425 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
11426 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11427 constant_alignment_word_strings.
11428 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
11429 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11430 (rs6000_constant_alignment): New function.
11431 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
11432 * config/riscv/riscv.c (riscv_constant_alignment): New function.
11433 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11434 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
11435 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11436 (rs6000_constant_alignment): New function.
11437 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
11438 * config/s390/s390.c (s390_constant_alignment): New function.
11439 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11440 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
11441 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11442 constant_alignment_word_strings.
11443 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
11444 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11445 (sparc_constant_alignment): New function.
11446 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
11447 * config/spu/spu.c (spu_constant_alignment): New function.
11448 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11449 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
11450 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11451 constant_alignment_word_strings.
11452 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
11453 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11454 constant_alignment_word_strings.
11455 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
11456 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11457 constant_alignment_word_strings.
11458 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
11459 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11460 (visium_constant_alignment): New function.
11461 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
11462 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11463 (xtensa_constant_alignment): New function.
11464 * system.h (CONSTANT_ALIGNMENT): Poison.
11466 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
11468 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11469 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
11470 (rs6000_builtin_valid_without_lhs): New helper function.
11471 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11472 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
11474 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
11476 * target.h (vec_perm_indices): Use unsigned short rather than
11478 (auto_vec_perm_indices): Likewise.
11479 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
11480 Use unsigned int rather than unsigned char.
11481 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
11483 2017-09-25 Richard Biener <rguenther@suse.de>
11485 * cfgloop.h (sort_sibling_loops): Declare.
11486 * cfgloop.c (sort_sibling_loops_cmp): New helper.
11487 (sort_sibling_loops): New function sorting the sibling loop list
11489 * graphite.c (graphite_transform_loops): Sort sibling loops.
11491 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
11493 * target.def (vec_perm_const_ok): Change sel parameter to
11495 * optabs-query.c (can_vec_perm_p): Update accordingly.
11496 * doc/tm.texi: Regenerate.
11497 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
11498 auto_vec_perm_indices and remove separate nelt field.
11499 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
11500 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
11501 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
11502 (aarch64_expand_vec_perm_const): Update accordingly.
11503 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
11504 to vec_perm_indices.
11505 * config/arm/arm.c (expand_vec_perm_d): Change perm to
11506 auto_vec_perm_indices and remove separate nelt field.
11507 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
11508 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
11509 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
11511 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
11512 to vec_perm_indices.
11513 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
11514 sel to vec_perm_indices.
11515 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
11516 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
11517 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
11519 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
11522 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
11525 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
11526 on the FUNCTION_DECL function context if it has a DIE that is a
11529 2017-09-25 Richard Biener <rguenther@suse.de>
11531 PR tree-optimization/82285
11532 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
11535 2017-09-25 Tom de Vries <tom@codesourcery.com>
11539 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
11542 2017-09-25 Richard Biener <rguenther@suse.de>
11544 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
11545 ISL errors other than isl_error_quota happen. Dump if the
11546 schedule is the same.
11547 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
11548 errors instead of aborting inside ISL.
11550 2017-09-25 Iain Sandoe <iain@codesourcery.com>
11553 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
11554 of libgcc_eh for m64.
11555 * config/i386/darwin64.h: Likewise.
11557 2017-09-25 Richard Biener <rguenther@suse.de>
11559 PR middle-end/82144
11560 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
11561 attribute for incomplete types nor twice for complete ones.
11563 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
11566 * config/i386/i386.c (ix86_print_operand_address_as): Encode
11567 %esp as %rsp to avoid 0x67 prefix if there is no index or base
11570 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
11573 * config/i386/i386.opt (mprefer-avx256): Use
11574 ix86_target_flags variable.
11575 * config/i386/i386.c (ix86_target_string): Move
11576 -mprefer-avx256 to flag2_opts.
11578 2017-09-22 Jakub Jelinek <jakub@redhat.com>
11580 PR middle-end/35691
11581 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
11582 and x != -1 | y != -1 into (x & y) != -1.
11584 2017-09-22 Steve Ellcey <sellcey@cavium.com>
11586 * config.gcc: Add new case statement to set
11587 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
11588 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
11589 s390x-*-linux* case statements. Added aarch64 to the list of
11590 supported architectures.
11592 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11594 PR tree-optimization/82289
11595 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
11596 STMT_VINFO_RELEVANT_P.
11598 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11599 Alan Hayward <alan.hayward@arm.com>
11600 David Sherwood <david.sherwood@arm.com>
11602 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
11603 for VR_RANGE only; don't allow VR_ANTI_RANGE.
11604 (extract_range_from_binary_expr_1): Don't call
11605 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
11607 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11608 Alan Hayward <alan.hayward@arm.com>
11609 David Sherwood <david.sherwood@arm.com>
11611 * target.def (preferred_vector_alignment): New hook.
11612 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
11614 * doc/tm.texi: Regenerate.
11615 * targhooks.h (default_preferred_vector_alignment): Declare.
11616 * targhooks.c (default_preferred_vector_alignment): New function.
11617 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
11619 (DR_TARGET_ALIGNMENT): New macro.
11620 (aligned_access_p): Update commentary.
11621 (vect_known_alignment_in_bytes): New function.
11622 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
11624 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
11625 Calculate the misalignment based on the target alignment rather than
11627 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
11628 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
11629 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
11630 the target alignment, rather than masking the element misalignment
11631 with the number of elements in a vector. Also use the target
11632 alignment when calculating the maximum number of peels.
11633 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
11634 instead of TYPE_ALIGN_UNIT.
11635 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
11636 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
11637 (vect_create_addr_base_for_vector_ref): Update call accordingly.
11638 (vect_create_data_ref_ptr): Likewise.
11639 (vect_setup_realignment): Realign by ANDing with
11640 -DR_TARGET_MISALIGNMENT.
11641 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
11642 the number of peels based on DR_TARGET_ALIGNMENT.
11643 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
11644 with the guaranteed alignment boundary when deciding whether
11646 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
11647 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
11648 (ensure_base_align): Remove stmt_info parameter. Get the
11649 target base alignment from DR_TARGET_ALIGNMENT.
11650 (vectorizable_store): Update call accordingly. Interpret
11651 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
11653 (vectorizable_load): Likewise.
11655 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11656 Alan Hayward <alan.hayward@arm.com>
11657 David Sherwood <david.sherwood@arm.com>
11659 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
11660 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
11661 (vect_enhance_data_refs_alignment): Likewise.
11663 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11665 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
11666 error. Only quit immediately if parsing is complete.
11667 (BEGIN): Initialize fatal_err and parse_done.
11668 (begin fpu, end fpu): Check number of arguments.
11669 (begin arch, end arch): Likewise.
11670 (begin cpu, end cpu): Likewise.
11671 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
11672 (optalias): Likewise.
11674 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11676 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
11677 * config/arm/arm-isa.h: Delete. Move definitions to ...
11678 * arm-cpus.in: ... here. Use new feature and fgroup values.
11679 * config/arm/arm.c (arm_option_override): Use lower case for feature
11681 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
11682 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
11683 * config/arm/parsecpu.awk (END): Add new command 'isa'.
11685 (print_isa_bits_for): New function.
11686 (gen_isa): New function.
11687 (gen_comm_data): Use print_isa_bits_for.
11688 (define feature): New keyword.
11689 (define fgroup): New keyword.
11690 * config/arm/t-arm (TM_H): Remove.
11691 (GTM_H): Add arm-isa.h.
11692 (arm-isa.h): Add rule to generate file.
11693 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
11694 case for feature bit names.
11696 2017-09-22 Richard Biener <rguenther@suse.de>
11698 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
11700 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
11701 print of no dependency loops ...
11702 * graphite.c (graphite_transform_loops): ... here.
11703 (canonicalize_loop_closed_ssa_form): Work from inner to outer
11705 (same_close_phi_node, remove_duplicate_close_phi,
11706 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
11707 (canonicalize_loop_closed_ssa): ... here and simplify.
11708 * graphite-optimize-isl.c: Include tree-vectorizer.h.
11709 (optimize_isl): Use dump_printf_loc to tell when we stopped
11710 optimizing because of an ISL timeout.
11712 2017-09-22 Richard Biener <rguenther@suse.de>
11714 PR tree-optimization/82291
11715 * tree-if-conv.c (predicate_mem_writes): Make sure to
11716 remove writes in blocks predicated with false.
11718 2017-09-22 Richard Biener <rguenther@suse.de>
11720 * sese.c: Include cfganal.h.
11721 (if_region_set_false_region): Remove.
11722 (create_if_region_on_edge): Likewise.
11723 (move_sese_in_condition): Re-implement without destroying
11726 2017-09-22 Richard Biener <rguenther@suse.de>
11728 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
11729 Verify both BBs contain loop PHI nodes before dispatching to
11730 copy_loop_phi_args.
11731 (graphite_regenerate_ast_isl): Do not recompute dominators,
11732 do not verify three times. Restructure for clarity.
11733 * graphite-scop-detection.c (same_close_phi_node,
11734 remove_duplicate_close_phi, make_close_phi_nodes_unique,
11735 defined_in_loop_p, canonicalize_loop_closed_ssa,
11736 canonicalize_loop_closed_ssa_form): Simplify, remove excess
11737 checking and SSA rewrite, move to ...
11738 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
11739 (graphite_initialize): Do not pass in ctx, do not reset the
11740 SCEV cache, compute only dominators.
11741 (graphite_transform_loops): Allocate ISL ctx after
11742 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
11743 Maintain post-dominators only around build_scops.
11744 * sese.c (if_region_set_false_region): Make static. Free
11745 and recompute dominators.
11746 (move_sese_in_condition): Assert we don't get called with
11747 post-dominators computed.
11748 * sese.h (if_region_set_false_region): Remove.
11750 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
11752 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
11753 mode attribute for TARGET_AVX512VL.
11755 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
11757 * config/i386/i386.opt (mprefer-avx256): New option.
11758 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
11760 (ix86_preferred_simd_mode): Return 256-bit AVX modes
11761 for TARGET_PREFER_AVX256.
11762 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
11764 2017-09-21 Jeff Law <law@redhat.com>
11766 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
11767 Fix dump output if the only stack space is for pushed registers.
11769 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11771 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
11774 2017-09-21 Martin Sebor <msebor@redhat.com>
11777 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
11778 code (in C++) or code that triggers warnings.
11780 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
11782 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
11784 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
11786 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
11787 * sched-rgn.c: Ditto.
11788 * sel-sched-ir.c: Ditto.
11790 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
11792 * toplev.h (set_random_seed): Adjust return type.
11793 * toplev.c (init_local_tick): Move eager initialization of random_seed
11794 to get_random_seed. Adjust comment.
11795 (init_random_seed): Inline to get_random_seed, delete.
11796 (get_random_seed): Initialize random_seed lazily.
11797 (set_random_seed): Do not return previous value.
11798 (print_switch_value): Do not call get_random_seed.
11800 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
11802 * cgraph.c (delete_function_version): New, broken out from...
11803 (cgraph_node::delete_function_version): ...here. Rename to
11804 cgraph_node::delete_function_version_by_decl. Update all uses.
11805 (cgraph_node::remove): Call delete_function_version.
11807 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11810 * tree-inline.c (expand_call_inline): Emit clobber stmts for
11811 VAR_DECLs to which addressable non-volatile parameters are mapped
11812 and for id->retvar after the return value assignment. Clear
11813 id->retval and id->retbnd after inlining.
11815 2017-09-21 Richard Biener <rguenther@suse.de>
11817 PR tree-optimization/82276
11818 PR tree-optimization/82244
11819 * tree-vrp.c (build_assert_expr_for): Set
11820 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
11822 (remove_range_assertions): Revert earlier change.
11824 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
11827 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
11829 2017-09-21 Richard Biener <rguenther@suse.de>
11831 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11832 Restore valid IL after code generation errors.
11833 * graphite.c (graphite_transform_loops): Diagnose code
11834 generation issues as MSG_MISSED_OPTIMIZATION and continue
11835 with processing SCOPs.
11837 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11838 Alan Hayward <alan.hayward@arm.com>
11839 David Sherwood <david.sherwood@arm.com>
11841 * calls.c (compute_argument_addresses): Use simplify_gen_binary
11842 rather than choosing between plus_constant and gen_rtx_<CODE>.
11843 * expr.c (emit_push_insn): Likewise.
11844 (expand_expr_real_2): Likewise.
11846 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11847 Alan Hayward <alan.hayward@arm.com>
11848 David Sherwood <david.sherwood@arm.com>
11850 * loop-unroll.c (split_iv): Call copy_rtx on the step.
11852 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11853 Alan Hayward <alan.hayward@arm.com>
11854 David Sherwood <david.sherwood@arm.com>
11856 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
11857 calling tree_to_uhwi.
11859 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11860 Alan Hayward <alan.hayward@arm.com>
11861 David Sherwood <david.sherwood@arm.com>
11863 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
11864 INTEGER_CST rather than a negative test for ADDR_EXPR.
11866 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11867 Alan Hayward <alan.hayward@arm.com>
11868 David Sherwood <david.sherwood@arm.com>
11870 * tree-vrp.c (extract_range_from_binary_expr_1): Check
11871 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
11873 2017-09-21 Richard Biener <rguenther@suse.de>
11875 PR tree-optimization/71351
11876 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
11877 graphite_create_new_loop_guard): Remove, fold remaining parts
11879 (translate_isl_ast_node_for): ... here and simplify.
11881 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11884 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
11885 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
11886 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
11887 alternative always use QI mode, for -Os imov (=R,R) alternative
11888 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
11891 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11892 Jeff Law <law@redhat.com>
11894 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
11895 (allocate_stack_space): New function, partially extracted from
11896 s390_emit_prologue.
11897 (s390_emit_prologue): Track offset to most recent stack probe.
11898 Code to allocate space moved into allocate_stack_space.
11899 Dump actions when no stack is allocated.
11900 (s390_prologue_plus_offset): New function.
11901 (s390_emit_stack_probe): Likewise.
11903 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
11905 * common.opt (Wa, Wl, Wp, g, gz=): Add
11907 (gno-column-info): Remove.
11908 (gcolumn-info): Drop RejectNegative.
11909 (gno-): New prefix.
11910 (gno-record-gcc-switches): Remove.
11911 (grecord-gcc-switches): Drop RejectNegative.
11912 (gno-split-dwarf): Remove.
11913 (gsplit-dwarf): Drop RejectNegative.
11914 (gno-strict-dwarf): Remove.
11915 (gstrict-dwarf): Drop RejectNegative.
11916 * config/darwin.opt (gfull, gused): Add RejectNegative.
11917 * dwarf2out.c (gen_producer_string): Drop
11918 gno-record-gcc-switches handler.
11919 * optc-gen.awk: Add g to prefixes with negative forms.
11920 * opts-common.c (remapping_prefix_p): New.
11921 (find_opt): Check it.
11922 (generate_canonical_option): Test g prefix.
11923 (option_map): Add -gno- mapping.
11924 (add_misspelling_candidates): Check remapping_prefix_p.
11926 2017-09-20 Jeff Law <law@redhat.com>
11928 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
11929 thinko in stack clash protection support.
11931 * explow.c (compute_stack_clash_protection_loop_data): Use
11932 CONST_INT_P instead of explicit test. Verify object is a
11933 CONST_INT_P before looking at INTVAL.
11934 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
11935 instead of explicit test.
11937 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
11940 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
11941 address instead of to r1 and r11.
11943 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
11945 * config.gcc: Support "knm".
11946 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
11947 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
11949 * config/i386/i386.c (m_KNM): Define.
11950 (processor_target_table): Add "knm".
11952 (ix86_option_override_internal): Add "knm".
11953 (ix86_issue_rate): Add PROCESSOR_KNM.
11954 (ix86_adjust_cost): Ditto.
11955 (ia32_multipass_dfa_lookahead): Ditto.
11956 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
11957 (fold_builtin_cpu): Add M_INTEL_KNM.
11958 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
11959 (processor_type): Add PROCESSOR_KNM.
11960 * config/i386/x86-tune.def: Add m_KNM.
11961 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
11963 2017-09-20 Richard Biener <rguenther@suse.de>
11965 PR tree-optimization/80213
11966 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
11967 are allowed in empty BBs as well.
11968 (canonicalize_loop_closed_ssa): Also look for other complex
11970 (scop_detection::get_sese): Include the loop-closed PHI block
11972 (scop_detection::merge_sese): Remove code adding extra blocks.
11973 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
11974 (build_scops): Assert the final returned scop is invalid.
11976 2017-09-20 Richard Biener <rguenther@suse.de>
11978 PR tree-optimization/82264
11979 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
11981 (vn_phi_lookup): Likewise.
11982 (vn_phi_insert): Likewise.
11984 2017-09-20 Jakub Jelinek <jakub@redhat.com>
11986 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
11987 that fits into uhwi or shwi, add DW_AT_const_value regardless
11988 of early_dwarf without going through RTL, using add_AT_unsigned
11991 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
11992 (DEBUG_LTO_ABBREV_SECTION): Likewise.
11993 (DEBUG_LTO_MACINFO_SECTION): Likewise.
11994 (DEBUG_MACRO_SECTION): Likewise.
11995 (DEBUG_LTO_MACRO_SECTION): Likewise.
11996 (DEBUG_STR_DWO_SECTION): Likewise.
11997 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
11998 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
11999 (DEBUG_LTO_DWO_LINE_SECTION): Define.
12000 (DEBUG_LTO_LINE_STR_SECTION): Define.
12001 (init_sections_and_labels): Initialize debug_line_str_section
12002 variable. Initialize debug_loc_section for -gdwarf-5 to
12003 DEBUG_LOCLISTS_SECTION. Formatting fixes.
12005 2017-09-20 Richard Biener <rguenther@suse.de>
12007 * graphite-sese-to-poly.c (extract_affine): Properly handle
12008 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
12010 2017-09-20 Richard Biener <rguenther@suse.de>
12012 PR tree-optimization/81373
12013 * graphite-scop-detection.c (build_cross_bb_scalars_def):
12014 Force SESE live-out defs to be handled even if they are
12017 2017-09-19 Jeff Law <law@redhat.com>
12019 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
12020 nothing for stack adjustments with REG_STACK_CHECK.
12021 * sched-deps.c (parse_add_or_inc): Reject insns with
12022 REG_STACK_CHECK from dependency breaking.
12023 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
12024 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
12025 * reg-notes.def (STACK_CHECK): New note.
12027 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
12028 (ix86_expand_prologue): Dump stack clash info as needed.
12029 Call ix86_adjust_stack_and_probe_stack_clash as needed.
12031 * function.c (dump_stack_clash_frame_info): New function.
12032 * function.h (dump_stack_clash_frame_info): Prototype.
12033 (enum stack_clash_probes): New enum.
12035 * config/alpha/alpha.c (alpha_expand_prologue): Also check
12036 flag_stack_clash_protection.
12037 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
12038 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
12039 (arm_frame_pointer_required): Likewise.
12040 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
12041 (ia64_expand_prologue): Likewise.
12042 * config/mips/mips.c (mips_expand_prologue): Likewise.
12043 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
12044 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
12045 (sparc_flat_expand_prologue): Likewise.
12046 * config/spu/spu.c (spu_expand_prologue): Likewise.
12048 * explow.c: Include "params.h".
12049 (anti_adjust_stack_and_probe_stack_clash): New function.
12050 (get_stack_check_protect): Likewise.
12051 (compute_stack_clash_protection_loop_data): Likewise.
12052 (emit_stack_clash_protection_loop_start): Likewise.
12053 (emit_stack_clash_protection_loop_end): Likewise.
12054 (allocate_dynamic_stack_space): Use get_stack_check_protect.
12055 Use anti_adjust_stack_and_probe_stack_clash.
12056 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
12057 (emit_stack_clash_protection_loop_start): Likewise.
12058 (emit_stack_clash_protection_loop_end): Likewise.
12059 * rtl.h (get_stack_check_protect): Prototype.
12060 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
12061 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
12062 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
12064 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
12066 * doc/tm.texi: Rebuilt.
12067 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
12068 get_stack_check_protect.
12069 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
12070 * config/arm/arm.c (arm_expand_prologue): Likewise.
12071 (arm_frame_pointer_required): Likewise.
12072 * config/i386/i386.c (ix86_expand_prologue): Likewise.
12073 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
12074 * config/mips/mips.c (mips_expand_prologue): Likewise.
12075 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
12076 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
12077 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
12078 (sparc_flat_expand_prologue): Likewise.
12080 * common.opt (-fstack-clash-protection): New option.
12081 * flag-types.h (enum stack_check_type): Note difference between
12082 -fstack-check= and -fstack-clash-protection.
12083 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
12084 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
12085 * toplev.c (process_options): Issue warnings/errors for cases
12086 not handled with -fstack-clash-protection.
12087 * doc/invoke.texi (-fstack-clash-protection): Document new option.
12088 (-fstack-check): Note additional problem with -fstack-check=generic.
12089 Note that -fstack-check is primarily for Ada and refer users
12090 to -fstack-clash-protection for stack-clash-protection.
12091 Document new params for stack clash protection.
12093 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
12095 * config/i386/i386.c (ix86_split_long_move): Do not handle
12096 address used for LEA in a special way.
12098 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
12100 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
12102 2017-09-19 Martin Sebor <msebor@redhat.com>
12105 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
12106 of incompatible types.
12108 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
12110 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12111 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
12112 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12113 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
12115 2017-09-19 Richard Biener <rguenther@suse.de>
12117 PR tree-optimization/82244
12118 * tree-vrp.c (remove_range_assertions): Do not propagate
12119 a constant to abnormals but replace the assert with a copy.
12121 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
12123 PR rtl-optimization/57878
12124 PR rtl-optimization/68988
12125 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
12126 avoidance test involving non_reload_pseudos. Move frequency test
12127 below the general fragmentation avoidance test.
12129 2017-09-19 Richard Biener <rguenther@suse.de>
12131 PR tree-optimization/69728
12132 * graphite-sese-to-poly.c (schedule_error): New global.
12133 (add_loop_schedule): Handle empty domain by failing the
12135 (build_original_schedule): Handle schedule_error.
12137 2017-09-19 Richard Biener <rguenther@suse.de>
12139 * graphite-scop-detection.c (scop_detection::can_represent_loop):
12140 Do not iterate to sibling loops but only to siblings of inner
12143 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
12146 * config/m68k/m68k.md (moveq feeding equality comparison): Check
12147 that the registers are different.
12149 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
12151 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
12152 to processor_model and "amdfam17h" to arch_names_table.
12153 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
12155 2017-09-18 Jakub Jelinek <jakub@redhat.com>
12158 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
12160 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12161 Alan Hayward <alan.hayward@arm.com>
12162 David Sherwood <david.sherwood@arm.com>
12164 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
12166 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
12168 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
12169 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
12170 vect_schedule_slp_instance.
12171 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
12172 Update call to vect_slp_analyze_node_operations. Simplify return
12174 (vect_slp_analyze_bb_1): Update call accordingly.
12175 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
12176 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
12177 (vect_schedule_slp): Update call accordingly.
12179 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12180 Alan Hayward <alan.hayward@arm.com>
12181 David Sherwood <david.sherwood@arm.com>
12183 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
12184 with types that aren't in fact scalar.
12186 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12188 * tree-vect-slp.c (vect_record_max_nunits): New function,
12190 (vect_build_slp_tree_1): ...here.
12191 (vect_build_slp_tree_2): Call it for phis too.
12193 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12195 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
12196 to vect_get_vec_def_for_operand when getting the mask operand.
12198 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12199 Alan Hayward <alan.hayward@arm.com>
12200 David Sherwood <david.sherwood@arm.com>
12202 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
12205 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12206 Alan Hayward <alan.hayward@arm.com>
12207 David Sherwood <david.sherwood@arm.com>
12209 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
12210 calculation for vector booleans.
12212 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12213 Alan Hayward <alan.hayward@arm.com>
12214 David Sherwood <david.sherwood@arm.com>
12216 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
12218 (vect_transform_stmt): ...here.
12219 (vect_analyze_stmt): Use it instead of calling
12220 vectorizable_live_operation directly.
12222 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
12224 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
12225 non-SIMT targets in acc vector loops.
12227 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
12229 * configure.ac: Add arc and check if assembler supports gdwarf2.
12230 * configure: Regenerate.
12232 2017-09-18 Richard Biener <rguenther@suse.de>
12234 PR tree-optimization/82220
12235 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
12236 epilogue niters from the min_profitable_iters compute.
12238 2017-09-18 Jakub Jelinek <jakub@redhat.com>
12241 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
12242 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
12243 (ix86_init_pic_reg): Revert 2017-09-01 changes.
12245 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
12248 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
12249 switching to a new text section.
12251 2017-09-18 Richard Biener <rguenther@suse.de>
12253 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
12255 (build_alias_set): Reject aliases with no access function.
12257 2017-09-18 Richard Biener <rguenther@suse.de>
12259 PR tree-optimization/79622
12260 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
12262 (build_cross_bb_scalars_use): Likewise.
12264 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
12266 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
12268 2017-09-18 Alan Modra <amodra@gmail.com>
12271 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
12272 stack_pointer_rtx for count 0. Update comments. Break up
12273 large rtl expression.
12275 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
12277 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
12278 Increase to 20 bytes.
12279 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
12280 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
12281 or avx version of the stub.
12283 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
12286 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
12287 compute the minimum stack alignment. Also update preferred stack
12288 boundary for leaf functions.
12290 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
12292 PR tree-optimization/82228
12293 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
12296 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
12298 * common/config/nds32/nds32-common.c
12299 (nds32_option_optimization_table): Refine formatting.
12300 (nds32_option_optimization_table): Use -fsched-pressure and
12301 -fomit-frame-pointer for specific optimization level.
12303 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
12305 * config/nds32/nds32.c: Refine formatting and comments.
12306 * config/nds32/nds32.h: Likewise.
12307 * config/nds32/nds32.md: Likewise.
12308 * config/nds32/nds32-cost.c: Likewise.
12309 * config/nds32/nds32-isr.c: Likewise.
12310 * config/nds32/nds32-md-auxiliary.c: Likewise.
12311 * config/nds32/nds32-multiple.md: Likewise.
12312 * config/nds32/nds32-predicates.c: Likewise.
12314 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
12315 Jakub Jelinek <jakub@redhat.com>
12317 Add support for -std=c++2a.
12318 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
12320 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
12322 2017-09-15 Steve Ellcey <sellcey@cavium.com>
12325 * doc/extend.texi (Common Function Attributes): Add
12326 references to ARM, AArch64, and S/390 specific attributes.
12327 (Function Specific Option Pragmas): Add AArch64 and S/390
12328 to list of back ends that support the target pragma.
12330 2017-09-15 Nathan Sidwell <nathan@acm.org>
12332 * doc/standards.texi: Fix C++17 description. Update URLs for
12335 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
12337 * common.opt (Wcast-align=strict): New warning option.
12338 * doc/invoke.texi: Document -Wcast-align=strict.
12340 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
12342 * cgraph.h (cgraph_thunk_info): Add comments.
12343 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
12344 assert for VIRTUAL_* arguments stricter.
12346 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
12348 PR tree-optimization/71026
12349 * match.pd: Move RDIV patterns from fold-const.c
12350 * fold-const.c (distribute_real_division): Removed.
12351 (fold_binary_loc): Remove calls to distribute_real_divison.
12353 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12355 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
12356 c++1z and gnu++1z as deprecated. Change other references to
12357 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
12358 Change -Wc++1z-compat to -Wc++17-compat.
12359 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
12360 * dwarf2out.c (highest_c_language): Handle C++17.
12361 (gen_compile_unit_die): Likewise.
12363 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12365 PR rtl-optimization/82192
12366 * combine.c (make_extraction): Don't look through non-paradoxical
12367 SUBREGs or TRUNCATE if pos + len is or might be bigger than
12370 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
12371 Alan Hayward <alan.hayward@arm.com>
12372 David Sherwood <david.sherwood@arm.com>
12374 * target.def (function_arg_offset): New hook.
12375 * targhooks.h (default_function_arg_offset): Declare.
12376 * targhooks.c (default_function_arg_offset): New function.
12377 * function.c (locate_and_pad_parm): Use
12378 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
12379 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
12380 (TARGET_FUNCTION_ARG_OFFSET): ...this.
12381 * doc/tm.texi: Regenerate.
12382 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
12383 * config/spu/spu.c (spu_function_arg_offset): New function.
12384 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
12385 * system.h (FUNCTION_ARG_OFFSET): Poison.
12387 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
12388 Alan Hayard <alan.hayward@arm.com>
12389 David Sherwood <david.sherwood@arm.com>
12391 * target.def (truly_noop_truncation): New hook.
12392 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
12393 than TRULY_NOOP_TRUNCATION.
12394 * hooks.h (hook_bool_uint_uint_true): Declare.
12395 * hooks.c (hook_bool_uint_uint_true): New function.
12396 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
12397 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
12398 * doc/tm.texi: Regenerate.
12399 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
12400 rather than TRULY_NOOP_TRUNCATION in comments.
12401 (simplify_comparison): Likewise.
12402 (record_truncated_value): Likewise.
12403 * expmed.c (extract_bit_field_1): Likewise.
12404 (extract_split_bit_field): Likewise.
12405 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
12406 instead of TRULY_NOOP_TRUNCATION.
12407 * function.c (assign_parm_setup_block): Likewise.
12408 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
12409 * rtlhooks.c: Include target.h.
12410 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
12411 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
12412 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
12413 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
12414 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
12415 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
12416 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
12417 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
12418 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
12419 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
12420 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
12421 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
12422 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
12423 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
12424 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
12425 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
12426 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
12427 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
12428 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
12429 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
12430 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
12431 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
12432 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
12433 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
12434 * config/mips/mips.c (mips_truly_noop_truncation): New function.
12435 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
12436 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
12437 rather than TRULY_NOOP_TRUNCATION in comments.
12438 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
12439 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
12440 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
12441 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
12442 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
12443 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
12444 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
12445 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
12446 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
12447 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
12448 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
12449 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
12450 rather than TRULY_NOOP_TRUNCATION in comments.
12451 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
12452 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
12453 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
12454 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
12455 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
12456 TRULY_NOOP_TRUNCATION condition.
12457 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
12458 (TRULY_NOOP_TRUNCATION): Delete.
12459 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
12460 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
12461 * config/spu/spu.c (spu_truly_noop_truncation): New function.
12462 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
12463 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
12464 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
12465 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
12466 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
12467 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
12468 rather than TRULY_NOOP_TRUNCATION in comments.
12469 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
12470 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
12471 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
12472 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
12473 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
12474 * system.h (TRULY_NOOP_TRUNCATION): Poison.
12476 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
12479 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
12480 (*cmp_ior): Likewise.
12481 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
12482 (*ior_scc_scc_cmp): Likewise.
12483 (*and_scc_scc): Likewise.
12484 (*and_scc_scc_cmp): Likewise.
12486 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
12487 Alan Hayard <alan.hayward@arm.com>
12488 David Sherwood <david.sherwood@arm.com>
12490 * target.def (can_change_mode_class): New hook.
12491 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
12492 (hard_regno_nregs): Likewise.
12493 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
12494 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
12495 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
12496 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
12497 (LOAD_EXTEND_OP): Update accordingly.
12498 * doc/tm.texi: Regenerate.
12499 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
12500 CANNOT_CHANGE_MODE_CLASS.
12501 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
12502 (REG_CAN_CHANGE_MODE_P): ...this new macro.
12503 * combine.c (simplify_set): Update accordingly.
12504 * emit-rtl.c (validate_subreg): Likewise.
12505 * recog.c (general_operand): Likewise.
12506 * regcprop.c (mode_change_ok): Likewise.
12507 * reload1.c (choose_reload_regs): Likewise.
12508 (inherit_piecemeal_p): Likewise.
12509 * rtlanal.c (simplify_subreg_regno): Likewise.
12510 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
12511 instead of CANNOT_CHANGE_MODE_CLASS.
12512 (reload_cse_simplify_operands): Likewise.
12513 * reload.c (push_reload): Use targetm.can_change_mode_class
12514 instead of CANNOT_CHANGE_MODE_CLASS.
12515 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
12516 REG_CANNOT_CHANGE_MODE_P.
12517 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12518 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
12519 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12520 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12521 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12522 (arm_can_change_mode_class): New function.
12523 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
12524 than CANNOT_CHANGE_MODE_CLASS in comments.
12525 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12526 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
12527 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
12528 (ix86_can_change_mode_class): ...this new function, inverting the
12529 sense of the return value.
12530 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12531 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12532 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12533 (ia64_can_change_mode_class): New function.
12534 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12535 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
12536 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
12537 (m32c_can_change_mode_class): ...this new function, inverting the
12538 sense of the return value.
12539 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12540 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12541 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
12542 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
12543 (mips_can_change_mode_class): ...this new function, inverting the
12544 sense of the return value.
12545 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12546 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12547 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12548 (msp430_can_change_mode_class): New function.
12549 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12550 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
12551 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12552 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12553 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12554 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
12555 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12556 (pa_cannot_change_mode_class): Replace with...
12557 (pa_can_change_mode_class): ...this new function, inverting the
12558 sense of the return value.
12559 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
12560 than CANNOT_CHANGE_MODE_CLASS in comments.
12561 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12562 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
12563 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12564 (pdp11_cannot_change_mode_class): Replace with...
12565 (pdp11_can_change_mode_class): ...this new function, inverting the
12566 sense of the return value.
12567 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12568 * config/powerpcspe/powerpcspe-protos.h
12569 (rs6000_cannot_change_mode_class_ptr): Delete.
12570 * config/powerpcspe/powerpcspe.c
12571 (rs6000_cannot_change_mode_class_ptr): Delete.
12572 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12573 (rs6000_option_override_internal): Assign to
12574 targetm.can_change_mode_class instead of
12575 rs6000_cannot_change_mode_class_ptr.
12576 (rs6000_cannot_change_mode_class): Replace with...
12577 (rs6000_can_change_mode_class): ...this new function, inverting the
12578 sense of the return value.
12579 (rs6000_debug_cannot_change_mode_class): Replace with...
12580 (rs6000_debug_can_change_mode_class): ...this new function.
12581 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12582 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
12583 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12584 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12585 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
12587 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
12588 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12589 (rs6000_option_override_internal): Assign to
12590 targetm.can_change_mode_class instead of
12591 rs6000_cannot_change_mode_class_ptr.
12592 (rs6000_cannot_change_mode_class): Replace with...
12593 (rs6000_can_change_mode_class): ...this new function, inverting the
12594 sense of the return value.
12595 (rs6000_debug_cannot_change_mode_class): Replace with...
12596 (rs6000_debug_can_change_mode_class): ...this new function.
12597 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12598 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
12599 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
12600 (s390_can_change_mode_class): ...this new function, inverting the
12601 sense of the return value.
12602 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12603 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12604 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
12605 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12606 (sh_cannot_change_mode_class): Replace with...
12607 (sh_can_change_mode_class): ...this new function, inverting the
12608 sense of the return value.
12609 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12610 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12611 (sparc_can_change_mode_class): New function.
12612 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12613 * config/spu/spu.c (spu_can_change_mode_class): New function.
12614 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12615 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12616 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12617 (visium_can_change_mode_class): New function.
12618 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
12620 2017-09-15 Richard Biener <rguenther@suse.de>
12622 PR tree-optimization/82217
12623 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
12624 but not undefined case.
12626 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12629 * postreload.c (reload_cse_simplify_operands): Skip
12630 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
12632 2017-09-15 Richard Biener <rguenther@suse.de>
12634 PR tree-optimization/68823
12635 * graphite-scop-detection.c (build_alias_set): If we have a
12636 possible dependence check whether we can handle them by just
12637 looking at the DRs DR_ACCESS_FNs.
12638 (build_scops): If build_alias_set fails, fail the SCOP.
12640 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
12642 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
12643 to support float128 built-in functions that require the ISA 3.0
12645 (BU_FLOAT128_3_HW): Likewise.
12646 (SQRTF128): Add support for the IEEE 128-bit square root and fma
12647 built-in functions.
12648 (FMAF128): Likewise.
12650 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
12651 support for built-in functions that need the ISA 3.0 IEEE 128-bit
12652 floating point instructions.
12653 (rs6000_invalid_builtin): Likewise.
12654 (rs6000_builtin_mask_names): Likewise.
12655 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
12656 (RS6000_BTM_FLOAT128_HW): Likewise.
12657 (RS6000_BTM_COMMON): Likewise.
12658 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
12660 * doc/extend.texi (RS/6000 built-in functions): Document the
12661 IEEE 128-bit floating point square root and fused multiply-add
12662 built-in functions.
12664 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
12666 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
12667 reg (r2) isn't in the set of registers defined in the prologue.
12669 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12670 Alan Hayward <alan.hayward@arm.com>
12671 David Sherwood <david.sherwood@arm.com>
12673 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
12674 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
12675 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
12676 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
12677 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
12679 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
12680 max_vectorization_factor.
12681 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
12683 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12684 Alan Hayward <alan.hayward@arm.com>
12685 David Sherwood <david.sherwood@arm.com>
12687 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
12688 (vect_worthwhile_without_simd_p): Declare.
12689 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
12690 (vectorizable_reduction): Use it.
12691 * tree-vect-stmts.c (vectorizable_shift): Likewise.
12692 (vectorizable_operation): Likewise.
12694 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12695 Alan Hayward <alan.hayward@arm.com>
12696 David Sherwood <david.sherwood@arm.com>
12698 * tree-vectorizer.h (vect_get_num_copies): New function.
12699 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
12700 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12701 (vectorizable_induction): Likewise.
12702 (vectorizable_live_operation): Likewise.
12703 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12704 (vectorizable_bswap): Likewise.
12705 (vectorizable_call): Likewise.
12706 (vectorizable_conversion): Likewise.
12707 (vectorizable_assignment): Likewise.
12708 (vectorizable_shift): Likewise.
12709 (vectorizable_operation): Likewise.
12710 (vectorizable_store): Likewise.
12711 (vectorizable_load): Likewise.
12712 (vectorizable_condition): Likewise.
12713 (vectorizable_comparison): Likewise.
12714 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
12716 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12717 Alan Hayward <alan.hayward@arm.com>
12718 David Sherwood <david.sherwood@arm.com>
12720 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
12721 of vect_init_vector.
12723 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12724 Alan Hayward <alan.hayward@arm.com>
12725 David Sherwood <david.sherwood@arm.com>
12727 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
12728 an inline wrapper that provides a location.
12729 (gimple_build_vector): Likewise.
12730 * gimple-fold.c (gimple_build_vector_from_val): New function.
12731 (gimple_build_vector): Likewise.
12732 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
12733 functions to build the initial value. Always return a gimple value.
12734 (get_initial_defs_for_reduction): Likewise. Only compute
12736 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
12737 vect_init_vector on the results from get_initial_def(s)_for_reduction.
12738 (vectorizable_induction): Use gimple_build_vector rather than
12741 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12742 Alan Hayward <alan.hayward@arm.com>
12743 David Sherwood <david.sherwood@arm.com>
12745 * target.h (vec_perm_indices): New typedef.
12746 (auto_vec_perm_indices): Likewise.
12747 * optabs-query.h: Include target.h
12748 (can_vec_perm_p): Take a vec_perm_indices *.
12749 * optabs-query.c (can_vec_perm_p): Likewise.
12750 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
12751 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12752 * tree-vect-generic.c (lower_vec_perm): Likewise.
12753 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
12754 (vect_grouped_load_supported): Likewise.
12755 (vect_shift_permute_load_chain): Likewise.
12756 (vect_permute_store_chain): Use auto_vec_perm_indices.
12757 (vect_permute_load_chain): Likewise.
12758 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
12759 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
12760 Update uses of can_vec_perm_p.
12761 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
12762 mode with a number of elements. Take a vec_perm_indices *.
12763 (vect_create_epilog_for_reduction): Update accordingly.
12764 Use auto_vec_perm_indices.
12765 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
12766 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
12767 (vect_transform_slp_perm_load): Likewise.
12768 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
12769 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
12770 (vect_gen_perm_mask_checked): Likewise.
12771 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
12772 (vect_gen_perm_mask_checked): Likewise.
12773 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
12774 (vectorizable_store): Likewise.
12775 (vectorizable_load): Likewise.
12776 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
12777 (vectorizable_bswap): Likewise.
12779 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12780 Alan Hayward <alan.hayward@arm.com>
12781 David Sherwood <david.sherwood@arm.com>
12783 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
12784 * tree.c (build_vector): Likewise.
12785 (build_vector_from_ctor): Update accordingly.
12786 (build_vector_from_val): Likewise.
12787 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
12788 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12789 * tree-vect-generic.c (add_rshift): Likewise.
12790 (expand_vector_divmod): Likewise.
12791 (optimize_vector_constructor): Likewise.
12792 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
12793 (vect_transform_slp_perm_load): Likewise.
12794 (vect_schedule_slp_instance): Likewise.
12795 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
12796 (vectorizable_call): Likewise.
12797 (vect_gen_perm_mask_any): Likewise. Add elements in order.
12798 * expmed.c (make_tree): Likewise.
12799 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
12800 a vector passed to build_vector.
12801 (fold_convert_const): Likewise.
12802 (exact_inverse): Likewise.
12803 (fold_ternary_loc): Likewise.
12804 (fold_relational_const): Likewise.
12805 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
12806 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
12807 (const_unop): Likewise. Store the reduction accumulator in a
12808 variable rather than an array.
12809 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
12810 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
12811 the new vector, rather than constructing it after the input arrays.
12812 (native_interpret_vector): Use auto_vec<tree> when building
12813 a vector passed to build_vector. Add elements in order.
12814 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
12815 auto_vec<tree> when building a vector passed to build_vector.
12816 (vect_create_epilog_for_reduction): Likewise.
12817 (vectorizable_induction): Likewise.
12818 (get_initial_def_for_reduction): Likewise. Fix indentation of
12820 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
12822 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
12823 passed to build_vector.
12825 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12826 Alan Hayward <alan.hayward@arm.com>
12827 David Sherwood <david.sherwood@arm.com>
12829 * tree-core.h (tree_base::u): Add an "nelts" field.
12830 (tree_vector): Use VECTOR_CST_NELTS as the length.
12831 * tree.c (tree_size): Likewise.
12832 (make_vector): Initialize VECTOR_CST_NELTS.
12833 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
12834 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
12835 TYPE_VECTOR_SUBPARTS.
12836 * expr.c (const_vector_mask_from_tree): Consistently use "units"
12837 as the number of units, setting it from VECTOR_CST_NELTS.
12838 (const_vector_from_tree): Likewise.
12839 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
12840 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
12841 (fold_negate_expr_1): Likewise.
12842 (fold_convert_const): Likewise.
12843 (const_binop): Likewise. Differentiate the number of output and
12845 (const_unop): Likewise.
12846 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
12847 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
12848 in cases that did the opposite.
12850 2017-09-14 Richard Biener <rguenther@suse.de>
12852 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
12855 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
12857 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
12859 2017-09-14 Jakub Jelinek <jakub@redhat.com>
12862 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
12863 if and where to split a bb, except for splitting before debug insn
12864 sequences followed by non-label real insn. Delete debug insns
12865 in between basic blocks.
12867 * combine.c (make_compound_operation_int): Formatting fixes.
12869 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
12870 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12871 * config/netbsd.h (LINK_EH_SPEC): Likewise.
12872 * config/sol2.h (LINK_EH_SPEC): Likewise.
12873 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12874 * config/s390/linux.h (LINK_SPEC): Likewise.
12875 * config/freebsd.h (LINK_EH_SPEC): Likewise.
12876 * config/openbsd.h (LINK_EH_SPEC): Likewise.
12877 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12878 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
12879 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
12880 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12881 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
12882 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
12884 2017-09-13 Jakub Jelinek <jakub@redhat.com>
12886 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
12888 (ENDFILE_LINUX_SPEC): Likewise.
12889 (LINK_EH_SPEC): Likewise.
12890 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
12891 (LINK_OS_LINUX_SPEC32): Likewise.
12892 (LINK_OS_LINUX_SPEC64): Likewise.
12893 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
12894 (LINK_OS_LINUX_SPEC): Likewise.
12896 2017-09-13 Martin Liska <mliska@suse.cz>
12898 PR middle-end/82154
12899 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
12900 CASE_HIGH is NULL_TREE.
12902 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12903 Alan Hayward <alan.hayward@arm.com>
12904 David Sherwood <david.sherwood@arm.com>
12906 * target.def (secondary_memory_needed): New hook.
12907 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
12908 instead of SECONDARY_MEMORY_NEEDED.
12909 (secondary_memory_needed_mode): Likewise.
12910 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
12911 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
12912 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
12913 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
12914 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
12915 * doc/tm.texi: Regenerate.
12916 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
12917 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
12918 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12919 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
12920 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
12921 * config/i386/i386.c (inline_secondary_memory_needed): Put the
12922 mode argument first and change the reg_class arguments to reg_class_t.
12923 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
12924 Make static. Update the call to inline_secondary_memory_needed.
12925 (ix86_register_move_cost): Update the call to
12926 inline_secondary_memory_needed.
12927 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12928 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
12930 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
12931 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12933 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
12934 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
12935 * config/mips/mips.c (mips_secondary_memory_needed): Make static
12936 and match hook interface. Add comment from mips.h.
12937 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12938 * config/mmix/mmix.md (truncdfsf2): Refer to
12939 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12941 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
12942 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
12943 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12944 (pa_secondary_memory_needed): New function.
12945 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
12946 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
12947 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12948 (pdp11_secondary_memory_needed): Make static and match hook interface.
12949 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
12950 * config/powerpcspe/powerpcspe-protos.h
12951 (rs6000_secondary_memory_needed_ptr): Delete.
12952 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
12954 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12955 (rs6000_option_override_internal): Assign to
12956 targetm.secondary_memory_needed rather than
12957 rs6000_secondary_memory_needed_ptr.
12958 (rs6000_secondary_memory_needed): Match hook interface.
12959 (rs6000_debug_secondary_memory_needed): Likewise.
12960 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
12961 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
12962 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
12963 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12964 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
12965 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
12967 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
12968 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12969 (rs6000_option_override_internal): Assign to
12970 targetm.secondary_memory_needed rather than
12971 rs6000_secondary_memory_needed_ptr.
12972 (rs6000_secondary_memory_needed): Match hook interface.
12973 (rs6000_debug_secondary_memory_needed): Likewise.
12974 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
12975 * config/s390/s390.c (s390_secondary_memory_needed): New function.
12976 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12977 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
12978 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12979 (sparc_secondary_memory_needed): New function.
12980 * lra-constraints.c (check_and_process_move): Refer to
12981 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12983 (curr_insn_transform): Likewise.
12984 (process_alt_operands): Use targetm.secondary_memory_needed
12985 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12986 (check_secondary_memory_needed_p): Likewise.
12987 (choose_split_class): Likewise.
12988 * reload.c: Unconditionally include code that was previously
12989 conditional on SECONDARY_MEMORY_NEEDED.
12990 (push_secondary_reload): Use targetm.secondary_memory_needed
12991 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12992 (push_reload): Likewise.
12993 * reload1.c: Unconditionally include code that was previously
12994 conditional on SECONDARY_MEMORY_NEEDED.
12995 (choose_reload_regs): Use targetm.secondary_memory_needed
12996 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12997 (gen_reload): Likewise.
12998 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
13000 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
13001 Alan Hayward <alan.hayward@arm.com>
13002 David Sherwood <david.sherwood@arm.com>
13004 * target.def (secondary_memory_needed_mode): New hook:
13005 * targhooks.c (default_secondary_memory_needed_mode): Declare.
13006 * targhooks.h (default_secondary_memory_needed_mode): New function.
13007 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
13008 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
13009 * doc/tm.texi: Regenerate.
13010 * lra-constraints.c (check_and_process_move): Use
13011 targetm.secondary_memory_needed_mode instead of
13012 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
13013 (curr_insn_transform): Likewise.
13014 * reload.c (get_secondary_mem): Likewise.
13015 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13016 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
13018 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13019 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13020 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
13021 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13022 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
13024 * config/powerpcspe/powerpcspe-protos.h
13025 (rs6000_secondary_memory_needed_mode): Delete.
13026 * config/powerpcspe/powerpcspe.c
13027 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13028 (rs6000_secondary_memory_needed_mode): Make static.
13029 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13030 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
13032 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
13034 (rs6000_secondary_memory_needed_mode): Make static.
13035 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13036 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
13037 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13038 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13039 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
13041 (sparc_secondary_memory_needed_mode): New function.
13042 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
13044 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
13046 * config/aarch64/constraints.md (Umq): New constraint.
13047 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
13049 (mov<mode>): Update condition.
13051 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13053 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
13054 when bitposition is the same.
13056 2017-09-13 Richard Biener <rguenther@suse.de>
13058 * dwarf2out.c (output_die_symbol): Remove.
13059 (output_die): Do not output a DIEs symbol.
13061 2017-09-13 Richard Biener <rguenther@suse.de>
13063 PR middle-end/82128
13064 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
13065 default-def to avoid breaking iterator update with the weird
13066 interaction with cgraph_update_edges_for_call_stmt_node.
13068 2017-09-13 Richard Biener <rguenther@suse.de>
13070 * tree-cfg.c (verify_gimple_assign_binary): Add verification
13071 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
13072 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
13073 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
13075 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
13077 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
13078 Disable pc relative literal load irrespective of
13079 TARGET_FIX_ERR_A53_84341 for default.
13081 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
13083 * config/sparc/sparc.c (output_return): Output the source location of
13084 the insn in the delay slot, if any.
13085 (output_sibcall): Likewise.
13087 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
13090 * common.opt (-static-pie): New alias.
13091 (shared): Negate static-pie.
13092 (-no-pie): Update help text.
13094 (static-pie): New option.
13095 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
13096 -static-pie support.
13097 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
13098 (LINK_EH_SPEC): Likewise.
13099 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13100 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
13101 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
13102 * gcc.c (LINK_COMMAND_SPEC): Likewise.
13103 (init_gcc_specs): Likewise.
13104 (init_spec): Likewise.
13105 (display_help): Update help message for -pie.
13106 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
13109 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
13111 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
13112 (movdi_aarch64): Likewise.
13113 (movti_aarch64): Likewise.
13115 2017-09-12 Simon Wright <simon@pushface.org>
13118 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
13119 calculation of the minor version, always output as 0.
13121 2017-09-12 Jakub Jelinek <jakub@redhat.com>
13124 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
13125 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
13126 on it early, rather than manual conversion late. For
13127 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
13128 instead of performing manual conversion.
13130 2017-09-12 Carl Love <cel@us.ibm.com>
13132 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
13133 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
13135 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13136 VMULOSW): Add definitions.
13137 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13138 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13139 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13140 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13141 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13143 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
13145 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
13147 (movti_aarch64): Likewise.
13148 (movdf_aarch64): Likewise.
13149 (movtf_aarch64): Likewise.
13150 (load_pairdi): Likewise.
13151 (store_pairdi): Likewise.
13152 (load_pairdf): Likewise.
13153 (store_pairdf): Likewise.
13154 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
13155 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
13156 (ldr_got_small_<mode>): Likewise.
13157 (ldr_got_small_28k_<mode>): Likewise.
13158 (ldr_got_tiny): Likewise.
13159 * config/aarch64/iterators.md (ldst_sz): New.
13160 (ldpstp_sz): Likewise.
13161 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
13163 (thunderx_load): Split load_8 to load_16.
13164 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
13166 (thunderx2t99_storepair_basic): Split store_8 to store_16.
13167 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
13168 (xgene1_store_pair): Split store_8 to store_16.
13169 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
13170 (falkor_st_0_st_sd): Split store_8 to store_16.
13172 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
13174 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
13175 and store1/2/3/4 to store_4/8/12/16.
13176 * config/aarch64/aarch64.md: Update for rename.
13177 * config/arm/arm.md: Likewise.: Likewise.
13178 * config/arm/arm.c: Likewise.
13179 * config/arm/thumb1.md: Likewise.
13180 * config/arm/thumb2.md: Likewise.
13181 * config/arm/vfp.md: Likewise.
13182 * config/arm/arm-generic.md: Likewise.
13183 * config/arm/arm1020e.md: Likewise.
13184 * config/arm/arm1026ejs.md: Likewise.
13185 * config/arm/arm1136jfs.md: Likewise.
13186 * config/arm/arm926ejs.md: Likewise.
13187 * config/arm/cortex-a15.md: Likewise.
13188 * config/arm/cortex-a17.md: Likewise.
13189 * config/arm/cortex-a5.md: Likewise.
13190 * config/arm/cortex-a53.md: Likewise.
13191 * config/arm/cortex-a57.md: Likewise.
13192 * config/arm/cortex-a7.md: Likewise.
13193 * config/arm/cortex-a8.md: Likewise.
13194 * config/arm/cortex-a9.md: Likewise.
13195 * config/arm/cortex-m4.md: Likewise.
13196 * config/arm/cortex-m7.md: Likewise.
13197 * config/arm/cortex-r4.md: Likewise.
13198 * config/arm/exynos-m1.md: Likewise.
13199 * config/arm/fa526.md: Likewise.
13200 * config/arm/fa606te.md: Likewise.
13201 * config/arm/fa626te.md: Likewise.
13202 * config/arm/fa726te.md: Likewise.
13203 * config/arm/fmp626.md: Likewise.
13204 * config/arm/iwmmxt.md: Likewise.
13205 * config/arm/ldmstm.md: Likewise.
13206 * config/arm/marvell-pj4.md: Likewise.
13207 * config/arm/xgene1.md: Likewise.
13208 * config/aarch64/thunderx.md: Likewise.
13209 * config/aarch64/thunderx2t99.md: Likewise.
13210 * config/aarch64/falkor.md: Likewise.
13212 2017-09-12 Martin Liska <mliska@suse.cz>
13214 * attribs.c (private_lookup_attribute): New function.
13215 * attribs.h (private_lookup_attribute): Declared here.
13216 (lookup_attribute): Called from this place.
13218 2017-09-12 Richard Biener <rguenther@suse.de>
13220 PR tree-optimization/82157
13221 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
13222 stmts with side-effects.
13224 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13225 Alan Hayward <alan.hayward@arm.com>
13226 David Sherwood <david.sherwood@arm.com>
13228 * target.def (hard_regno_nregs): New hook.
13229 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
13230 * targhooks.h (default_hard_regno_nregs): Declare.
13231 * targhooks.c (default_hard_regno_nregs): New function.
13232 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
13233 (TARGET_HARD_REGNO_NREGS): ...this hook.
13234 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
13235 (CLASS_MAX_NREGS): Likewise.
13236 * doc/tm.texi: Regenerate.
13237 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
13238 instead of HARD_REGNO_NREGS.
13239 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
13240 HARD_REGNO_NREGS in the comment.
13241 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
13242 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
13243 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
13244 Return an unsigned int.
13245 (TARGET_HARD_REGNO_NREGS): Redefine.
13246 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
13247 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
13248 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
13249 (arc_hard_regno_nregs): New function.
13250 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
13251 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
13252 (arm_hard_regno_nregs): New function.
13253 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
13254 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
13255 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
13256 (TARGET_HARD_REGNO_NREGS): Redefine.
13257 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
13258 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
13259 (HARD_REGNO_NREGS): Delete.
13260 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
13261 (cr16_hard_regno_nregs): New function.
13262 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
13263 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
13264 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
13265 (cris_hard_regno_nregs): New function.
13266 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
13267 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
13268 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
13269 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
13270 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
13271 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
13272 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
13273 (frv_hard_regno_nregs): Make static. Take and return an
13275 (frv_class_max_nregs): Remove outdated copy of documentation.
13276 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
13277 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
13278 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
13279 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
13280 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
13281 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
13282 (TARGET_HARD_REGNO_NREGS): Redefine.
13283 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
13284 (CLASS_MAX_NREGS): Update comment.
13285 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
13286 (ia64_hard_regno_nregs): New function.
13287 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
13288 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
13289 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
13290 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
13291 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
13293 (m32c_hard_regno_nregs): Likewise. Make static.
13294 (TARGET_HARD_REGNO_NREGS): Redefine.
13295 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
13296 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
13297 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
13298 (m68k_hard_regno_nregs): New function.
13299 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
13300 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
13301 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
13302 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
13303 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
13304 Take and return an unsigned int.
13305 (TARGET_HARD_REGNO_NREGS): Redefine.
13306 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
13307 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
13308 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
13309 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
13310 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
13311 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
13312 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
13313 (msp430_hard_regno_nregs): Make static. Take and return an
13315 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
13316 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
13317 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
13318 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
13319 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
13320 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
13321 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
13322 (TARGET_HARD_REGNO_NREGS): Redefine.
13323 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
13324 (PA_HARD_REGNO_NREGS): ...this.
13325 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
13326 (PA_HARD_REGNO_NREGS): ...this.
13327 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
13328 (pa_hard_regno_nregs): New function.
13329 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
13330 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
13331 (pdp11_hard_regno_nregs): New function.
13332 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
13333 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
13334 (rs6000_hard_regno_nregs_hook): New function.
13335 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
13336 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
13337 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
13338 Take and return an unsigned int. Move earlier in file.
13339 (TARGET_HARD_REGNO_NREGS): Redefine.
13340 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
13341 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
13342 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
13343 (rl78_hard_regno_nregs): Make static. Take and return an
13345 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
13346 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
13347 (rs6000_hard_regno_nregs_hook): New function.
13348 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
13349 * config/rx/rx.c (rx_hard_regno_nregs): New function.
13350 (TARGET_HARD_REGNO_NREGS): Redefine.
13351 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
13352 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
13353 instead of HARD_REGNO_NREGS.
13354 (s390_hard_regno_nregs): New function.
13355 (s390_hard_regno_mode_ok): Add comment from s390.h.
13356 (TARGET_HARD_REGNO_NREGS): Redefine.
13357 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
13358 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
13359 (sh_hard_regno_nregs): New function.
13360 (sh_pass_in_reg_p): Use it.
13361 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
13362 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
13363 (sparc_hard_regno_nregs): New function.
13364 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
13365 * config/spu/spu.c (spu_hard_regno_nregs): New function.
13366 (spu_function_arg_advance): Use it, supplying a valid register number.
13367 (TARGET_HARD_REGNO_NREGS): Redefine.
13368 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
13369 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
13370 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
13371 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
13372 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
13373 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
13374 (CLASS_MAX_NREGS): Remove copy of old documentation.
13375 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
13376 (visium_hard_regno_nregs): New function.
13377 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
13378 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
13379 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
13380 xtensa_hard_regno_nregs): New function.
13381 * system.h (HARD_REGNO_NREGS): Poison.
13383 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13385 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
13386 hard_regno_nregs instead of HARD_REGNO_NREGS.
13387 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
13388 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
13389 (c6x_expand_epilogue): Likewise.
13390 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
13391 (frv_read_iacc_argument): Likewise.
13392 * config/sh/sh.c: Include regs.h.
13393 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
13394 (regs_used): Likewise.
13395 (output_stack_adjust): Likewise.
13396 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
13397 * expmed.c: Include regs.h.
13398 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
13399 * ree.c: Include regs.h.
13400 (combine_reaching_defs): Use hard_regno_nregs instead of
13402 (add_removable_extension): Likewise.
13404 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13406 * regs.h (hard_regno_nregs): Turn into a function.
13407 (end_hard_regno): Update accordingly.
13408 * caller-save.c (setup_save_areas): Likewise.
13409 (save_call_clobbered_regs): Likewise.
13410 (replace_reg_with_saved_mem): Likewise.
13411 (insert_restore): Likewise.
13412 (insert_save): Likewise.
13413 * combine.c (can_change_dest_mode): Likewise.
13414 (move_deaths): Likewise.
13415 (distribute_notes): Likewise.
13416 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
13417 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
13418 (rs6000_split_multireg_move): Likewise.
13419 (rs6000_register_move_cost): Likewise.
13420 (rs6000_memory_move_cost): Likewise.
13421 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
13422 (rs6000_split_multireg_move): Likewise.
13423 (rs6000_register_move_cost): Likewise.
13424 (rs6000_memory_move_cost): Likewise.
13425 * cselib.c (cselib_reset_table): Likewise.
13426 (cselib_lookup_1): Likewise.
13427 * emit-rtl.c (set_mode_and_regno): Likewise.
13428 * function.c (aggregate_value_p): Likewise.
13429 * ira-color.c (setup_profitable_hard_regs): Likewise.
13430 (check_hard_reg_p): Likewise.
13431 (calculate_saved_nregs): Likewise.
13432 (assign_hard_reg): Likewise.
13433 (improve_allocation): Likewise.
13434 (calculate_spill_cost): Likewise.
13435 * ira-emit.c (modify_move_list): Likewise.
13436 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
13437 (ira_hard_reg_in_set_p): Likewise.
13438 * ira.c (setup_reg_mode_hard_regset): Likewise.
13439 (clarify_prohibited_class_mode_regs): Likewise.
13440 (check_allocation): Likewise.
13441 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13442 (lra_setup_reg_renumber): Likewise.
13443 (setup_try_hard_regno_pseudos): Likewise.
13444 (spill_for): Likewise.
13445 (assign_hard_regno): Likewise.
13446 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
13447 * lra-constraints.c (in_class_p): Likewise.
13448 (lra_constraint_offset): Likewise.
13449 (simplify_operand_subreg): Likewise.
13450 (lra_constraints): Likewise.
13451 (split_reg): Likewise.
13452 (split_if_necessary): Likewise.
13453 (invariant_p): Likewise.
13454 (inherit_in_ebb): Likewise.
13455 * lra-lives.c (process_bb_lives): Likewise.
13456 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
13457 (get_hard_regs): Likewise.
13458 (do_remat): Likewise.
13459 * lra-spills.c (assign_spill_hard_regs): Likewise.
13460 * mode-switching.c (create_pre_exit): Likewise.
13461 * postreload.c (reload_combine_recognize_pattern): Likewise.
13462 * recog.c (peep2_find_free_register): Likewise.
13463 * regcprop.c (kill_value_regno): Likewise.
13464 (set_value_regno): Likewise.
13465 (copy_value): Likewise.
13466 (maybe_mode_change): Likewise.
13467 (find_oldest_value_reg): Likewise.
13468 (copyprop_hardreg_forward_1): Likewise.
13469 * regrename.c (check_new_reg_p): Likewise.
13470 (regrename_do_replace): Likewise.
13471 * reload.c (push_reload): Likewise.
13472 (combine_reloads): Likewise.
13473 (find_dummy_reload): Likewise.
13474 (operands_match_p): Likewise.
13475 (find_reloads): Likewise.
13476 (find_equiv_reg): Likewise.
13477 (reload_adjust_reg_for_mode): Likewise.
13478 * reload1.c (count_pseudo): Likewise.
13479 (count_spilled_pseudo): Likewise.
13480 (find_reg): Likewise.
13481 (clear_reload_reg_in_use): Likewise.
13482 (free_for_value_p): Likewise.
13483 (allocate_reload_reg): Likewise.
13484 (choose_reload_regs): Likewise.
13485 (reload_adjust_reg_for_temp): Likewise.
13486 (emit_reload_insns): Likewise.
13487 (delete_output_reload): Likewise.
13488 * rtlanal.c (subreg_get_info): Likewise.
13489 * sched-deps.c (sched_analyze_reg): Likewise.
13490 * sel-sched.c (init_regs_for_mode): Likewise.
13491 (mark_unavailable_hard_regs): Likewise.
13492 (choose_best_reg_1): Likewise.
13493 (verify_target_availability): Likewise.
13494 * valtrack.c (dead_debug_insert_temp): Likewise.
13495 * var-tracking.c (track_loc_p): Likewise.
13496 (emit_note_insn_var_location): Likewise.
13497 * varasm.c (make_decl_rtl): Likewise.
13498 * reginfo.c (choose_hard_reg_mode): Likewise.
13499 (init_reg_modes_target): Refer directly to
13500 this_target_regs->x_hard_regno_nregs.
13502 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13504 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
13505 instead of hard_regno_nregs.
13507 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13509 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
13510 end_hard_regno instead of hard_regno_nregs.
13511 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
13512 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
13513 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
13514 * ira-color.c (improve_allocation): Likewise.
13515 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13516 * lra-lives.c (mark_regno_live): Likewise.
13517 (mark_regno_dead): Likewise.
13518 * lra-remat.c (operand_to_remat): Likewise.
13519 * lra.c (collect_non_operand_hard_regs): Likewise.
13520 * postreload.c (reload_combine_note_store): Likewise.
13521 (move2add_valid_value_p): Likewise.
13522 * reload.c (regno_clobbered_p): Likewise.
13524 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13526 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
13528 * config/v850/v850.c (v850_reorg): Likewise.
13529 * reload.c (refers_to_regno_for_reload_p): Likewise.
13530 (find_equiv_reg): Likewise.
13531 * reload1.c (reload_reg_reaches_end_p): Likewise.
13533 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13535 * caller-save.c (add_used_regs): Use REG_NREGS instead of
13537 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
13538 * config/arm/arm.c (output_move_neon): Likewise.
13539 (arm_attr_length_move_neon): Likewise.
13540 (neon_split_vcombine): Likewise.
13541 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
13542 (c6x_mark_reg_written): Likewise.
13543 (c6x_dwarf_register_span): Likewise.
13544 * config/i386/i386.c (ix86_save_reg): Likewise.
13545 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
13546 (rws_access_reg): Likewise.
13547 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
13548 * mode-switching.c (create_pre_exit): Likewise.
13549 * ree.c (combine_reaching_defs): Likewise.
13550 (add_removable_extension): Likewise.
13551 * regcprop.c (find_oldest_value_reg): Likewise.
13552 (copyprop_hardreg_forward_1): Likewise.
13553 * reload.c (reload_inner_reg_of_subreg): Likewise.
13554 (push_reload): Likewise.
13555 (combine_reloads): Likewise.
13556 (find_dummy_reload): Likewise.
13557 (reload_adjust_reg_for_mode): Likewise.
13558 * reload1.c (find_reload_regs): Likewise.
13559 (forget_old_reloads_1): Likewise.
13560 (reload_reg_free_for_value_p): Likewise.
13561 (reload_adjust_reg_for_temp): Likewise.
13562 (emit_reload_insns): Likewise.
13563 (delete_output_reload): Likewise.
13564 * sel-sched.c (choose_best_reg_1): Likewise.
13565 (choose_best_pseudo_reg): Likewise.
13567 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13568 Alan Hayward <alan.hayward@arm.com>
13569 David Sherwood <david.sherwood@arm.com>
13571 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
13572 * target.def (slow_unaligned_access): New hook.
13573 * targhooks.h (default_slow_unaligned_access): Declare.
13574 * targhooks.c (default_slow_unaligned_access): New function.
13575 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
13576 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
13577 * doc/tm.texi: Regenerate.
13578 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
13579 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
13580 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
13582 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
13583 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
13585 (rs6000_slow_unaligned_access): New function.
13586 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13587 (expand_block_compare): Likewise.
13588 (expand_strn_compare): Likewise.
13589 (rs6000_rtx_costs): Likewise.
13590 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
13591 (riscv_slow_unaligned_access): Likewise.
13592 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
13593 (riscv_slow_unaligned_access_p): ...this and make static.
13594 (riscv_option_override): Update accordingly.
13595 (riscv_slow_unaligned_access): New function.
13596 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13597 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
13598 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13599 (rs6000_slow_unaligned_access): New function.
13600 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13601 (rs6000_rtx_costs): Likewise.
13602 * config/rs6000/rs6000-string.c (expand_block_compare)
13603 (expand_strn_compare): Use targetm.slow_unaligned_access instead
13604 of SLOW_UNALIGNED_ACCESS.
13605 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
13606 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
13607 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
13608 of SLOW_UNALIGNED_ACCESS.
13609 * expmed.c (simple_mem_bitfield_p): Likewise.
13610 * expr.c (alignment_for_piecewise_move): Likewise.
13611 (emit_group_load_1): Likewise.
13612 (emit_group_store): Likewise.
13613 (copy_blkmode_from_reg): Likewise.
13614 (emit_push_insn): Likewise.
13615 (expand_assignment): Likewise.
13616 (store_field): Likewise.
13617 (expand_expr_real_1): Likewise.
13618 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
13619 * lra-constraints.c (simplify_operand_subreg): Likewise.
13620 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13621 * gimple-ssa-store-merging.c: Likewise in block comment at start
13623 * tree-ssa-strlen.c: Include target.h.
13624 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
13625 of SLOW_UNALIGNED_ACCESS.
13626 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
13628 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13630 PR rtl-optimization/82185
13631 * expmed.c (emit_store_flag_int): Only test tem if it has been
13634 2017-09-12 Richard Biener <rguenther@suse.de>
13636 PR middle-end/82149
13637 * match.pd ((FTYPE) N CMP CST): Fix typo.
13639 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
13641 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
13643 (mips_near_type_p): Add 'short_call' attribute as a synonym
13645 * doc/extend.texi (short_call): Document new function attribute.
13647 2017-09-12 Jakub Jelinek <jakub@redhat.com>
13650 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
13651 assertion check that in the condition.
13652 (get_atomic_generic_size): Likewise. Before testing if parameter
13653 has pointer type, if it has array type, call for C++
13654 default_conversion to perform array-to-pointer conversion.
13656 2017-09-12 Richard Biener <rguenther@suse.de>
13658 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
13659 for operations we cannot scalarize.
13661 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
13663 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
13664 vectors heap vectors. Clean up comments.
13665 Make visited_bbs a reference.
13666 (profitable_jump_thread_path): Make GC
13667 vectors heap vectors. Clean up comments.
13669 (convert_and_register_jump_thread_path): Make GC vectors heap
13671 (check_subpath_and_update_thread_path): Same. Clean up comments.
13672 Make visited_bbs a reference.
13673 (handle_phi): Abstract common code to to
13674 register_jump_thread_path_if_profitable.
13675 Rename VAR_BB to DEF_BB.
13677 Make GC vectors heap vectors.
13678 Make visited_bbs a reference.
13679 (handle_assignment): Same.
13680 (register_jump_thread_path_if_profitable): New.
13681 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
13683 Make GC vectors heap vectors. Clean up comments.
13684 Make visited_bbs a reference.
13685 (find_jump_threads_backwards): Make visited_bbs live in the stack.
13686 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
13689 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
13692 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
13693 words of E_DImode object are reachable by xtensa_uimm8x4 access.
13695 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
13697 Revert r251800 and r251799.
13699 2017-09-11 Martin Jambor <mjambor@suse.cz>
13702 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
13703 arguments in advance.
13704 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
13705 use it to find predecessor edges.
13706 (naive_outof_ssa): Collect vector of predecessors.
13708 2017-09-08 Jason Merrill <jason@redhat.com>
13710 PR c++/70029 - ICE with ref-qualifier and -flto
13711 * langhooks.h (struct lang_hooks_for_types): Add
13712 copy_lang_qualifiers.
13713 * attribs.c (build_type_attribute_qual_variant): Use it.
13714 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
13716 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
13717 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
13719 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
13722 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
13723 (*mulsi3_sp64): New instruction.
13724 (mulsi3): New expander.
13726 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
13728 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
13730 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13732 * sancov.c: Include memmodel.h.
13734 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
13737 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
13740 2017-09-07 Carl Love <cel@us.ibm.com>
13742 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
13743 the sldi instruction.
13745 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
13747 * sancov.c: Include tm_p.h.
13749 2017-09-07 Jakub Jelinek <jakub@redhat.com>
13752 * output.h (switch_to_other_text_partition): New declaration.
13753 * varasm.c (switch_to_other_text_partition): New function.
13754 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
13755 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
13756 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
13757 to the other text partition before emitting LCL label and switch back
13758 after emitting the word after it.
13760 2017-09-07 Richard Biener <rguenther@suse.de>
13762 * passes.def (pass_split_crit_edges): Remove instance before PRE.
13763 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
13764 critical edges here, after loop init.
13765 (pass_data_pre): Remove PROP_no_crit_edges flags.
13766 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
13767 for valueization of call args to avoid leaking VN_TOP.
13768 (visit_use): Assert we do not visit default defs.
13769 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
13770 Use error_mark_node to more easily detect leaking VN_TOP.
13771 All default-defs are varying, not VN_TOP. Mark them visited.
13772 (run_scc_vn): Make code match comment.
13774 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13776 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
13777 OPTION_MASK_FLOAT128_KEYWORD.
13778 (POWERPC_MASKS): Likewise.
13779 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
13780 support for the -mfloat128-type option, and make -mfloat128
13781 default on PowerPC Linux systems. Define or undefine
13782 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
13783 Define __float128 to be __ieee128 if IEEE 128-bit support is
13784 enabled, or undefine it.
13785 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
13786 Delete defining __FLOAT128_TYPE__.
13787 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
13788 -mfloat128-type option and make -mfloat128 default on PowerPC
13790 (TARGET_FLOAT128_TYPE): Likewise.
13791 (-mfloat128-type): Likewise.
13792 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13793 Delete the -mfloat128-type option and make -mfloat128 default on
13794 PowerPC Linux systems. Always use __ieee128 to be the keyword for
13795 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
13796 128-bit floating point is enabled. Change tests from using
13797 -mfloat128-type to -mfloat128.
13798 (rs6000_mangle_type): Use the correct mangling for the __float128
13799 type even if normal long double is restricted to 64-bits.
13800 (floatn_mode): Enable the _Float128 type by default on VSX Linux
13802 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
13803 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
13804 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
13806 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
13807 documentation for -mfloat128.
13809 2017-09-06 Olivier Hainque <hainque@adacore.com>
13811 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
13813 2017-09-06 Wish Wu <wishwu007@gmail.com>
13814 Jakub Jelinek <jakub@redhat.com>
13816 * asan.c (initialize_sanitizer_builtins): Add
13817 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
13818 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
13819 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
13820 BT_FN_VOID_UINT64_PTR variables.
13821 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
13822 (BT_FN_VOID_UINT16_UINT16): Likewise.
13823 (BT_FN_VOID_UINT32_UINT32): Likewise.
13824 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
13825 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
13826 (BT_FN_VOID_UINT64_PTR): Likewise.
13827 * common.opt (flag_sanitize_coverage): New variable.
13828 (fsanitize-coverage=trace-pc): Remove.
13829 (fsanitize-coverage=): Add.
13830 * flag-types.h (enum sanitize_coverage_code): New enum.
13831 * fold-const.c (fold_range_test): Disable non-short-circuit
13832 optimization if flag_sanitize_coverage.
13833 (fold_truth_andor): Likewise.
13834 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
13835 * opts.c (COVERAGE_SANITIZER_OPT): Define.
13836 (coverage_sanitizer_opts): New array.
13837 (get_closest_sanitizer_option): Add OPTS argument, handle also
13838 OPT_fsanitize_coverage_.
13839 (parse_sanitizer_options): Adjusted to also handle
13840 OPT_fsanitize_coverage_.
13841 (common_handle_option): Add OPT_fsanitize_coverage_.
13842 * sancov.c (instrument_comparison, instrument_switch): New function.
13843 (sancov_pass): Add trace-cmp support.
13844 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
13845 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
13846 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
13847 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
13848 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
13849 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
13850 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
13851 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
13852 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
13853 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
13855 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13857 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13858 error. Only quit immediately if parsing is complete.
13859 (BEGIN): Initialize fatal_err and parse_done.
13860 (begin fpu, end fpu): Check number of arguments.
13861 (begin arch, end arch): Likewise.
13862 (begin cpu, end cpu): Likewise.
13863 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13864 (optalias): Likewise.
13866 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13868 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13869 * config/arm/arm-isa.h: Delete. Move definitions to ...
13870 * arm-cpus.in: ... here. Use new feature and fgroup values.
13871 * config/arm/arm.c (arm_option_override): Use lower case for feature
13873 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13874 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13875 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13877 (print_isa_bits_for): New function.
13878 (gen_isa): New function.
13879 (gen_comm_data): Use print_isa_bits_for.
13880 (define feature): New keyword.
13881 (define fgroup): New keyword.
13882 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
13883 (arm-isa.h): Add rule to generate file.
13884 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13885 case for feature bit names.
13887 2017-09-06 Richard Biener <rguenther@suse.de>
13889 * tree-ssa-pre.c (NECESSARY): Remove.
13890 (create_expression_by_pieces): Do not touch pass-local flags.
13891 (insert_into_preds_of_block): Likewise.
13892 (do_pre_regular_insertion): Likewise.
13893 (eliminate_insert): Likewise.
13894 (eliminate_dom_walker::before_dom_children): Likewise.
13895 (fini_eliminate): Do not look at inserted_exprs.
13896 (mark_operand_necessary): Remove.
13897 (remove_dead_inserted_code): Replace with simple work-list
13898 algorithm based on inserted_exprs and SSA uses.
13899 (pass_pre::execute): Re-order fini_eliminate and
13900 remove_dead_inserted_code.
13902 2017-09-06 Olivier Hainque <hainque@adacore.com>
13904 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
13905 for VxWorks 7. Adjust surrounding comments.
13907 2017-09-06 Richard Biener <rguenther@suse.de>
13909 * gimple-ssa-strength-reduction.c
13910 (find_candidates_dom_walker::before_dom_children): Also allow
13913 2017-09-06 Richard Biener <rguenther@suse.de>
13915 PR tree-optimization/82108
13916 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
13917 for gap in the non-permutation SLP case.
13919 2017-09-06 Martin Jambor <mjambor@suse.cz>
13921 PR tree-optimization/82078
13922 * tree-sra.c (sort_and_splice_var_accesses): Move call to
13923 add_access_to_work_queue...
13924 (build_accesses_from_assign): ...here.
13925 (propagate_all_subaccesses): Make sure racc is the group
13926 representative, if there is one.
13928 2017-09-06 Jakub Jelinek <jakub@redhat.com>
13930 PR middle-end/82095
13931 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
13934 2017-09-06 Richard Biener <rguenther@suse.de>
13936 * gimple-ssa-strength-reduction.c
13937 (find_candidates_dom_walker::before_doom_children): Use a
13938 type and not a mode check.
13940 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
13943 * config/arm/predicates.md (arm_general_adddi_operand): Create new
13945 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
13947 2017-09-05 Jeff Law <law@redhat.com>
13949 PR tree-optimization/64910
13950 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
13951 cases where we have 3 or more operands.
13953 2017-09-05 Jakub Jelinek <jakub@redhat.com>
13955 PR middle-end/81768
13956 * omp-low.c (lower_omp_for): Recompute tree invariant if
13957 gimple_omp_for_initial/final is ADDR_EXPR.
13959 PR middle-end/81768
13960 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
13961 into gimple val before gimplification fo the COND_EXPR.
13963 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
13965 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
13966 REGION_COPY argument.
13967 (thread_through_all_blocks): Remove unused argument to
13968 duplicate_thread_path.
13970 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13971 Alan Hayward <alan.hayward@arm.com>
13972 David Sherwood <david.sherwood@arm.com>
13974 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
13975 Take a scalar_mode rather than a machine_mode.
13976 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13977 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
13978 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13979 (aarch64_gen_adjusted_ldpstp): Likewise.
13980 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
13982 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13983 Alan Hayward <alan.hayward@arm.com>
13984 David Sherwood <david.sherwood@arm.com>
13986 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
13987 Take a scalar_int_mode instead of a machine_mode.
13988 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13989 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13990 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13991 (aarch64_simd_attr_length_rglist): Delete.
13992 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
13993 a scalar_int_mode instead of a machine_mode.
13994 (aarch64_add_offset): Likewise.
13995 (aarch64_internal_mov_immediate): Likewise
13996 (aarch64_add_constant_internal): Likewise.
13997 (aarch64_add_constant): Likewise.
13998 (aarch64_movw_imm): Likewise.
13999 (aarch64_rtx_arith_op_extract_p): Likewise.
14000 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
14001 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
14002 Remove assert that the mode isn't a vector.
14003 (aarch64_output_scalar_simd_mov_immediate): Likewise.
14004 (aarch64_expand_mov_immediate): Update calls after above changes.
14005 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
14006 (aarch64_and_bitmask_imm): Check for scalar integer modes.
14007 (aarch64_move_imm): Likewise.
14008 (aarch64_can_const_movi_rtx_p): Likewise.
14009 (aarch64_strip_extend): Likewise.
14010 (aarch64_extr_rtx_p): Likewise.
14011 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
14012 a CONST_INT when the mode parameter is VOIDmode.
14013 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
14015 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14017 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
14018 * stor-layout.c (bitwise_mode_for_mode): Likewise.
14019 (bitwise_type_for_mode): Update accordingly.
14021 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14023 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
14024 * stor-layout.c (mode_for_size_tree): Likewise.
14025 (mode_for_array): Update accordingly.
14026 (layout_decl): Likewise.
14027 (compute_record_mode): Likewise. Only set the mode once.
14029 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14031 * target.def (get_mask_mode): Change return type to opt_mode.
14033 * doc/tm.texi: Regenerate.
14034 * targhooks.h (default_get_mask_mode): Return an opt_mode.
14035 * targhooks.c (default_get_mask_mode): Likewise.
14036 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
14037 * optabs-query.c (can_vec_mask_load_store_p): Update use of
14038 targetm.get_mask_mode.
14039 * tree.c (build_truth_vector_type): Likewise.
14041 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14043 * machmode.h (mode_for_vector): Return an opt_mode.
14044 * stor-layout.c (mode_for_vector): Likewise.
14045 (mode_for_int_vector): Update accordingly.
14046 (layout_type): Likewise.
14047 * config/i386/i386.c (emit_memmov): Likewise.
14048 (ix86_expand_set_or_movmem): Likewise.
14049 (ix86_expand_vector_init): Likewise.
14050 (ix86_get_mask_mode): Likewise.
14051 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
14053 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
14054 * expmed.c (extract_bit_field_1): Likewise.
14055 * expr.c (expand_expr_real_2): Likewise.
14056 * optabs-query.c (can_vec_perm_p): Likewise.
14057 (can_vec_mask_load_store_p): Likewise.
14058 * optabs.c (expand_vec_perm): Likewise.
14059 * targhooks.c (default_get_mask_mode): Likewise.
14060 * tree-vect-stmts.c (vectorizable_store): Likewise.
14061 (vectorizable_load): Likewise.
14062 (get_vectype_for_scalar_type_and_size): Likewise.
14064 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14066 * machmode.h (mode_for_int_vector): New function.
14067 * stor-layout.c (mode_for_int_vector): Likewise.
14068 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
14069 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
14070 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
14071 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
14072 (s390_expand_vcond): Likewise.
14074 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14076 * machmode.h (opt_machine_mode): New type.
14077 (opt_mode<T>): Allow construction from anything that can be
14079 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
14080 (mode_for_size): Return an opt_machine_mode.
14081 * stor-layout.c (mode_for_size): Likewise.
14082 (mode_for_size_tree): Update call accordingly.
14083 (bitwise_mode_for_mode): Likewise.
14084 (make_fract_type): Likewise.
14085 (make_accum_type): Likewise.
14086 * caller-save.c (replace_reg_with_saved_mem): Update call
14088 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14089 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14090 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14091 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14092 * expmed.c (extract_bit_field_1): Likewise.
14093 * reload.c (get_secondary_mem): Likewise.
14094 * varasm.c (assemble_integer): Likewise.
14095 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
14098 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14100 * machmode.h (decimal_float_mode_for_size): New function.
14101 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
14102 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
14103 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
14104 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
14105 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
14106 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
14108 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14110 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
14111 (get_builtin_sync_mode): Likewise.
14112 (expand_ifn_atomic_compare_exchange): Likewise.
14113 (expand_builtin_atomic_clear): Likewise.
14114 (expand_builtin_atomic_test_and_set): Likewise.
14115 (fold_builtin_atomic_always_lock_free): Likewise.
14116 * calls.c (compute_argument_addresses): Likewise.
14117 (emit_library_call_value_1): Likewise.
14118 (store_one_arg): Likewise.
14119 * combine.c (combine_instructions): Likewise.
14120 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
14121 * config/arm/arm.c (arm_function_value): Likewise.
14122 (aapcs_allocate_return_reg): Likewise.
14123 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
14124 * config/i386/i386.c (construct_container): Likewise.
14125 (ix86_gimplify_va_arg): Likewise.
14126 (ix86_expand_sse_cmp): Likewise.
14127 (emit_memmov): Likewise.
14128 (emit_memset): Likewise.
14129 (expand_small_movmem_or_setmem): Likewise.
14130 (ix86_expand_pextr): Likewise.
14131 (ix86_expand_pinsr): Likewise.
14132 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
14133 * config/microblaze/microblaze.c (microblaze_block_move_straight):
14135 * config/mips/mips.c (mips_function_value_1) Likewise.
14136 (mips_block_move_straight): Likewise.
14137 (mips_expand_ins_as_unaligned_store): Likewise.
14138 * config/powerpcspe/powerpcspe.c
14139 (rs6000_darwin64_record_arg_advance_flush): Likewise.
14140 (rs6000_darwin64_record_arg_flush): Likewise.
14141 * config/rs6000/rs6000.c
14142 (rs6000_darwin64_record_arg_advance_flush): Likewise.
14143 (rs6000_darwin64_record_arg_flush): Likewise.
14144 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
14145 (sparc_function_value_1): Likewise.
14146 * config/spu/spu.c (adjust_operand): Likewise.
14147 (spu_emit_branch_or_set): Likewise.
14148 (arith_immediate_p): Likewise.
14149 * emit-rtl.c (gen_lowpart_common): Likewise.
14150 * expr.c (expand_expr_real_1): Likewise.
14151 * function.c (assign_parm_setup_block): Likewise.
14152 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
14153 * reload1.c (alter_reg): Likewise.
14154 * stor-layout.c (mode_for_vector): Likewise.
14155 (layout_type): Likewise.
14157 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14159 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
14160 (spu_convert_move): Likewise.
14161 * lower-subreg.c (resolve_simple_move): Likewise.
14163 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14166 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
14167 define_insn to a define_expand.
14168 (altivec_vsum2sws_direct): New define_insn.
14169 (altivec_vsumsws): Convert from a define_insn to a define_expand.
14171 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
14173 * config/arm/arm.c (arm_option_params_internal): Improve setting of
14176 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
14181 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
14182 realign stack if stack alignment needed is less than incoming
14185 2017-09-05 Marek Polacek <polacek@redhat.com>
14188 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
14191 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
14193 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
14195 2017-09-05 Richard Biener <rguenther@suse.de>
14197 PR tree-optimization/82084
14198 * fold-const.c (can_native_encode_string_p): Handle wide characters.
14200 2017-09-05 Richard Biener <rguenther@suse.de>
14202 PR tree-optimization/82102
14203 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
14205 2017-09-05 Martin Liska <mliska@suse.cz>
14207 PR tree-optimization/82032
14208 * tree-cfg.c (generate_range_test): New function.
14209 * tree-cfg.h (generate_range_test): Declared here.
14210 * tree-cfgcleanup.c (convert_single_case_switch): New function.
14211 (cleanup_control_expr_graph): Use it.
14212 * tree-switch-conversion.c (try_switch_expansion): Remove
14214 (emit_case_nodes): Use generate_range_test.
14216 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
14219 * config/i386/i386.md (*<btsc><mode>_mask): Add
14220 TARGET_USE_BT to insn constraint.
14221 (*btr<mode>_mask): Ditto.
14223 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
14225 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
14226 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
14228 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
14231 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
14232 TARGET_NEON and TARGET_IWMMXT.
14233 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
14234 TARGET_NEON and TARGET_IWMMXT.
14235 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
14237 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
14239 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
14240 (ix86_rewrite_tls_address): Ditto.
14241 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
14242 (ix86_rewrite_tls_address_1): Ditto.
14243 (ix86_rewrite_tls_address): Ditto.
14244 * config/i386/predicates.md (tls_address_pattern): New predicate.
14245 * config/i386/i386.md (TLS address splitter): New splitter.
14247 2017-09-04 Richard Biener <rguenther@suse.de>
14249 PR tree-optimization/82084
14250 * fold-const.h (can_native_encode_string_p): Declare.
14251 * fold-const.c (can_native_encode_string_p): Factor out from ...
14252 (native_encode_string): ... here.
14253 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
14254 vectorizing stores from constants we later cannot handle.
14256 2017-09-04 Marek Polacek <polacek@redhat.com>
14259 * doc/invoke.texi: Update -Wtautological-compare documentation.
14261 2017-09-04 Jeff Law <law@redhat.com>
14263 PR tree-optimization/64910
14264 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
14265 swap the first and last operand if the last is a constant.
14267 2017-09-04 Marek Polacek <polacek@redhat.com>
14270 * convert.c (do_narrow): When sanitizing signed integer overflows,
14271 bail out for signed types.
14272 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
14274 2017-09-04 Richard Biener <rguenther@suse.de>
14276 PR tree-optimization/82060
14277 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
14278 Move devirtualization after stmt folding and before EH/AB/noreturn
14279 cleanup to get the stmt refs canonicalized. Use a bool instead
14280 of gimple_modified_p since that doesn't work for NOPs. Schedule
14281 NOPs generated by folding for removal.
14283 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14284 Alan Hayward <alan.hayward@arm.com>
14285 David Sherwood <david.sherwood@arm.com>
14287 * coretypes.h (pad_direction): New enum.
14288 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
14289 (FUNCTION_ARG_PADDING): Likewise.
14290 * target.def (function_arg_padding): New hook.
14291 * targhooks.h (default_function_arg_padding): Declare.
14292 * targhooks.c (default_function_arg_padding): New function.
14293 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
14294 (TARGET_FUNCTION_ARG_PADDING): ...this.
14295 * doc/tm.texi: Regenerate.
14296 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
14297 instead of direction.
14298 (compute_argument_addresses): Likewise.
14299 (load_register_parameters): Likewise.
14300 (emit_library_call_value_1): Likewise.
14301 (store_one_arg): Use targetm.calls.function_arg_padding instead
14302 of FUNCTION_ARG_PADDING.
14303 (must_pass_in_stack_var_size_or_pad): Likewise.
14304 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
14305 (emit_group_store): Likewise.
14306 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
14307 instead of FUNCTION_ARG_PADDING.
14308 (emit_push_insn): Likewise, and propagate enum change throughout
14310 * function.h (direction): Delete.
14311 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
14313 * function.c (assign_parm_find_stack_rtl): Likewise.
14314 (assign_parm_setup_block_p): Likewise.
14315 (assign_parm_setup_block): Likewise.
14316 (gimplify_parameters): Likewise.
14317 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
14318 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
14320 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
14321 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
14322 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
14323 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
14324 (aarch64_function_arg_padding): ...this new function.
14325 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
14326 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14327 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
14328 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
14329 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
14330 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14331 (arm_pad_arg_upward): Replace with...
14332 (arm_function_arg_padding): ...this new function.
14333 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
14335 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
14336 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
14337 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14338 (ia64_hpux_function_arg_padding): Replace with...
14339 (ia64_function_arg_padding): ...this new function. Use pad_direction
14340 instead of direction. Check for TARGET_HPUX.
14341 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
14342 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14343 (iq2000_function_arg_padding): New function.
14344 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
14345 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
14346 (mips_function_arg_padding): ...this new function.
14347 (mips_pad_reg_upward): Update accordingly.
14348 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14349 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
14350 targetm.calls.function_arg_padding.
14351 (FUNCTION_ARG_PADDING): Delete.
14352 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
14353 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
14354 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
14355 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
14356 (nios2_block_reg_padding): Return pad_direction instead of direction.
14357 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
14358 instead of direction.
14359 (nios2_function_arg_padding): Likewise. Make static.
14360 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14361 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
14362 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
14363 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
14364 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14365 (pa_function_arg_padding): Make static. Return pad_direction instead
14367 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
14368 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
14369 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
14370 instead of direction. Use targetm.calls.function_arg_padding.
14371 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
14372 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
14373 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
14374 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
14375 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
14377 (function_arg_padding): Rename to...
14378 (rs6000_function_arg_padding): ...this. Make static. Return
14379 pad_direction instead of direction.
14380 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
14381 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
14382 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
14383 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
14384 instead of direction. Use targetm.calls.function_arg_padding.
14385 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
14386 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
14387 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
14388 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
14389 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14390 (function_arg_padding): Rename to...
14391 (rs6000_function_arg_padding): ...this. Make static. Return
14392 pad_direction instead of direction.
14393 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
14394 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
14395 * config/s390/s390.c (s390_function_arg_padding): New function.
14396 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14397 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
14398 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
14399 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14400 (function_arg_padding): Rename to...
14401 (sparc_function_arg_padding): ...this. Make static. Return
14402 pad_direction instead of direction.
14403 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
14404 * config/spu/spu.c (spu_function_arg_padding): New function.
14405 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14406 * system.h (FUNCTION_ARG_PADDING): Poison.
14408 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14409 Alan Hayward <alan.hayward@arm.com>
14410 David Sherwood <david.sherwood@arm.com>
14412 * target.def (modes_tieable_p): New hook.
14413 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
14414 (TARGET_MODES_TIEABLE_P): ...this.
14415 * doc/tm.texi.in: Regenerate.
14416 * hooks.h (hook_bool_mode_mode_true): Declare.
14417 * hooks.c (hook_bool_mode_mode_true): New function.
14418 * combine.c (subst): Use targetm.modes_tieable_p instead of
14420 * dse.c (find_shift_sequence): Likewise.
14421 * expmed.c (extract_low_bits): Likewise.
14422 * lower-subreg.c: Include target.h.
14423 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
14425 * rtlanal.c (rtx_cost): Likewise.
14426 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
14427 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
14428 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
14429 (TARGET_MODES_TIEABLE_P): Redefine.
14430 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
14431 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
14432 (TARGET_MODES_TIEABLE_P): Redefine.
14433 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
14434 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
14435 (arc_modes_tieable_p): New function.
14436 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
14437 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
14438 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
14439 (arm_modes_tieable_p): Make static.
14440 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
14441 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
14442 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
14443 (TARGET_MODES_TIEABLE_P): Redefine.
14444 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
14445 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
14446 (TARGET_MODES_TIEABLE_P): Redefine.
14447 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
14448 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
14449 (cr16_modes_tieable_p): New function.
14450 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
14451 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
14452 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
14453 (TRULY_NOOP_TRUNCATION): Update comment.
14454 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
14455 (TRULY_NOOP_TRUNCATION): Update comment.
14456 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
14457 (frv_modes_tieable_p): New function.
14458 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
14459 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
14460 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
14461 (TARGET_MODES_TIEABLE_P): Redefine.
14462 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
14463 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
14464 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
14465 (TARGET_MODES_TIEABLE_P): Redefine.
14466 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
14467 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
14468 (ia64_modes_tieable_p): New function.
14469 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
14470 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
14471 (iq2000_modes_tieable_p): New function.
14472 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
14473 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
14474 (lm32_modes_tieable_p): New function.
14475 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
14476 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
14477 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
14478 (TARGET_MODES_TIEABLE_P): Redefine.
14479 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
14480 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
14481 (m32r_modes_tieable_p): New function.
14482 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
14483 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
14484 (m68k_modes_tieable_p): New function.
14485 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
14486 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
14487 (mcore_modes_tieable_p): New function.
14488 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
14489 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
14491 (TARGET_MODES_TIEABLE_P): Redefine.
14492 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
14493 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
14494 * config/mips/mips.c (mips_modes_tieable_p): Make static.
14495 (TARGET_MODES_TIEABLE_P): Redefine.
14496 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
14497 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
14498 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
14499 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
14500 (mn10300_modes_tieable_p): ...this and make static.
14501 (TARGET_MODES_TIEABLE_P): Redefine.
14502 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
14503 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
14504 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
14505 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
14506 (msp430_modes_tieable_p): Make static.
14507 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
14508 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
14509 (TARGET_MODES_TIEABLE_P): Redefine.
14510 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
14511 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
14512 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
14513 (TARGET_MODES_TIEABLE_P): Redefine.
14514 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
14515 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
14516 * config/pa/pa.c (pa_modes_tieable_p): Make static.
14517 (TARGET_MODES_TIEABLE_P): Redefine.
14518 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
14519 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
14520 (pdp11_modes_tieable_p): New function.
14521 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
14522 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
14523 (rs6000_modes_tieable_p): New function.
14524 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
14525 * config/powerpcspe/powerpcspe.md: Update comment.
14526 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
14527 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
14528 (TARGET_MODES_TIEABLE_P): Redefine.
14529 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
14530 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
14531 (rl78_modes_tieable_p): New function.
14532 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
14533 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
14534 (rs6000_modes_tieable_p): New function.
14535 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
14536 * config/rs6000/rs6000.md: Update comment.
14537 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
14538 * config/rx/rx.c (rx_modes_tieable_p): New function.
14539 (TARGET_MODES_TIEABLE_P): Redefine.
14540 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
14541 * config/s390/s390.c (s390_modes_tieable_p): New function.
14542 (TARGET_MODES_TIEABLE_P): Redefine.
14543 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
14544 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
14545 (sh_modes_tieable_p): New function.
14546 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
14547 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
14548 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
14549 (sparc_modes_tieable_p): Make static.
14550 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
14551 * config/spu/spu.c (spu_modes_tieable_p): New function.
14552 (TARGET_MODES_TIEABLE_P): Redefine.
14553 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
14554 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
14555 (TARGET_MODES_TIEABLE_P): Redefine.
14556 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
14557 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
14558 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
14559 * config/v850/v850.c (v850_modes_tieable_p): New function.
14560 (TARGET_MODES_TIEABLE_P): Redefine.
14561 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
14562 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
14563 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
14564 (visium_modes_tieable_p): New function.
14565 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
14566 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
14567 (xtensa_modes_tieable_p): New function.
14568 * system.h (MODES_TIEABLE_P): Poison.
14570 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14571 Alan Hayward <alan.hayward@arm.com>
14572 David Sherwood <david.sherwood@arm.com>
14574 * target.def (hard_regno_mode_ok): New hook.
14575 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
14576 (TARGET_HARD_REGNO_MODE_OK): ...this.
14577 * doc/tm.texi.in: Regenerate.
14578 * hooks.h (hook_bool_uint_mode_true): Declare.
14579 * hooks.c (hook_bool_uint_mode_true): New function.
14580 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
14581 HARD_REGNO_MODE_OK.
14582 * genpreds.c (write_insn_preds_c): Add an include of target.h.
14583 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
14584 instead of HARD_REGNO_MODE_OK.
14585 * caller-save.c: Include target.h.
14586 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
14587 HARD_REGNO_MODE_OK.
14588 * combine.c (can_combine_p): Likewise.
14589 (combinable_i3pat): Likewise.
14590 (can_change_dest_mode): Likewise.
14591 * expr.c (init_expr_target): Likewise.
14592 (convert_move): Likewise.
14593 (convert_modes): Likewise.
14594 * ira.c (setup_prohibited_class_mode_regs): Likewise.
14595 (setup_prohibited_mode_move_regs): Likewise.
14596 * ira.h (target_ira): Likewise.
14597 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14598 * lra-constraints.c (process_alt_operands): Likewise.
14599 (split_reg): Likewise.
14600 * recog.c (peep2_find_free_register): Likewise.
14601 * ree.c (combine_reaching_defs): Likewise.
14602 * regcprop.c (maybe_mode_change): Likewise.
14603 * reginfo.c (init_reg_sets_1): Likewise.
14604 (choose_hard_reg_mode): Likewise.
14605 (simplifiable_subregs): Likewise.
14606 * regrename.c (check_new_reg_p): Likewise.
14607 * reload.c (find_valid_class): Likewise.
14608 (find_valid_class_1): Likewise.
14609 (reload_inner_reg_of_subreg): Likewise.
14610 (push_reload): Likewise.
14611 (combine_reloads): Likewise.
14612 (find_dummy_reload): Likewise.
14613 (find_reloads): Likewise.
14614 * reload1.c (find_reg): Likewise.
14615 (set_reload_reg): Likewise.
14616 (allocate_reload_reg): Likewise.
14617 (choose_reload_regs): Likewise.
14618 (reload_adjust_reg_for_temp): Likewise.
14619 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
14620 (simplify_subreg_regno): Likewise.
14621 * sel-sched.c (init_regs_for_mode): Likewise.
14622 * varasm.c (make_decl_rtl): Likewise.
14623 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
14624 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
14625 HARD_REGNO_MODE_OK.
14626 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
14627 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
14628 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14629 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
14630 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
14631 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14632 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
14633 (arc_mode_class): Delete.
14634 (HARD_REGNO_MODE_OK): Delete.
14635 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14636 (arc_hard_regno_mode_ok): Rename old array to...
14637 (arc_hard_regno_mode_ok_modes): ...this.
14638 (arc_conditional_register_usage): Update accordingly.
14639 (arc_mode_class): Make static.
14640 (arc_hard_regno_mode_ok): New function.
14641 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
14642 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
14643 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14644 (arm_hard_regno_mode_ok): Make static.
14645 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
14646 HARD_REGNO_MODE_OK.
14647 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
14648 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
14649 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
14651 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14652 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
14653 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
14654 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
14655 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
14656 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14657 * config/bfin/predicates.md (valid_reg_operand): Use
14658 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
14659 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
14660 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
14661 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14662 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
14663 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
14664 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14665 (cr16_hard_regno_mode_ok): Make static and return a bool.
14666 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
14667 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14668 (cris_hard_regno_mode_ok): New function.
14669 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
14670 (epiphany_mode_class): Delete.
14671 (HARD_REGNO_MODE_OK): Delete.
14672 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
14673 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14674 (hard_regno_mode_ok): Rename to...
14675 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
14676 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
14677 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
14678 HARD_REGNO_MODE_OK.
14679 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
14680 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
14681 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14682 (frv_hard_regno_mode_ok): Make static and return a bool.
14683 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
14684 HARD_REGNO_MODE_OK.
14685 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
14686 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
14687 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
14688 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
14690 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14691 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
14692 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
14693 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
14695 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14696 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
14697 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14698 (ia64_hard_regno_mode_ok): New function.
14699 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
14700 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14701 (iq2000_hard_regno_mode_ok): New function.
14702 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
14703 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14704 (lm32_hard_regno_mode_ok): New function.
14705 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
14706 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
14707 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
14708 instead of HARD_REGNO_MODE_OK.
14709 (m32c_hard_regno_ok): Rename to...
14710 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
14711 (m32c_cannot_change_mode_class): Update accordingly.
14712 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14713 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
14714 (m32r_mode_class): Delete.
14715 (HARD_REGNO_MODE_OK): Delete.
14716 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14717 (m32r_hard_regno_mode_ok): Rename to...
14718 (m32r_hard_regno_modes): ...this.
14719 (m32r_mode_class): Make static.
14720 (m32r_hard_regno_mode_ok): New function.
14721 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
14722 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
14723 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14724 (m68k_hard_regno_mode_ok): Make static.
14725 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
14726 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14727 (mcore_hard_regno_mode_ok): New function.
14728 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
14729 (HARD_REGNO_MODE_OK): Delete.
14730 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
14732 (microblaze_hard_regno_mode_ok_p): ...this and make static.
14733 (microblaze_hard_regno_mode_ok): New function.
14734 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14735 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
14736 (mips_hard_regno_mode_ok): Delete.
14737 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
14738 (mips_hard_regno_mode_ok_p): ...this and make static.
14739 (mips_hard_regno_mode_ok_p): Rename to...
14740 (mips_hard_regno_mode_ok_uncached): ...this.
14741 (mips_hard_regno_mode_ok): New function.
14742 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
14743 of HARD_REGNO_MODE_OK.
14744 (mips_option_override): Update after above name changes.
14745 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14746 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
14747 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
14748 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
14749 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
14750 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14751 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
14752 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
14753 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
14754 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14755 (msp430_hard_regno_mode_ok): Make static and return a bool.
14756 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
14757 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
14758 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
14760 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14761 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
14762 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
14763 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
14764 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
14765 (PA_HARD_REGNO_MODE_OK): ...this
14766 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
14767 (PA_HARD_REGNO_MODE_OK): ...this.
14768 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14769 (pa_hard_regno_mode_ok): New function.
14770 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
14771 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14772 (pdp11_hard_regno_mode_ok): New function.
14773 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
14774 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
14776 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
14778 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14779 (rs6000_hard_regno_mode_ok): Rename to...
14780 (rs6000_hard_regno_mode_ok_uncached): ...this.
14781 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14782 (rs6000_hard_regno_mode_ok): New function.
14783 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
14784 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
14785 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
14786 (riscv_hard_regno_mode_ok): ...this and make static.
14787 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14788 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
14789 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
14790 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14791 (rl78_hard_regno_mode_ok): Make static and return bool.
14792 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
14793 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
14795 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
14796 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14797 (rs6000_hard_regno_mode_ok): Rename to...
14798 (rs6000_hard_regno_mode_ok_uncached): ...this.
14799 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14800 (rs6000_hard_regno_mode_ok): New function.
14801 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
14802 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
14803 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14804 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
14805 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
14806 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
14807 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14808 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
14809 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
14810 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14811 (sh_hard_regno_mode_ok): Make static.
14812 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
14813 instead of HARD_REGNO_MODE_OK.
14814 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
14815 (sparc_mode_class): Delete.
14816 (HARD_REGNO_MODE_OK): Delete.
14817 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14818 (hard_regno_mode_classes): Make static.
14819 (sparc_mode_class): Likewise.
14820 (sparc_hard_regno_mode_ok): New function.
14821 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
14822 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
14823 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
14825 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14826 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
14827 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
14828 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
14829 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
14830 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14831 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
14832 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
14833 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14834 (visium_hard_regno_mode_ok): New function.
14835 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
14836 instead of HARD_REGNO_MODE_OK.
14837 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
14838 (HARD_REGNO_MODE_OK): Delete.
14839 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
14840 (xtensa_hard_regno_mode_ok_p): ...this and make static.
14841 (xtensa_option_override): Update accordingly.
14842 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14843 (xtensa_hard_regno_mode_ok): New function.
14844 * system.h (HARD_REGNO_MODE_OK): Poison.
14846 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14847 Alan Hayward <alan.hayward@arm.com>
14848 David Sherwood <david.sherwood@arm.com>
14850 * target.def (hard_regno_call_part_clobbered): New hook.
14851 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
14852 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
14853 * doc/tm.texi: Regenerate.
14854 * hooks.h (hook_bool_uint_mode_false): Declare.
14855 * hooks.c (hook_bool_uint_mode_false): New function.
14856 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14857 * cselib.c (cselib_process_insn): Use
14858 targetm.hard_regno_call_part_clobbered instead of
14859 HARD_REGNO_CALL_PART_CLOBBERED.
14860 * ira-conflicts.c (ira_build_conflicts): Likewise.
14861 * ira-costs.c (ira_tune_allocno_costs): Likewise.
14862 * lra-constraints.c (need_for_call_save_p): Likewise.
14863 * lra-lives.c: Include target.h.
14864 (check_pseudos_live_through_calls): Use
14865 targetm.hard_regno_call_part_clobbered instead of
14866 HARD_REGNO_CALL_PART_CLOBBERED.
14867 * regcprop.c: Include target.h.
14868 (copyprop_hardreg_forward_1): Use
14869 targetm.hard_regno_call_part_clobbered instead of
14870 HARD_REGNO_CALL_PART_CLOBBERED.
14871 * reginfo.c (choose_hard_reg_mode): Likewise.
14872 * regrename.c (check_new_reg_p): Likewise.
14873 * reload.c (find_equiv_reg): Likewise.
14874 * reload1.c (emit_reload_insns): Likewise.
14875 * sched-deps.c (deps_analyze_insn): Likewise.
14876 * sel-sched.c (init_regs_for_mode): Likewise.
14877 (mark_unavailable_hard_regs): Likewise.
14878 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
14879 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14880 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
14882 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14883 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14884 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
14886 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
14888 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14889 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14890 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
14892 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14893 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14894 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
14896 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14897 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
14899 * config/powerpcspe/powerpcspe.c
14900 (rs6000_hard_regno_call_part_clobbered): New function.
14901 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14902 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14903 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
14905 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14906 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14907 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
14909 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14910 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14911 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
14913 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14914 Alan Hayward <alan.hayward@arm.com>
14915 David Sherwood <david.sherwood@arm.com>
14917 * rtl.h (subreg_memory_offset): Declare.
14918 * emit-rtl.c (subreg_memory_offset): New function.
14919 * expmed.c (store_bit_field_1): Use it.
14920 * expr.c (undefined_operand_subword_p): Likewise.
14921 * simplify-rtx.c (simplify_subreg): Likewise.
14923 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
14925 PR rtl-optimization/57448
14928 * optabs.c (expand_atomic_load): Place compiler memory barriers if
14929 using atomic_load pattern.
14930 (expand_atomic_store): Likewise.
14932 2017-09-04 Jakub Jelinek <jakub@redhat.com>
14935 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
14936 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
14937 handling. Use replace_call_with_value with NULL instead of
14938 gsi_replace, unlink_stmt_vdef and release_defs.
14940 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
14943 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
14945 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14948 * rtl.h (emit_library_call_value_1): Declare.
14949 (emit_library_call): Replace declaration with a series of overloads.
14950 Remove the parameter count argument.
14951 (emit_library_call_value): Likewise.
14952 * calls.c (emit_library_call_value_1): Make global. Replace varargs
14953 with an "rtx_mode_t *".
14954 (emit_library_call_value): Delete.
14955 (emit_library_call): Likewise.
14956 * asan.c (asan_emit_stack_protection): Update calls accordingly.
14957 (asan_emit_allocas_unpoison): Likewise.
14958 * builtins.c (expand_builtin_powi): Likewise.
14959 (expand_asan_emit_allocas_unpoison): Likewise.
14960 * cfgexpand.c (expand_main_function): Likewise.
14961 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
14962 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
14963 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
14964 * config/arm/arm.c (arm_trampoline_init): Likewise.
14965 (arm_call_tls_get_addr): Likewise.
14966 (arm_expand_divmod_libfunc): Likewise.
14967 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
14968 (smulsi3_highpart): Likewise.
14969 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
14970 (c6x_expand_compare): Likewise.
14971 (c6x_expand_movmem): Likewise.
14972 * config/frv/frv.c (frv_trampoline_init): Likewise.
14973 * config/i386/i386.c (ix86_trampoline_init): Likewise.
14974 (ix86_expand_divmod_libfunc): Likewise.
14975 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
14976 (ia64_expand_compare): Likewise.
14977 (ia64_profile_hook): Likewise.
14978 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
14979 (nonlocal_goto): Likewise.
14980 (restore_stack_nonlocal): Likewise.
14981 * config/m32r/m32r.c (block_move_call): Likewise.
14982 (m32r_trampoline_init): Likewise.
14983 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
14984 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
14985 (m68k_call_m68k_read_tp): Likewise.
14986 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
14987 (microblaze_expand_divide): Likewise.
14988 * config/mips/mips.h (mips_args): Likewise.
14989 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
14990 (MIPS_ICACHE_SYNC): Likewise.
14991 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
14992 (nios2_trampoline_init): Likewise.
14993 * config/pa/pa.c (hppa_tls_call): Likewise.
14994 (pa_trampoline_init): Likewise.
14995 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
14996 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
14997 (expand_strn_compare): Likewise.
14998 (rs6000_generate_compare): Likewise.
14999 (rs6000_expand_float128_convert): Likewise.
15000 (output_profile_hook): Likewise.
15001 (rs6000_trampoline_init): Likewise.
15002 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
15003 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
15004 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
15005 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
15006 (rs6000_generate_compare): Likewise.
15007 (rs6000_expand_float128_convert): Likewise.
15008 (output_profile_hook): Likewise.
15009 (rs6000_trampoline_init): Likewise.
15010 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
15011 * config/sh/sh.c (sh_trampoline_init): Likewise.
15012 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
15013 (sparc_emit_float_lib_cmp): Likewise.
15014 (sparc32_initialize_trampoline): Likewise.
15015 (sparc64_initialize_trampoline): Likewise.
15016 (sparc_profile_hook): Likewise.
15017 * config/spu/spu.c (ea_load_store): Likewise.
15018 * config/spu/spu.md (floatunssidf2): Likewise.
15019 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
15020 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
15021 * config/visium/visium.c (expand_block_move_4): Likewise.
15022 (expand_block_move_2): Likewise.
15023 (expand_block_move_1): Likewise.
15024 (expand_block_set_4): Likewise.
15025 (expand_block_set_2): Likewise.
15026 (expand_block_set_1): Likewise.
15027 (visium_trampoline_init): Likewise.
15028 (visium_profile_hook): Likewise.
15029 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
15030 (xtensa_setup_frame_addresses): Likewise.
15031 (xtensa_trampoline_init): Likewise.
15032 * except.c (sjlj_emit_function_enter): Likewise.
15033 (sjlj_emit_function_exit): Likewise.
15034 * explow.c (allocate_dynamic_stack_space): Likewise.
15035 (probe_stack_range): Likewise.
15036 * expr.c (convert_mode_scalar): Likewise.
15037 * optabs.c (expand_binop): Likewise.
15038 (expand_twoval_binop_libfunc): Likewise.
15039 (expand_unop): Likewise.
15040 (prepare_cmp_insn): Likewise.
15041 (prepare_float_lib_cmp): Likewise.
15042 (expand_float): Likewise.
15043 (expand_fix): Likewise.
15044 (expand_fixed_convert): Likewise.
15045 (maybe_emit_sync_lock_test_and_set): Likewise.
15046 (expand_atomic_compare_and_swap): Likewise.
15047 (expand_mem_thread_fence): Likewise.
15048 (expand_atomic_fetch_op): Likewise.
15050 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
15052 * doc/generic.texi (OpenACC): Adjust URL.
15053 * doc/invoke.texi (C Dialect Options): Ditto.
15055 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
15057 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
15058 predicate for operand 1. Add (m,<S>) constraint.
15059 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
15060 Prevent memory operand 1 with register operand 2.
15062 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
15064 PR rtl-optimization/82024
15065 * combine.c (try_combine): If the combination result is a PARALLEL,
15066 and we only need to retain the SET in there that would be placed
15067 at I2, check that we can place that at I3 instead, before doing so.
15069 2017-09-01 Jakub Jelinek <jakub@redhat.com>
15072 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
15074 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
15075 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
15078 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
15079 Jeff Law <law@redhat.com>
15081 * varasm.c (bss_initializer_p): Do not put constants into .bss
15082 (categorize_decl_for_section): Handle bss_initializer_p returning
15083 false when DECL_INITIAL is NULL.
15085 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15088 * config/s390/s390.c (s390_can_inline_p): New function.
15090 2017-09-01 Jeff Law <law@redhat.com>
15092 PR tree-optimization/82052
15093 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
15094 Always initialize the returned slot after a hash table miss
15095 when INSERT is true.
15097 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
15099 * config/s390/s390.md (mem_signal_fence): Remove.
15100 * doc/md.texi (mem_signal_fence): Remove.
15101 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
15103 * target-insns.def (mem_signal_fence): Remove.
15105 2017-09-01 Jakub Jelinek <jakub@redhat.com>
15108 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
15111 * asan.c (create_odr_indicator): Strip name encoding from assembler
15112 name before appending it after __odr_asan_.
15114 2017-09-01 Martin Liska <mliska@suse.cz>
15116 PR tree-optimization/82059
15117 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
15118 frequency only when an edge is redirected.
15120 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15122 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
15123 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
15124 (arc_conditional_register_usage): Remove ARC600 lp_count
15126 (arc_file_start): Emit Tag_ARC_CPU_variation.
15127 (arc_can_use_doloop_p): New conditions to use ZOLs.
15128 (hwloop_fail): New function.
15129 (hwloop_optimize): Likewise.
15130 (hwloop_pattern_reg): Likewise.
15131 (arc_doloop_hooks): New struct, to be used with reorg_loops.
15132 (arc_reorg_loops): New function, calls reorg_loops.
15133 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
15134 (arc600_corereg_hazard): Remove ZOL checking, case handled by
15136 (arc_loop_hazard): Remove function, functionality moved into
15138 (arc_hazard): Remove arc_loop_hazard call.
15139 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
15140 into hwloop_optimize.
15141 (arc_label_align): Remove ZOL handling.
15142 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
15143 * config/arc/arc.md (doloop_begin): Remove pattern.
15144 (doloop_begin_i): Likewise.
15145 (doloop_end_i): Likewise.
15146 (doloop_fallback): Likewise.
15147 (doloop_fallback_m): Likewise.
15148 (doloop_end): Reimplement expand.
15149 (arc_lp): New pattern for LP instruction.
15150 (loop_end): New pattern.
15151 (loop_fail): Likewise.
15152 (decrement_and_branch_until_zero): Likewise.
15153 * config/arc/arc.opt (mlpc-width): New option.
15154 * doc/invoke.texi (mlpc-width): Document option.
15156 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15158 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
15159 (arc_ccfsm_advance): Fix checking for delay slots.
15160 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
15162 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15164 * config/arc/arc.md (movqi_insn): Add stores to save constant long
15166 (movhi_insn): Update store instruction constraint which are saving
15167 6-bit short immediates.
15168 (movsi_insn): Consider also short scaled load operations.
15169 (zero_extendhisi2_i): Use Usd constraint instead of T.
15170 (extendhisi2_i): Add q constraint.
15171 (arc_clzsi2): Add type and length attributes.
15172 (arc_ctzsi2): Likewise.
15173 * config/arc/constraints.md (Usc): Update constraint, the
15174 assembler can parse two relocations for a single instruction.
15176 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15178 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
15179 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
15181 2017-08-31 Olivier Hainque <hainque@adacore.com>
15183 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
15184 match as powerpc-wrs-vxworks*.
15186 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
15188 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
15189 register constraint for by-element operand.
15190 (aarch64_mls_elt_merge<mode>): Likewise.
15192 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
15194 * config/arc/arc.c (arc_can_follow_jump): Check for short
15197 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
15199 * config.gcc: Use g.opt for arc.
15200 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
15201 functionality moved to ...
15202 (legitimate_scaled_address_p): New function, ...here.
15203 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
15204 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
15205 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
15207 (arc_override_options): Handle G option.
15208 (arc_output_pic_addr_const): Correct function definition.
15209 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
15210 (arc_decl_anon_ns_mem_p): Delete.
15211 (arc_in_small_data_p): Overhaul this function to take into
15212 consideration the value given via G option.
15213 (arc_rewrite_small_data_1): Renamed and corrected old
15214 arc_rewrite_small_data function.
15215 (arc_rewrite_small_data): New function.
15216 (small_data_pattern): Don't use pic_offset_table_rtx.
15217 * config/arc/arc.h (CC1_SPEC): Recognize G option.
15218 * config/arc/simdext.md (movmisalignv2hi): Use
15219 prepare_move_operands function.
15221 (movmisalign*): Likewise.
15222 * doc/invoke.texi (ARC options): Document -G option.
15224 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
15226 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
15228 * config/arc/arc.c (arc_print_operand): Output scalled address for
15229 sdata whenever is possible.
15230 (arc_in_small_data_p): Allow sdata for 64bit datum when double
15231 load/stores are available.
15232 (compact_sda_memory_operand): Check for the alignment required by
15233 code density instructions.
15234 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
15236 * config/arc/constraints.md (Usd): Update constraint.
15237 (Us0): New constraint.
15238 (Usc): Update constraint.
15240 2017-08-31 Richard Biener <rguenther@suse.de>
15242 PR middle-end/82054
15243 * dwarf2out.c (dwarf2out_early_global_decl): Process each
15244 function only once.
15246 2017-08-31 Tamar Christina <tamar.christina@arm.com>
15248 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
15249 Resize type_signature.
15251 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
15252 Alan Hayward <alan.hayward@arm.com>
15253 David Sherwood <david.sherwood@arm.com>
15255 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
15256 subregs whose inner modes can be stored in GPRs.
15257 (aarch64_classify_index): Likewise.
15259 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
15260 Alan Hayward <alan.hayward@arm.com>
15261 David Sherwood <david.sherwood@arm.com>
15263 * config/aarch64/iterators.md (V_cmp_result): Rename to...
15264 (V_INT_EQUIV): ...this.
15265 (v_cmp_result): Rename to...
15266 (v_int_equiv): ...this.
15267 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
15268 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
15269 (copysign<mode>3): Likewise.
15270 (aarch64_simd_bsl<mode>_internal): Likewise.
15271 (aarch64_simd_bsl<mode>): Likewise.
15272 (vec_cmp<mode><mode>): Likewise.
15273 (vcond<mode><mode>): Likewise.
15274 (vcond<v_cmp_mixed><mode>): Likewise.
15275 (vcondu<mode><v_cmp_mixed>): Likewise.
15276 (aarch64_cm<optab><mode>): Likewise.
15277 (aarch64_cmtst<mode>): Likewise.
15278 (aarch64_fac<optab><mode>): Likewise.
15279 (vec_perm_const<mode>): Likewise.
15280 (vcond_mask_<mode><v_cmp_result>): Rename to...
15281 (vcond_mask_<mode><v_int_equiv>): ...this.
15282 (vec_cmp<mode><v_cmp_result>): Rename to...
15283 (vec_cmp<mode><v_int_equiv>): ...this.
15285 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
15286 Alan Hayward <alan.hayward@arm.com>
15287 David Sherwood <david.sherwood@arm.com>
15289 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
15291 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
15292 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
15293 (UNSPEC_LD4_DREG): New unspecs.
15294 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
15295 (aarch64_ld2<mode>_dreg_be): Replace with...
15296 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
15298 (aarch64_ld3<mode>_dreg_le)
15299 (aarch64_ld3<mode>_dreg_be): Replace with...
15300 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
15302 (aarch64_ld4<mode>_dreg_le)
15303 (aarch64_ld4<mode>_dreg_be): Replace with...
15304 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
15307 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15309 PR tree-optimization/81987
15310 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
15311 insert an initializer in a location not dominated by the stride
15314 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
15316 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
15317 on the entire header of the finally block in the fallthru case.
15319 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
15321 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
15323 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
15325 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
15326 rs6000_emit_move_from_cr and call renamed function.
15327 (rs6000_emit_prologue): Call renamed functions.
15328 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
15329 movesi_from_cr, remove volatile CRs.
15331 2017-08-30 Jon Beniston <jon@beniston.com>
15332 Richard Biener <rguenther@suse.de>
15334 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
15335 of VECTOR_MODE_P check.
15336 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
15337 element vector types.
15339 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15341 * df.h (df_read_modify_subreg_p): Remove in favor of...
15342 * rtl.h (read_modify_subreg_p): ...this new function. Take a
15343 const_rtx instead of an rtx.
15344 * cprop.c (local_cprop_find_used_regs): Update accordingly.
15345 * df-problems.c (df_word_lr_mark_ref): Likewise.
15346 * ira-lives.c (mark_pseudo_reg_live): Likewise.
15347 (mark_pseudo_reg_dead): Likewise.
15348 (mark_ref_dead): Likewise.
15349 * reginfo.c (init_subregs_of_mode): Likewise.
15350 * sched-deps.c (sched_analyze_1): Likewise.
15351 * df-scan.c (df_def_record_1): Likewise.
15352 (df_uses_record): Likewise.
15353 (df_read_modify_subreg_p): Remove in favor of...
15354 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
15355 const_rtx instead of an rtx.
15357 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15358 Alan Hayward <alan.hayward@arm.com>
15359 David Sherwood <david.sherwood@arm.com>
15361 * rtl.h (partial_subreg_p): New function.
15362 * caller-save.c (save_call_clobbered_regs): Use it.
15363 * calls.c (expand_call): Likewise.
15364 * combine.c (combinable_i3pat): Likewise.
15365 (simplify_set): Likewise.
15366 (make_extraction): Likewise.
15367 (make_compound_operation_int): Likewise.
15368 (gen_lowpart_or_truncate): Likewise.
15369 (force_to_mode): Likewise.
15370 (make_field_assignment): Likewise.
15371 (reg_truncated_to_mode): Likewise.
15372 (record_truncated_value): Likewise.
15373 (move_deaths): Likewise.
15374 * cse.c (record_jump_cond): Likewise.
15375 (cse_insn): Likewise.
15376 * cselib.c (cselib_lookup_1): Likewise.
15377 * expmed.c (extract_bit_field_using_extv): Likewise.
15378 * function.c (assign_parm_setup_reg): Likewise.
15379 * ifcvt.c (noce_convert_multiple_sets): Likewise.
15380 * ira-build.c (create_insn_allocnos): Likewise.
15381 * lra-coalesce.c (merge_pseudos): Likewise.
15382 * lra-constraints.c (match_reload): Likewise.
15383 (simplify_operand_subreg): Likewise.
15384 (curr_insn_transform): Likewise.
15385 * lra-lives.c (process_bb_lives): Likewise.
15386 * lra.c (new_insn_reg): Likewise.
15387 (lra_substitute_pseudo): Likewise.
15388 * regcprop.c (mode_change_ok): Likewise.
15389 (maybe_mode_change): Likewise.
15390 (copyprop_hardreg_forward_1): Likewise.
15391 * reload.c (push_reload): Likewise.
15392 (find_reloads): Likewise.
15393 (find_reloads_subreg_address): Likewise.
15394 * reload1.c (alter_reg): Likewise.
15395 (eliminate_regs_1): Likewise.
15396 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15398 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
15400 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
15401 back to if statements, including unpack.
15403 2017-08-30 Martin Liska <mliska@suse.cz>
15405 PR inline-asm/82001
15406 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
15408 (func_checker::compare_asm_inputs_outputs): ... this function.
15409 (func_checker::compare_gimple_asm): Use the function to compare
15410 also ASM constrains.
15411 * ipa-icf-gimple.h: Rename the function.
15413 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15414 Alan Hayward <alan.hayward@arm.com>
15415 David Sherwood <david.sherwood@arm.com>
15417 * coretypes.h (complex_mode): New type.
15418 * gdbhooks.py (build_pretty_printer): Handle it.
15419 * machmode.h (complex_mode): New class.
15420 (complex_mode::includes_p): New function.
15421 (is_complex_int_mode): Likewise.
15422 (is_complex_float_mode): Likewise.
15423 * genmodes.c (get_mode_class): Handle complex mode classes.
15424 * function.c (expand_function_end): Use is_complex_int_mode.
15426 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15427 Alan Hayward <alan.hayward@arm.com>
15428 David Sherwood <david.sherwood@arm.com>
15430 * coretypes.h (scalar_mode_pod): New typedef.
15431 * gdbhooks.py (build_pretty_printer): Handle it.
15432 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
15433 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
15434 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
15435 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
15436 as_a <scalar_mode>.
15438 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15439 Alan Hayward <alan.hayward@arm.com>
15440 David Sherwood <david.sherwood@arm.com>
15442 * machmode.h (mode_for_vector): Take a scalar_mode instead
15444 * stor-layout.c (mode_for_vector): Likewise.
15445 * explow.c (promote_mode): Use as_a <scalar_mode>.
15446 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
15448 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15449 Alan Hayward <alan.hayward@arm.com>
15450 David Sherwood <david.sherwood@arm.com>
15452 * target.def (preferred_simd_mode): Take a scalar_mode
15453 instead of a machine_mode.
15454 * targhooks.h (default_preferred_simd_mode): Likewise.
15455 * targhooks.c (default_preferred_simd_mode): Likewise.
15456 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
15457 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
15458 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
15459 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
15460 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
15461 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
15462 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
15463 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
15465 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
15466 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
15467 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
15468 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
15469 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
15470 * doc/tm.texi: Regenerate.
15471 * optabs-query.c (can_vec_mask_load_store_p): Return false for
15474 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15475 Alan Hayward <alan.hayward@arm.com>
15476 David Sherwood <david.sherwood@arm.com>
15478 * target.def (scalar_mode_supported_p): Take a scalar_mode
15479 instead of a machine_mode.
15480 * targhooks.h (default_scalar_mode_supported_p): Likewise.
15481 * targhooks.c (default_scalar_mode_supported_p): Likewise.
15482 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
15483 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
15484 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
15485 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
15486 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
15487 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
15488 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
15489 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
15490 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
15491 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
15492 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
15493 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
15495 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
15496 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
15497 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
15498 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
15499 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
15501 * doc/tm.texi: Regenerate.
15503 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15504 Alan Hayward <alan.hayward@arm.com>
15505 David Sherwood <david.sherwood@arm.com>
15507 * coretypes.h (opt_scalar_mode): New typedef.
15508 * gdbhooks.py (build_pretty_printers): Handle it.
15509 * machmode.h (mode_iterator::get_2xwider): Add overload for
15511 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
15513 * expr.c (convert_mode_scalar): Likewise.
15514 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15515 * optabs.c (expand_float): Likewise.
15516 (expand_fix): Likewise.
15517 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15519 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15520 Alan Hayward <alan.hayward@arm.com>
15521 David Sherwood <david.sherwood@arm.com>
15523 * optabs.c (expand_float): Explicitly check for scalars before
15524 using a branching expansion.
15525 (expand_fix): Likewise.
15527 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15528 Alan Hayward <alan.hayward@arm.com>
15529 David Sherwood <david.sherwood@arm.com>
15531 * expr.c (convert_mode): Split scalar handling out into...
15532 (convert_mode_scalar): ...this new function. Treat the modes
15535 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15536 Alan Hayward <alan.hayward@arm.com>
15537 David Sherwood <david.sherwood@arm.com>
15539 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
15541 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
15543 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15544 Alan Hayward <alan.hayward@arm.com>
15545 David Sherwood <david.sherwood@arm.com>
15547 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
15548 rather than a machine_mode.
15549 (fixed_from_string): Likewise.
15550 (fixed_convert): Likewise.
15551 (fixed_convert_from_int): Likewise.
15552 (fixed_convert_from_real): Likewise.
15553 (real_convert_from_fixed): Likewise.
15554 * fixed-value.c (fixed_from_double_int): Likewise.
15555 (fixed_from_string): Likewise.
15556 (fixed_convert): Likewise.
15557 (fixed_convert_from_int): Likewise.
15558 (fixed_convert_from_real): Likewise.
15559 (real_convert_from_fixed): Likewise.
15560 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
15562 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15563 Alan Hayward <alan.hayward@arm.com>
15564 David Sherwood <david.sherwood@arm.com>
15566 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
15567 of separate mode class checks. Do not allow vector modes here.
15568 (immed_wide_int_const): Use as_a <scalar_mode>.
15569 * explow.c (trunc_int_for_mode): Likewise.
15570 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
15571 (wi::shwi): Likewise.
15572 (wi::min_value): Likewise.
15573 (wi::max_value): Likewise.
15574 * dwarf2out.c (loc_descriptor): Likewise.
15575 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
15576 for CONST_WIDE_INT.
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 * tree.h (SCALAR_TYPE_MODE): New macro.
15583 * expr.c (expand_expr_addr_expr_1): Use it.
15584 (expand_expr_real_2): Likewise.
15585 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
15586 (fold_convert_const_fixed_from_int): Likewise.
15587 (fold_convert_const_fixed_from_real): Likewise.
15588 (native_encode_fixed): Likewise
15589 (native_encode_complex): Likewise
15590 (native_encode_vector): Likewise.
15591 (native_interpret_fixed): Likewise.
15592 (native_interpret_real): Likewise.
15593 (native_interpret_complex): Likewise.
15594 (native_interpret_vector): Likewise.
15595 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
15596 (simd_clone_adjust_argument_types): Likewise.
15597 (simd_clone_init_simd_arrays): Likewise.
15598 (simd_clone_adjust): Likewise.
15599 * stor-layout.c (layout_type): Likewise.
15600 * tree.c (build_minus_one_cst): Likewise.
15601 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15602 * tree-inline.c (estimate_move_cost): Likewise.
15603 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
15604 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
15605 (vectorizable_reduction): Likewise.
15606 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
15607 (vect_recog_mixed_size_cond_pattern): Likewise.
15608 (check_bool_pattern): Likewise.
15609 (adjust_bool_pattern): Likewise.
15610 (search_type_for_mask_1): Likewise.
15611 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
15612 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15613 (vectorizable_load): Likewise.
15614 (vectorizable_store): Likewise.
15615 * ubsan.c (ubsan_encode_value): Likewise.
15616 * varasm.c (output_constant): Likewise.
15618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15619 Alan Hayward <alan.hayward@arm.com>
15620 David Sherwood <david.sherwood@arm.com>
15622 * coretypes.h (scalar_mode): New class.
15623 * machmode.h (scalar_mode): Likewise.
15624 (scalar_mode::includes_p): New function.
15625 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
15626 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
15627 * genmodes.c (get_mode_class): Handle remaining scalar modes.
15628 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
15629 * expmed.c (store_bit_field_1): Likewise.
15630 (extract_bit_field_1): Likewise.
15631 * expr.c (write_complex_part): Likewise.
15632 (read_complex_part): Likewise.
15633 (emit_move_complex_push): Likewise.
15634 (expand_expr_real_2): Likewise.
15635 * function.c (assign_parm_setup_reg): Likewise.
15636 (assign_parms_unsplit_complex): Likewise.
15637 * optabs.c (expand_binop): Likewise.
15638 * rtlanal.c (subreg_get_info): Likewise.
15639 * simplify-rtx.c (simplify_immed_subreg): Likewise.
15640 * varasm.c (output_constant_pool_2): Likewise.
15642 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15643 Alan Hayward <alan.hayward@arm.com>
15644 David Sherwood <david.sherwood@arm.com>
15646 * expmed.c (extract_high_half): Use scalar_int_mode and remove
15648 (expmed_mult_highpart_optab): Likewise.
15649 (expmed_mult_highpart): Likewise.
15651 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15652 Alan Hayward <alan.hayward@arm.com>
15653 David Sherwood <david.sherwood@arm.com>
15655 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
15656 instead of a machine_mode.
15657 (builtin_memset_read_str): Likewise.
15658 * builtins.c (c_readstr): Likewise.
15659 (builtin_memcpy_read_str): Likewise.
15660 (builtin_strncpy_read_str): Likewise.
15661 (builtin_memset_read_str): Likewise.
15662 (builtin_memset_gen_str): Likewise.
15663 (expand_builtin_signbit): Use scalar_int_mode for local variables.
15664 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
15665 instead of a machine_mode.
15666 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
15668 (make_extraction): Likewise.
15669 (try_widen_shift_mode): Take and return scalar_int_modes instead
15671 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
15672 a scalar_int_mode instead of a machine_mode.
15673 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
15674 (avr_addr_space_pointer_mode): Likewise.
15675 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
15676 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
15677 (msp430_unwind_word_mode): Likewise.
15678 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
15679 (spu_addr_space_pointer_mode): Likewise.
15680 (spu_addr_space_address_mode): Likewise.
15681 (spu_libgcc_cmp_return_mode): Likewise.
15682 (spu_libgcc_shift_count_mode): Likewise.
15683 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
15684 (rl78_addr_space_pointer_mode): Likewise.
15685 (fl78_unwind_word_mode): Likewise.
15686 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
15688 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
15689 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
15690 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
15691 (mips_valid_pointer_mode): Likewise.
15692 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
15693 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
15694 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
15696 (ft32_addr_space_address_mode): Likewise.
15697 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
15698 scalar_int_mode instead of a machine_mode.
15699 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
15701 (m32c_addr_space_address_mode): Likewise.
15702 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
15703 (rs6000_eh_return_filter_mode): Likewise.
15704 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
15705 (rs6000_eh_return_filter_mode): Likewise.
15706 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
15707 (s390_libgcc_shift_count_mode): Likewise.
15708 (s390_unwind_word_mode): Likewise.
15709 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
15711 * target.def (mode_rep_extended): Likewise.
15712 (valid_pointer_mode): Likewise.
15713 (addr_space.valid_pointer_mode): Likewise.
15714 (eh_return_filter_mode): Return a scalar_int_mode rather than
15716 (libgcc_cmp_return_mode): Likewise.
15717 (libgcc_shift_count_mode): Likewise.
15718 (unwind_word_mode): Likewise.
15719 (addr_space.pointer_mode): Likewise.
15720 (addr_space.address_mode): Likewise.
15721 * doc/tm.texi: Regenerate.
15722 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
15724 (do_jump): Use scalar_int_mode for local variables.
15725 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
15726 rather than a machine_mode.
15727 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
15728 (scompare_loc_descriptor_wide): Likewise.
15729 (scompare_loc_descriptor_narrow): Likewise.
15730 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
15732 * except.c (sjlj_emit_dispatch_table): Likewise.
15733 (expand_builtin_eh_copy_values): Likewise.
15734 * explow.c (convert_memory_address_addr_space_1): Likewise.
15735 Take a scalar_int_mode rather than a machine_mode.
15736 (convert_memory_address_addr_space): Take a scalar_int_mode rather
15737 than a machine_mode.
15738 (memory_address_addr_space): Use scalar_int_mode for local variables.
15739 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
15740 rather than a machine_mode.
15741 * expmed.c (mask_rtx): Likewise.
15742 (init_expmed_one_conv): Likewise.
15743 (expand_mult_highpart_adjust): Likewise.
15744 (extract_high_half): Likewise.
15745 (expmed_mult_highpart_optab): Likewise.
15746 (expmed_mult_highpart): Likewise.
15747 (expand_smod_pow2): Likewise.
15748 (expand_sdiv_pow2): Likewise.
15749 (emit_store_flag_int): Likewise.
15750 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
15752 (extract_low_bits): Likewise.
15753 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
15755 * expr.c (pieces_addr::adjust): Likewise.
15756 (can_store_by_pieces): Likewise.
15757 (store_by_pieces): Likewise.
15758 (clear_by_pieces_1): Likewise.
15759 (expand_expr_addr_expr_1): Likewise.
15760 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
15761 (expand_expr_real_1): Likewise.
15762 (try_casesi): Likewise.
15763 * final.c (shorten_branches): Likewise.
15764 * fold-const.c (fold_convert_const_int_from_fixed): Change the
15765 type of "mode" to machine_mode.
15766 * internal-fn.c (expand_arith_overflow_result_store): Take a
15767 scalar_int_mode rather than a machine_mode.
15768 (expand_mul_overflow): Use scalar_int_mode for local variables.
15769 * loop-doloop.c (doloop_modify): Likewise.
15770 (doloop_optimize): Likewise.
15771 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
15772 than a machine_mode.
15773 (expand_doubleword_shift_condmove): Likewise.
15774 (expand_doubleword_shift): Likewise.
15775 (expand_doubleword_clz): Likewise.
15776 (expand_doubleword_popcount): Likewise.
15777 (expand_doubleword_parity): Likewise.
15778 (expand_absneg_bit): Use scalar_int_mode for local variables.
15779 (prepare_float_lib_cmp): Likewise.
15780 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
15781 rather than a machine_mode.
15782 (convert_memory_address_addr_space): Likewise.
15783 (get_mode_bounds): Likewise.
15784 (get_address_mode): Return a scalar_int_mode rather than a
15786 * rtlanal.c (get_address_mode): Likewise.
15787 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
15788 than a machine_mode.
15789 * targhooks.c (default_mode_rep_extended): Likewise.
15790 (default_valid_pointer_mode): Likewise.
15791 (default_addr_space_valid_pointer_mode): Likewise.
15792 (default_eh_return_filter_mode): Return a scalar_int_mode rather
15793 than a machine_mode.
15794 (default_libgcc_cmp_return_mode): Likewise.
15795 (default_libgcc_shift_count_mode): Likewise.
15796 (default_unwind_word_mode): Likewise.
15797 (default_addr_space_pointer_mode): Likewise.
15798 (default_addr_space_address_mode): Likewise.
15799 * targhooks.h (default_eh_return_filter_mode): Likewise.
15800 (default_libgcc_cmp_return_mode): Likewise.
15801 (default_libgcc_shift_count_mode): Likewise.
15802 (default_unwind_word_mode): Likewise.
15803 (default_addr_space_pointer_mode): Likewise.
15804 (default_addr_space_address_mode): Likewise.
15805 (default_mode_rep_extended): Take a scalar_int_mode rather than
15807 (default_valid_pointer_mode): Likewise.
15808 (default_addr_space_valid_pointer_mode): Likewise.
15809 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
15811 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
15812 rather than a machine_mode.
15813 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
15814 for local variables.
15815 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15816 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
15817 than a machine_mode.
15819 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15820 Alan Hayward <alan.hayward@arm.com>
15821 David Sherwood <david.sherwood@arm.com>
15823 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
15824 the mode argument to scalar_int_mode.
15825 (do_jump_by_parts_zero_rtx): Likewise.
15826 (do_jump_by_parts_equality_rtx): Likewise.
15827 (do_jump_by_parts_greater): Take a mode argument.
15828 (do_jump_by_parts_equality): Likewise.
15829 (do_jump_1): Update calls accordingly.
15831 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15832 Alan Hayward <alan.hayward@arm.com>
15833 David Sherwood <david.sherwood@arm.com>
15835 * is-a.h (safe_dyn_cast): New function.
15836 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
15837 (jump_table_for_label): Likewise.
15838 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
15839 instead of an rtx_insn *.
15840 (shorten_branches): Use dyn_cast instead of LABEL_P and
15841 JUMP_TABLE_DATA_P. Use jump_table_for_label and
15842 rtx_jump_table_data::get_data_mode.
15843 (final_scan_insn): Likewise.
15845 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15846 Alan Hayward <alan.hayward@arm.com>
15847 David Sherwood <david.sherwood@arm.com>
15849 * combine.c (try_combine): Use is_a <scalar_int_mode> when
15850 trying to combine a full-register integer set with a subreg
15853 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15854 Alan Hayward <alan.hayward@arm.com>
15855 David Sherwood <david.sherwood@arm.com>
15857 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
15858 that is always either address_mode or pointer_mode.
15860 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15861 Alan Hayward <alan.hayward@arm.com>
15862 David Sherwood <david.sherwood@arm.com>
15864 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
15865 when the two are known to be equal.
15867 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15868 Alan Hayward <alan.hayward@arm.com>
15869 David Sherwood <david.sherwood@arm.com>
15871 * simplify-rtx.c (simplify_const_unary_operation): Use
15872 is_a <scalar_int_mode> instead of checking for a nonzero
15873 precision. Forcibly convert op_mode to a scalar_int_mode
15874 in that case. More clearly differentiate the operand and
15875 result modes and use the former when deciding what the value
15876 of a count-bits operation should be. Use is_int_mode instead
15877 of checking for a MODE_INT. Remove redundant check for whether
15878 this mode has a zero precision.
15880 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15881 Alan Hayward <alan.hayward@arm.com>
15882 David Sherwood <david.sherwood@arm.com>
15884 * optabs.c (widen_leading): Change the type of the mode argument
15885 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
15886 (widen_bswap): Likewise.
15887 (expand_parity): Likewise.
15888 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
15889 (expand_ffs): Likewise.
15890 (epand_unop): Check for scalar integer modes before calling the
15893 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15894 Alan Hayward <alan.hayward@arm.com>
15895 David Sherwood <david.sherwood@arm.com>
15897 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
15899 (expand_expr_real_1): Update call accordingly.
15901 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15902 Alan Hayward <alan.hayward@arm.com>
15903 David Sherwood <david.sherwood@arm.com>
15905 * expmed.c (store_bit_field_using_insv): Add op0_mode and
15906 value_mode arguments. Use scalar_int_mode internally.
15907 (store_bit_field_1): Rename the new integer mode from imode
15908 to op0_mode and use it instead of GET_MODE (op0). Update calls
15909 to store_split_bit_field, store_bit_field_using_insv and
15910 store_fixed_bit_field.
15911 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
15912 Use scalar_int_mode internally. Use a bit count rather than a mode
15913 when calculating the largest bit size for get_best_mode.
15914 Update calls to store_split_bit_field and store_fixed_bit_field_1.
15915 (store_fixed_bit_field_1): Add mode and value_mode arguments.
15916 Remove assertion that OP0 has a scalar integer mode.
15917 (store_split_bit_field): Add op0_mode and value_mode arguments.
15918 Update calls to extract_fixed_bit_field.
15919 (extract_bit_field_using_extv): Add an op0_mode argument.
15920 Use scalar_int_mode internally.
15921 (extract_bit_field_1): Rename the new integer mode from imode to
15922 op0_mode and use it instead of GET_MODE (op0). Update calls to
15923 extract_split_bit_field, extract_bit_field_using_extv and
15924 extract_fixed_bit_field.
15925 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
15926 to extract_split_bit_field and extract_fixed_bit_field_1.
15927 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
15928 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
15929 on the target mode.
15930 (extract_split_bit_field): Add an op0_mode argument. Update call
15931 to extract_fixed_bit_field.
15933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15934 Alan Hayward <alan.hayward@arm.com>
15935 David Sherwood <david.sherwood@arm.com>
15937 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
15938 * explow.c (hard_function_value): Likewise.
15939 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
15940 convert_to_mode call outside the loop.
15941 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
15942 for the mode iterator. Require the mode specified by max_pieces
15944 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
15946 (copy_blkmode_to_reg): Likewise.
15947 (set_storage_via_setmem): Likewise.
15948 * optabs.c (prepare_cmp_insn): Likewise.
15949 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15950 * stor-layout.c (finish_bitfield_representative): Likewise.
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 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
15957 * expr.c (convert_move): Use them.
15958 (convert_modes): Likewise.
15959 (store_expr_with_bounds): Likewise.
15961 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15962 Alan Hayward <alan.hayward@arm.com>
15963 David Sherwood <david.sherwood@arm.com>
15965 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
15966 parameter for the mode of "x". Remove the "known_x", "known_mode"
15967 and "known_ret" arguments. Change the type of the mode argument
15968 to scalar_int_mode.
15969 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
15970 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
15971 (reg_num_sign_bit_copies_for_combine): Likewise.
15972 * rtlanal.c (nonzero_bits1): Likewise.
15973 (num_sign_bit_copies1): Likewise.
15974 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
15975 (reg_num_sign_bit_copies_general): Likewise.
15976 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
15977 (reg_nonzero_bits_general): Likewise.
15979 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15980 Alan Hayward <alan.hayward@arm.com>
15981 David Sherwood <david.sherwood@arm.com>
15983 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
15984 than in subroutines. Return 1 for non-integer modes.
15985 (cached_num_sign_bit_copies): Change the type of the mode parameter
15986 to scalar_int_mode.
15987 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
15988 classes. Handle CONST_INT_P first and then check whether X also
15989 has a scalar integer mode. Check the same thing for inner registers
15990 of a SUBREG and for values that are being extended or truncated.
15992 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15993 Alan Hayward <alan.hayward@arm.com>
15994 David Sherwood <david.sherwood@arm.com>
15996 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
15997 in subroutines. Return the mode mask for non-integer modes.
15998 (cached_nonzero_bits): Change the type of the mode parameter
15999 to scalar_int_mode.
16000 (nonzero_bits1): Likewise. Remove early exit for other mode
16001 classes. Handle CONST_INT_P first and then check whether X
16002 also has a scalar integer mode.
16004 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16005 Alan Hayward <alan.hayward@arm.com>
16006 David Sherwood <david.sherwood@arm.com>
16008 * expr.c (widest_int_mode_for_size): Make the comment match the code.
16009 Return a scalar_int_mode and assert that the size is greater than
16011 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
16012 (op_by_pieces_d::op_by_pieces_d): Likewise.
16013 (op_by_pieces_d::run): Likewise.
16014 (can_store_by_pieces): Likewise.
16016 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16017 Alan Hayward <alan.hayward@arm.com>
16018 David Sherwood <david.sherwood@arm.com>
16020 * combine.c (extract_left_shift): Add a mode argument and update
16022 (make_compound_operation_int): Change the type of the mode parameter
16023 to scalar_int_mode and update the call to extract_left_shift.
16025 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16026 Alan Hayward <alan.hayward@arm.com>
16027 David Sherwood <david.sherwood@arm.com>
16029 * combine.c (simplify_and_const_int): Change the type of the mode
16030 parameter to scalar_int_mode.
16031 (simplify_and_const_int_1): Likewise. Update recursive call.
16033 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16034 Alan Hayward <alan.hayward@arm.com>
16035 David Sherwood <david.sherwood@arm.com>
16037 * combine.c (simplify_compare_const): Check that the mode is a
16038 scalar_int_mode (rather than VOIDmode) before testing its
16040 (simplify_comparison): Move COMPARISON_P handling out of the
16041 loop and restrict the latter part of the loop to scalar_int_modes.
16042 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
16043 and when considering SUBREG_REGs. Use is_int_mode instead of
16044 checking GET_MODE_CLASS against MODE_INT.
16046 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16047 Alan Hayward <alan.hayward@arm.com>
16048 David Sherwood <david.sherwood@arm.com>
16050 * combine.c (try_widen_shift_mode): Move check for equal modes to...
16051 (simplify_shift_const_1): ...here. Use scalar_int_mode for
16052 shift_unit_mode and for modes involved in scalar shifts.
16054 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16055 Alan Hayward <alan.hayward@arm.com>
16056 David Sherwood <david.sherwood@arm.com>
16058 * combine.c (force_int_to_mode): New function, split out from...
16059 (force_to_mode): ...here. Keep xmode up-to-date and use it
16060 instead of GET_MODE (x).
16062 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16063 Alan Hayward <alan.hayward@arm.com>
16064 David Sherwood <david.sherwood@arm.com>
16066 * optabs-query.h (extraction_insn::struct_mode): Change type to
16067 opt_scalar_int_mode and update comment.
16068 (extraction_insn::field_mode): Change type to scalar_int_mode.
16069 (extraction_insn::pos_mode): Likewise.
16070 * combine.c (make_extraction): Update accordingly.
16071 * optabs-query.c (get_traditional_extraction_insn): Likewise.
16072 (get_optab_extraction_insn): Likewise.
16073 * recog.c (simplify_while_replacing): Likewise.
16074 * expmed.c (narrow_bit_field_mem): Change the type of the mode
16075 parameter to opt_scalar_int_mode.
16077 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16078 Alan Hayward <alan.hayward@arm.com>
16079 David Sherwood <david.sherwood@arm.com>
16081 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
16082 to a scalar_int_mode instead of a machine_mode.
16083 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
16084 (get_best_mode): Return a boolean and use a pointer argument to store
16085 the selected mode. Replace the limit mode parameter with a bit limit.
16086 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
16087 for the values returned by bit_field_mode_iterator::next_mode.
16088 (store_bit_field): Update call to get_best_mode.
16089 (store_fixed_bit_field): Likewise.
16090 (extract_fixed_bit_field): Likewise.
16091 * expr.c (optimize_bitfield_assignment_op): Likewise.
16092 * fold-const.c (optimize_bit_field_compare): Likewise.
16093 (fold_truth_andor_1): Likewise.
16094 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
16095 Update for new type of m_mode.
16096 (get_best_mode): As above.
16098 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16099 Alan Hayward <alan.hayward@arm.com>
16100 David Sherwood <david.sherwood@arm.com>
16102 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
16103 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
16104 (store_bit_field): Check is_a <scalar_int_mode> before calling
16105 strict_volatile_bitfield_p.
16106 (extract_bit_field): Likewise.
16108 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16109 Alan Hayward <alan.hayward@arm.com>
16110 David Sherwood <david.sherwood@arm.com>
16112 * target.def (cstore_mode): Return a scalar_int_mode.
16113 * doc/tm.texi: Regenerate.
16114 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
16115 * targhooks.h (default_cstore_mode): Likewise.
16116 * targhooks.c (default_cstore_mode): Likewise, using a forced
16118 * expmed.c (emit_cstore): Expect the target of the cstore to be
16121 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16122 Alan Hayward <alan.hayward@arm.com>
16123 David Sherwood <david.sherwood@arm.com>
16125 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
16127 (niter_desc): Likewise mode.
16128 (iv_analyze): Add a mode parameter.
16130 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
16131 and change its type to scalar_int_mode.
16132 * loop-iv.c: Update commentary at head of file.
16133 (iv_constant): Pass the mode paraeter before the rtx it describes
16134 and change its type to scalar_int_mode. Remove VOIDmode handling.
16135 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
16136 (iv_extend): Likewise.
16137 (shorten_into_mode): Likewise.
16138 (iv_add): Use scalar_int_mode.
16139 (iv_mult): Likewise.
16140 (iv_shift): Likewise.
16141 (canonicalize_iv_subregs): Likewise.
16142 (get_biv_step_1): Pass the outer_mode parameter before the rtx
16143 it describes and change its mode to scalar_int_mode. Also change
16144 the type of the returned inner_mode to scalar_int_mode.
16145 (get_biv_step): Likewise, turning outer_mode from a pointer
16146 into a direct parameter. Update call to get_biv_step_1.
16147 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
16148 iv_constant and get_biv_step.
16149 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
16150 and change its type to scalar_int_mode. Don't initialise iv->mode
16151 to VOIDmode and remove later checks for its still being VOIDmode.
16152 Update calls to iv_analyze_op and iv_analyze_expr. Check
16153 is_a <scalar_int_mode> when changing the mode under consideration.
16154 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
16155 Update call to iv_analyze_expr.
16156 (iv_analyze_op): Add a mode parameter. Reject subregs whose
16157 inner register is not also a scalar_int_mode. Update call to
16159 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
16160 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
16161 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
16162 separate mode class checks. Update calls to iv_analyze. Remove
16163 fix-up of VOIDmodes after iv_analyze_biv.
16164 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
16165 don't have a scalar_int_mode. Update call to biv_p.
16167 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16168 Alan Hayward <alan.hayward@arm.com>
16169 David Sherwood <david.sherwood@arm.com>
16171 * cfgexpand.c (convert_debug_memory_address): Use
16172 as_a <scalar_int_mode>.
16173 * combine.c (expand_compound_operation): Likewise.
16174 (make_extraction): Likewise.
16175 (change_zero_ext): Likewise.
16176 (simplify_comparison): Likewise.
16177 * cse.c (cse_insn): Likewise.
16178 * dwarf2out.c (minmax_loc_descriptor): Likewise.
16179 (mem_loc_descriptor): Likewise.
16180 (loc_descriptor): Likewise.
16181 * expmed.c (init_expmed_one_mode): Likewise.
16182 (synth_mult): Likewise.
16183 (emit_store_flag_1): Likewise.
16184 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
16185 of a comparison with size.
16186 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
16187 (reduce_to_bit_field_precision): Likewise.
16188 * function.c (expand_function_end): Likewise.
16189 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
16190 * loop-doloop.c (doloop_modify): Likewise.
16191 * optabs.c (expand_binop): Likewise.
16192 (expand_unop): Likewise.
16193 (expand_copysign_absneg): Likewise.
16194 (prepare_cmp_insn): Likewise.
16195 (maybe_legitimize_operand): Likewise.
16196 * recog.c (const_scalar_int_operand): Likewise.
16197 * rtlanal.c (get_address_mode): Likewise.
16198 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16199 (simplify_cond_clz_ctz): Likewise.
16200 * tree-nested.c (get_nl_goto_field): Likewise.
16201 * tree.c (build_vector_type_for_mode): Likewise.
16202 * var-tracking.c (use_narrower_mode): Likewise.
16204 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16205 Alan Hayward <alan.hayward@arm.com>
16206 David Sherwood <david.sherwood@arm.com>
16208 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
16209 * builtins.c (expand_builtin_signbit): Use it.
16210 * cfgexpand.c (expand_debug_expr): Likewise.
16211 * dojump.c (do_jump): Likewise.
16212 (do_compare_and_jump): Likewise.
16213 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
16214 * expmed.c (make_tree): Likewise.
16215 * expr.c (expand_expr_real_2): Likewise.
16216 (expand_expr_real_1): Likewise.
16217 (try_casesi): Likewise.
16218 * fold-const-call.c (fold_const_call_ss): Likewise.
16219 * fold-const.c (unextend): Likewise.
16220 (extract_muldiv_1): Likewise.
16221 (fold_single_bit_test): Likewise.
16222 (native_encode_int): Likewise.
16223 (native_encode_string): Likewise.
16224 (native_interpret_int): Likewise.
16225 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
16226 * internal-fn.c (expand_addsub_overflow): Likewise.
16227 (expand_neg_overflow): Likewise.
16228 (expand_mul_overflow): Likewise.
16229 (expand_arith_overflow): Likewise.
16230 * match.pd: Likewise.
16231 * stor-layout.c (layout_type): Likewise.
16232 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
16233 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
16234 * tree-ssanames.c (get_range_info): Likewise.
16235 * tree-switch-conversion.c (array_value_type) Likewise.
16236 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
16237 (vect_recog_divmod_pattern): Likewise.
16238 (vect_recog_mixed_size_cond_pattern): Likewise.
16239 * tree-vrp.c (extract_range_basic): Likewise.
16240 (simplify_float_conversion_using_ranges): Likewise.
16241 * tree.c (int_fits_type_p): Likewise.
16242 * ubsan.c (instrument_bool_enum_load): Likewise.
16243 * varasm.c (mergeable_string_section): Likewise.
16244 (narrowing_initializer_constant_valid_p): Likewise.
16245 (output_constant): Likewise.
16247 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16248 Alan Hayward <alan.hayward@arm.com>
16249 David Sherwood <david.sherwood@arm.com>
16251 * machmode.h (NARROWEST_INT_MODE): New macro.
16252 * expr.c (alignment_for_piecewise_move): Use it instead of
16253 GET_CLASS_NARROWEST_MODE (MODE_INT).
16254 (push_block): Likewise.
16255 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
16257 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16259 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16260 Alan Hayward <alan.hayward@arm.com>
16261 David Sherwood <david.sherwood@arm.com>
16263 * postreload.c (move2add_valid_value_p): Change the type of the
16264 mode parameter to scalar_int_mode.
16265 (move2add_use_add2_insn): Add a mode parameter and use it instead
16267 (move2add_use_add3_insn): Likewise.
16268 (reload_cse_move2add): Update accordingly.
16270 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16271 Alan Hayward <alan.hayward@arm.com>
16272 David Sherwood <david.sherwood@arm.com>
16274 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
16276 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
16277 * optabs.c (expand_unop): Likewise.
16279 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16280 Alan Hayward <alan.hayward@arm.com>
16281 David Sherwood <david.sherwood@arm.com>
16283 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
16284 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
16285 (popcount_loc_descriptor): Likewise.
16286 (bswap_loc_descriptor): Likewise.
16287 (rotate_loc_descriptor): Likewise.
16288 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
16289 calling the functions above.
16291 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16292 Alan Hayward <alan.hayward@arm.com>
16293 David Sherwood <david.sherwood@arm.com>
16295 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
16297 (try_combine): Likewise.
16298 (simplify_if_then_else): Likewise.
16299 * cse.c (cse_insn): Likewise.
16300 * dwarf2out.c (mem_loc_descriptor): Likewise.
16301 * emit-rtl.c (gen_lowpart_common): Likewise.
16302 * simplify-rtx.c (simplify_truncation): Likewise.
16303 (simplify_binary_operation_1): Likewise.
16304 (simplify_const_relational_operation): Likewise.
16305 (simplify_ternary_operation): Likewise.
16306 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
16308 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16309 Alan Hayward <alan.hayward@arm.com>
16310 David Sherwood <david.sherwood@arm.com>
16312 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
16313 * reload.c (push_reload): Likewise.
16314 (find_reloads): Likewise.
16316 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16317 Alan Hayward <alan.hayward@arm.com>
16318 David Sherwood <david.sherwood@arm.com>
16320 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
16321 (make_compound_operation_int): Likewise.
16322 (change_zero_ext): Likewise.
16323 * expr.c (convert_move): Likewise.
16324 (convert_modes): Likewise.
16325 * fwprop.c (forward_propagate_subreg): Likewise.
16326 * loop-iv.c (get_biv_step_1): Likewise.
16327 * optabs.c (widen_operand): Likewise.
16328 * postreload.c (move2add_valid_value_p): Likewise.
16329 * recog.c (simplify_while_replacing): Likewise.
16330 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16331 (simplify_binary_operation_1): Likewise. Remove redundant
16332 mode equality check.
16334 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16335 Alan Hayward <alan.hayward@arm.com>
16336 David Sherwood <david.sherwood@arm.com>
16338 * combine.c (combine_simplify_rtx): Add checks for
16339 is_a <scalar_int_mode>.
16340 (simplify_if_then_else): Likewise.
16341 (make_field_assignment): Likewise.
16342 (simplify_comparison): Likewise.
16343 * ifcvt.c (noce_try_bitop): Likewise.
16344 * loop-invariant.c (canonicalize_address_mult): Likewise.
16345 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16347 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16348 Alan Hayward <alan.hayward@arm.com>
16349 David Sherwood <david.sherwood@arm.com>
16351 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
16352 is_a <scalar_int_mode> instead of != BLKmode.
16354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16355 Alan Hayward <alan.hayward@arm.com>
16356 David Sherwood <david.sherwood@arm.com>
16358 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
16359 instead of != VOIDmode.
16360 * combine.c (if_then_else_cond): Likewise.
16361 (change_zero_ext): Likewise.
16362 * dwarf2out.c (mem_loc_descriptor): Likewise.
16363 (loc_descriptor): Likewise.
16364 * rtlanal.c (canonicalize_condition): Likewise.
16365 * simplify-rtx.c (simplify_relational_operation_1): 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 * simplify-rtx.c (simplify_binary_operation_1): Use
16372 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
16374 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16375 Alan Hayward <alan.hayward@arm.com>
16376 David Sherwood <david.sherwood@arm.com>
16378 * wide-int.h (int_traits<unsigned char>) New class.
16379 (int_traits<unsigned short>) Likewise.
16380 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
16381 Use GET_MODE_UNIT_PRECISION and remove redundant test for
16383 * combine.c (set_nonzero_bits_and_sign_copies): Use
16384 is_a <scalar_int_mode>.
16385 (find_split_point): Likewise.
16386 (combine_simplify_rtx): Likewise.
16387 (simplify_logical): Likewise.
16388 (expand_compound_operation): Likewise.
16389 (expand_field_assignment): Likewise.
16390 (make_compound_operation): Likewise.
16391 (extended_count): Likewise.
16392 (change_zero_ext): Likewise.
16393 (simplify_comparison): Likewise.
16394 * dwarf2out.c (scompare_loc_descriptor): Likewise.
16395 (ucompare_loc_descriptor): Likewise.
16396 (minmax_loc_descriptor): Likewise.
16397 (mem_loc_descriptor): Likewise.
16398 (loc_descriptor): Likewise.
16399 * expmed.c (init_expmed_one_mode): Likewise.
16400 * lra-constraints.c (lra_constraint_offset): Likewise.
16401 * optabs.c (prepare_libcall_arg): Likewise.
16402 * postreload.c (move2add_note_store): Likewise.
16403 * reload.c (operands_match_p): Likewise.
16404 * rtl.h (load_extend_op): Likewise.
16405 * rtlhooks.c (gen_lowpart_general): Likewise.
16406 * simplify-rtx.c (simplify_truncation): Likewise.
16407 (simplify_unary_operation_1): Likewise.
16408 (simplify_binary_operation_1): Likewise.
16409 (simplify_const_binary_operation): Likewise.
16410 (simplify_const_relational_operation): Likewise.
16411 (simplify_subreg): Likewise.
16412 * stor-layout.c (bitwise_mode_for_mode): Likewise.
16413 * var-tracking.c (adjust_mems): Likewise.
16414 (prepare_call_arguments): Likewise.
16416 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16417 Alan Hayward <alan.hayward@arm.com>
16418 David Sherwood <david.sherwood@arm.com>
16420 * machmode.h (is_int_mode): New fuction.
16421 * combine.c (find_split_point): Use it.
16422 (combine_simplify_rtx): Likewise.
16423 (simplify_if_then_else): Likewise.
16424 (simplify_set): Likewise.
16425 (simplify_shift_const_1): Likewise.
16426 (simplify_comparison): Likewise.
16427 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
16428 * cse.c (notreg_cost): Likewise.
16429 (cse_insn): Likewise.
16430 * cselib.c (cselib_lookup_1): Likewise.
16431 * dojump.c (do_jump_1): Likewise.
16432 (do_compare_rtx_and_jump): Likewise.
16433 * dse.c (get_call_args): Likewise.
16434 * dwarf2out.c (rtl_for_decl_init): Likewise.
16435 (native_encode_initializer): Likewise.
16436 * expmed.c (emit_store_flag_1): Likewise.
16437 (emit_store_flag): Likewise.
16438 * expr.c (convert_modes): Likewise.
16439 (store_field): Likewise.
16440 (expand_expr_real_1): Likewise.
16441 * fold-const.c (fold_read_from_constant_string): Likewise.
16442 * gimple-ssa-sprintf.c (get_format_string): Likewise.
16443 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
16444 * optabs.c (expand_binop): Likewise.
16445 (expand_unop): Likewise.
16446 (expand_abs_nojump): Likewise.
16447 (expand_one_cmpl_abs_nojump): Likewise.
16448 * simplify-rtx.c (mode_signbit_p): Likewise.
16449 (val_signbit_p): Likewise.
16450 (val_signbit_known_set_p): Likewise.
16451 (val_signbit_known_clear_p): Likewise.
16452 (simplify_relational_operation_1): Likewise.
16453 * tree.c (vector_type_mode): Likewise.
16455 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16456 Alan Hayward <alan.hayward@arm.com>
16457 David Sherwood <david.sherwood@arm.com>
16459 * machmode.h (smallest_mode_for_size): Fix formatting.
16460 (smallest_int_mode_for_size): New function.
16461 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
16462 instead of smallest_mode_for_size.
16463 * combine.c (make_extraction): Likewise.
16464 * config/arc/arc.c (arc_expand_movmem): Likewise.
16465 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
16466 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
16467 * config/s390/s390.c (s390_expand_insv): Likewise.
16468 * config/sparc/sparc.c (assign_int_registers): Likewise.
16469 * config/spu/spu.c (spu_function_value): Likewise.
16470 (spu_function_arg): Likewise.
16471 * coverage.c (get_gcov_type): Likewise.
16472 (get_gcov_unsigned_t): Likewise.
16473 * dse.c (find_shift_sequence): Likewise.
16474 * expmed.c (store_bit_field_1): Likewise.
16475 * expr.c (convert_move): Likewise.
16476 (store_field): Likewise.
16477 * internal-fn.c (expand_arith_overflow): Likewise.
16478 * optabs-query.c (get_best_extraction_insn): Likewise.
16479 * optabs.c (expand_twoval_binop_libfunc): Likewise.
16480 * stor-layout.c (layout_type): Likewise.
16481 (initialize_sizetypes): Likewise.
16482 * targhooks.c (default_get_mask_mode): Likewise.
16483 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
16485 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16486 Alan Hayward <alan.hayward@arm.com>
16487 David Sherwood <david.sherwood@arm.com>
16489 * machmode.h (opt_mode::else_blk): New function.
16490 (int_mode_for_mode): Declare.
16491 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
16492 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
16494 * cfgexpand.c (expand_debug_expr): Likewise.
16495 * combine.c (gen_lowpart_or_truncate): Likewise.
16496 (gen_lowpart_for_combine): Likewise.
16497 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
16498 * config/avr/avr.c (avr_to_int_mode): Likewise.
16499 (avr_out_plus_1): Likewise.
16500 (avr_out_plus): Likewise.
16501 (avr_out_round): Likewise.
16502 * config/i386/i386.c (ix86_split_to_parts): Likewise.
16503 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
16504 (s390_expand_vcond): Likewise.
16505 * config/spu/spu.c (spu_split_immediate): Likewise.
16506 (spu_expand_mov): Likewise.
16507 * dse.c (get_stored_val): Likewise.
16508 * expmed.c (store_bit_field_1): Likewise.
16509 (convert_extracted_bit_field): Use int_mode_for_mode instead of
16511 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
16512 (extract_low_bits): Likewise.
16513 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
16514 handling rather than repeating the check.
16515 (emit_group_store): Likewise.
16516 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
16517 * optabs.c (expand_absneg_bit): Likewise.
16518 (expand_copysign_absneg): Likewise.
16519 (expand_copysign_bit): Likewise.
16520 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
16521 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
16522 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
16523 * var-tracking.c (prepare_call_arguments): Likewise.
16524 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
16525 int_mode_for_mode instead of mode_for_size.
16526 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
16528 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16529 Alan Hayward <alan.hayward@arm.com>
16530 David Sherwood <david.sherwood@arm.com>
16532 * machmode.h (int_mode_for_size): New function.
16533 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
16534 instead of mode_for_size.
16535 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
16537 * combine.c (expand_field_assignment): Use int_mode_for_size
16538 instead of mode_for_size.
16539 (make_extraction): Likewise.
16540 (simplify_shift_const_1): Likewise.
16541 (simplify_comparison): Likewise.
16542 * dojump.c (do_jump): Likewise.
16543 * dwarf2out.c (mem_loc_descriptor): Likewise.
16544 * emit-rtl.c (init_derived_machine_modes): Likewise.
16545 * expmed.c (flip_storage_order): Likewise.
16546 (convert_extracted_bit_field): Likewise.
16547 * expr.c (copy_blkmode_from_reg): Likewise.
16548 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
16549 * internal-fn.c (expand_mul_overflow): Likewise.
16550 * lower-subreg.c (simple_move): Likewise.
16551 * optabs-libfuncs.c (init_optabs): Likewise.
16552 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16553 * tree.c (vector_type_mode): Likewise.
16554 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
16555 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
16556 * tree-vect-generic.c (expand_vector_parallel): Likewise.
16557 * tree-vect-stmts.c (vectorizable_load): Likewise.
16558 (vectorizable_store): Likewise.
16560 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16561 Alan Hayward <alan.hayward@arm.com>
16562 David Sherwood <david.sherwood@arm.com>
16564 * coretypes.h (pod_mode): New type.
16565 (scalar_int_mode_pod): New typedef.
16566 * machmode.h (pod_mode): New class.
16567 (int_n_data_t::m): Change type to scalar_int_mode_pod.
16568 * genmodes.c (emit_mode_int_n): Update accordingly.
16569 * lower-subreg.h (target_lower_subreg): Change type to
16570 scalar_int_mode_pod.
16571 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
16572 scalar_int_mode_pod.
16574 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16575 Alan Hayward <alan.hayward@arm.com>
16576 David Sherwood <david.sherwood@arm.com>
16578 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
16579 machine_mode to scalar_int_mode.
16580 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
16581 (rs6000_option_override_internal): Remove cast to int.
16582 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
16583 machine_mode to scalar_int_mode.
16584 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
16585 (rs6000_option_override_internal): Remove cast to int.
16586 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
16587 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
16589 * config/s390/s390.c (s390_expand_builtin): Likewise.
16590 * coretypes.h (scalar_int_mode): New type.
16591 (opt_scalar_int_mode): New typedef.
16592 * machmode.h (scalar_int_mode): New class.
16593 (scalar_int_mode::includes_p): New function.
16594 (byte_mode): Change type to scalar_int_mode.
16595 (word_mode): Likewise.
16596 (ptr_mode): Likewise.
16597 * emit-rtl.c (byte_mode): Likewise.
16598 (word_mode): Likewise.
16599 (ptr_mode): Likewise.
16600 (init_derived_machine_modes): Update accordingly.
16601 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
16602 and MODE_PARTIAL_INT.
16603 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
16604 opt_scalar_int_mode.
16606 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16607 Alan Hayward <alan.hayward@arm.com>
16608 David Sherwood <david.sherwood@arm.com>
16610 * target.def (libgcc_floating_mode_supported_p): Take a
16612 * doc/tm.texi: Regenerate.
16613 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
16615 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16616 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
16619 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16620 Alan Hayward <alan.hayward@arm.com>
16621 David Sherwood <david.sherwood@arm.com>
16623 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
16624 * doc/tm.texi: Regenerate.
16625 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
16626 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
16627 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
16628 * targhooks.h (default_floatn_mode): Likewise.
16629 * targhooks.c (default_floatn_mode): Likewise.
16630 * tree.c (build_common_tree_nodes): Update accordingly.
16632 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16633 Alan Hayward <alan.hayward@arm.com>
16634 David Sherwood <david.sherwood@arm.com>
16636 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
16637 (mode_iterator::iterate_p): Likewise.
16638 (mode_iterator::get_wider): Likewise.
16639 * expr.c (init_expr_target): Use opt_scalar_float_mode.
16641 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16642 Alan Hayward <alan.hayward@arm.com>
16643 David Sherwood <david.sherwood@arm.com>
16645 * coretypes.h (opt_scalar_float_mode): New typedef.
16646 * machmode.h (float_mode_for_size): New function.
16647 * emit-rtl.c (double_mode): Delete.
16648 (init_emit_once): Use float_mode_for_size.
16649 * stor-layout.c (layout_type): Likewise.
16650 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
16652 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16653 Alan Hayward <alan.hayward@arm.com>
16654 David Sherwood <david.sherwood@arm.com>
16656 * output.h (assemble_real): Take a scalar_float_mode.
16657 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
16658 * config/arm/arm.md (consttable_4): Likewise.
16659 (consttable_8): Likewise.
16660 (consttable_16): Likewise.
16661 * config/mips/mips.md (consttable_float): Likewise.
16662 * config/s390/s390.c (s390_output_pool_entry): Likewise.
16663 * varasm.c (assemble_real): Take a scalar_float_mode.
16664 (output_constant_pool_2): Update accordingly.
16665 (output_constant): Likewise.
16667 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16668 Alan Hayward <alan.hayward@arm.com>
16669 David Sherwood <david.sherwood@arm.com>
16671 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
16672 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
16673 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
16674 (native_encode_real): Likewise.
16675 (native_interpret_real): Likewise.
16676 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
16677 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16679 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16680 Alan Hayward <alan.hayward@arm.com>
16681 David Sherwood <david.sherwood@arm.com>
16683 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
16684 <scalar_float_mode>. Simplify.
16685 (gen_extend_conv_libfunc): Likewise.
16687 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16688 Alan Hayward <alan.hayward@arm.com>
16689 David Sherwood <david.sherwood@arm.com>
16691 * coretypes.h (scalar_float_mode): New type.
16692 * machmode.h (mode_traits::from_int): Use machine_mode if
16693 USE_ENUM_MODES is defined.
16694 (is_a): New function.
16696 (dyn_cast): Likewise.
16697 (scalar_float_mode): New class.
16698 (scalar_float_mode::includes_p): New function.
16699 (is_float_mode): Likewise.
16700 * gdbhooks.py (MachineModePrinter): New class.
16701 (build_pretty_printer): Use it for scalar_float_mode.
16702 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
16703 (format_helper::format_helper): Turn into a template.
16704 * genmodes.c (get_mode_class): New function.
16705 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
16706 or machine_mode if none.
16707 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
16708 as_a <scalar_float_mode>.
16709 * dwarf2out.c (mem_loc_descriptor): Likewise.
16710 (insert_float): Likewise.
16711 (add_const_value_attribute): Likewise.
16712 * simplify-rtx.c (simplify_immed_subreg): Likewise.
16713 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
16714 (expand_unop): Update accordingly.
16715 (expand_abs_nojump): Likewise.
16716 (expand_copysign_absneg): Take a scalar_float_mode.
16717 (expand_copysign_bit): Likewise.
16718 (expand_copysign): Update accordingly.
16720 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16721 Alan Hayward <alan.hayward@arm.com>
16722 David Sherwood <david.sherwood@arm.com>
16724 * coretypes.h (opt_mode): New class.
16725 * machmode.h (opt_mode): Likewise.
16726 (opt_mode::else_void): New function.
16727 (opt_mode::require): Likewise.
16728 (opt_mode::exists): Likewise.
16729 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
16730 (GET_MODE_2XWIDER_MODE): Likewise.
16731 (mode_iterator::get_wider): Update accordingly.
16732 (mode_iterator::get_2xwider): Likewise.
16733 (mode_iterator::get_known_wider): Likewise, turning into a template.
16734 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
16735 forcing a wider mode to exist.
16736 * config/cr16/cr16.h (LONG_REG_P): Likewise.
16737 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16738 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
16739 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
16740 * lower-subreg.c (init_lower_subreg): Likewise.
16741 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
16742 on the final iteration.
16743 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
16744 a wider mode exists before asking for a move pattern.
16745 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
16746 forcing a wider mode to exist.
16747 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
16748 returning false if no such mode exists.
16749 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
16750 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
16751 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
16752 Avoid checking for a MODE_INT if we already know the mode is not a
16754 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
16755 forcing a wider mode to exist.
16756 (expmed_mult_highpart_optab): Likewise.
16757 (expmed_mult_highpart): Likewise.
16758 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
16760 * lto-streamer-in.c (lto_input_mode_table): Likewise.
16761 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
16762 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16763 * internal-fn.c (expand_mul_overflow): Update use of
16764 GET_MODE_2XWIDER_MODE.
16765 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16766 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
16767 GET_MODE_WIDER_MODE.
16768 (convert_plusminus_to_widen): Likewise.
16769 * tree-switch-conversion.c (array_value_type): Likewise.
16770 * var-tracking.c (emit_note_insn_var_location): Likewise.
16771 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16772 Return false inside rather than outside the loop if no wider mode
16774 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
16775 and GET_MODE_2XWIDER_MODE
16776 (can_compare_p): Use else_void.
16777 * gdbhooks.py (OptMachineModePrinter): New class.
16778 (build_pretty_printer): Use it for opt_mode.
16780 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16781 Alan Hayward <alan.hayward@arm.com>
16782 David Sherwood <david.sherwood@arm.com>
16784 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
16785 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
16787 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16788 Alan Hayward <alan.hayward@arm.com>
16789 David Sherwood <david.sherwood@arm.com>
16791 * machmode.h (mode_traits): New structure.
16792 (get_narrowest_mode): New function.
16793 (mode_iterator::start): Likewise.
16794 (mode_iterator::iterate_p): Likewise.
16795 (mode_iterator::get_wider): Likewise.
16796 (mode_iterator::get_known_wider): Likewise.
16797 (mode_iterator::get_2xwider): Likewise.
16798 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
16799 (FOR_EACH_MODE): Likewise.
16800 (FOR_EACH_MODE_FROM): Likewise.
16801 (FOR_EACH_MODE_UNTIL): Likewise.
16802 (FOR_EACH_WIDER_MODE): Likewise.
16803 (FOR_EACH_2XWIDER_MODE): Likewise.
16804 * builtins.c (expand_builtin_strlen): Use new mode iterators.
16805 * combine.c (simplify_comparison): Likewise
16806 * config/i386/i386.c (type_natural_mode): Likewise.
16807 * cse.c (cse_insn): Likewise.
16808 * dse.c (find_shift_sequence): Likewise.
16809 * emit-rtl.c (init_derived_machine_modes): Likewise.
16810 (init_emit_once): Likewise.
16811 * explow.c (hard_function_value): Likewise.
16812 * expmed.c (extract_fixed_bit_field_1): Likewise.
16813 (extract_bit_field_1): Likewise.
16814 (expand_divmod): Likewise.
16815 (emit_store_flag_1): Likewise.
16816 * expr.c (init_expr_target): Likewise.
16817 (convert_move): Likewise.
16818 (alignment_for_piecewise_move): Likewise.
16819 (widest_int_mode_for_size): Likewise.
16820 (emit_block_move_via_movmem): Likewise.
16821 (copy_blkmode_to_reg): Likewise.
16822 (set_storage_via_setmem): Likewise.
16823 (compress_float_constant): Likewise.
16824 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16825 * optabs-query.c (get_best_extraction_insn): Likewise.
16826 * optabs.c (expand_binop): Likewise.
16827 (expand_twoval_unop): Likewise.
16828 (expand_twoval_binop): Likewise.
16829 (widen_leading): Likewise.
16830 (widen_bswap): Likewise.
16831 (expand_parity): Likewise.
16832 (expand_unop): Likewise.
16833 (prepare_cmp_insn): Likewise.
16834 (prepare_float_lib_cmp): Likewise.
16835 (expand_float): Likewise.
16836 (expand_fix): Likewise.
16837 (expand_sfix_optab): Likewise.
16838 * postreload.c (move2add_use_add2_insn): Likewise.
16839 * reg-stack.c (reg_to_stack): Likewise.
16840 * reginfo.c (choose_hard_reg_mode): Likewise.
16841 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16842 * stor-layout.c (mode_for_size): Likewise.
16843 (smallest_mode_for_size): Likewise.
16844 (mode_for_vector): Likewise.
16845 (finish_bitfield_representative): Likewise.
16846 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
16847 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
16848 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16849 * var-tracking.c (prepare_call_arguments): Likewise.
16851 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16852 Alan Hayward <alan.hayward@arm.com>
16853 David Sherwood <david.sherwood@arm.com>
16855 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
16856 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
16857 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
16858 * machmode.h (mode_size): Move earlier in file.
16859 (mode_precision): Likewise.
16860 (mode_inner): Likewise.
16861 (mode_nunits): Likewise.
16862 (mode_unit_size): Likewise.
16863 (unit_unit_precision): Likewise.
16864 (mode_wider): Likewise.
16865 (mode_2xwider): Likewise.
16866 (machine_mode): New class.
16867 (mode_to_bytes): New function.
16868 (mode_to_bits): Likewise.
16869 (mode_to_precision): Likewise.
16870 (mode_to_inner): Likewise.
16871 (mode_to_unit_size): Likewise.
16872 (mode_to_unit_precision): Likewise.
16873 (mode_to_nunits): Likewise.
16874 (GET_MODE_SIZE): Use mode_to_bytes.
16875 (GET_MODE_BITSIZE): Use mode_to_bits.
16876 (GET_MODE_PRECISION): Use mode_to_precision.
16877 (GET_MODE_INNER): Use mode_to_inner.
16878 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
16879 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
16880 (GET_MODE_NUNITS): Use mode_to_nunits.
16881 * system.h (ALWAYS_INLINE): New macro.
16882 * config/powerpcspe/powerpcspe-c.c
16883 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
16884 int for arg1_mode and arg2_mode.
16886 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16887 Alan Hayward <alan.hayward@arm.com>
16888 David Sherwood <david.sherwood@arm.com>
16890 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16891 Prefix mode names with E_ in case statements.
16892 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16893 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
16894 (aarch64_split_simd_move): Likewise.
16895 (aarch64_gen_storewb_pair): Likewise.
16896 (aarch64_gen_loadwb_pair): Likewise.
16897 (aarch64_gen_store_pair): Likewise.
16898 (aarch64_gen_load_pair): Likewise.
16899 (aarch64_get_condition_code_1): Likewise.
16900 (aarch64_constant_pool_reload_icode): Likewise.
16901 (get_rsqrte_type): Likewise.
16902 (get_rsqrts_type): Likewise.
16903 (get_recpe_type): Likewise.
16904 (get_recps_type): Likewise.
16905 (aarch64_gimplify_va_arg_expr): Likewise.
16906 (aarch64_simd_container_mode): Likewise.
16907 (aarch64_emit_load_exclusive): Likewise.
16908 (aarch64_emit_store_exclusive): Likewise.
16909 (aarch64_expand_compare_and_swap): Likewise.
16910 (aarch64_gen_atomic_cas): Likewise.
16911 (aarch64_emit_bic): Likewise.
16912 (aarch64_emit_atomic_swap): Likewise.
16913 (aarch64_emit_atomic_load_op): Likewise.
16914 (aarch64_evpc_trn): Likewise.
16915 (aarch64_evpc_uzp): Likewise.
16916 (aarch64_evpc_zip): Likewise.
16917 (aarch64_evpc_ext): Likewise.
16918 (aarch64_evpc_rev): Likewise.
16919 (aarch64_evpc_dup): Likewise.
16920 (aarch64_gen_ccmp_first): Likewise.
16921 (aarch64_gen_ccmp_next): Likewise.
16922 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
16923 (alpha_emit_xfloating_libcall): Likewise.
16924 (emit_insxl): Likewise.
16925 (alpha_arg_type): Likewise.
16926 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
16927 (arc_preferred_simd_mode): Likewise.
16928 (arc_secondary_reload): Likewise.
16929 (get_arc_condition_code): Likewise.
16930 (arc_print_operand): Likewise.
16931 (arc_legitimate_constant_p): Likewise.
16932 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16933 * config/arc/arc.md (casesi_load): Likewise.
16934 (casesi_compact_jump): Likewise.
16935 * config/arc/predicates.md (proper_comparison_operator): Likewise.
16936 (cc_use_register): Likewise.
16937 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16938 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
16939 (arm_init_iwmmxt_builtins): Likewise.
16940 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
16941 (neon_expand_vector_init): Likewise.
16942 (arm_attr_length_move_neon): Likewise.
16943 (maybe_get_arm_condition_code): Likewise.
16944 (arm_emit_vector_const): Likewise.
16945 (arm_preferred_simd_mode): Likewise.
16946 (arm_output_iwmmxt_tinsr): Likewise.
16947 (thumb1_output_casesi): Likewise.
16948 (thumb2_output_casesi): Likewise.
16949 (arm_emit_load_exclusive): Likewise.
16950 (arm_emit_store_exclusive): Likewise.
16951 (arm_expand_compare_and_swap): Likewise.
16952 (arm_evpc_neon_vuzp): Likewise.
16953 (arm_evpc_neon_vzip): Likewise.
16954 (arm_evpc_neon_vrev): Likewise.
16955 (arm_evpc_neon_vtrn): Likewise.
16956 (arm_evpc_neon_vext): Likewise.
16957 (arm_validize_comparison): Likewise.
16958 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
16959 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
16960 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
16961 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
16962 (c6x_preferred_simd_mode): Likewise.
16963 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
16964 (epiphany_rtx_costs): Likewise.
16965 * config/epiphany/predicates.md (proper_comparison_operator):
16967 * config/frv/frv.c (condexec_memory_operand): Likewise.
16968 (frv_emit_move): Likewise.
16969 (output_move_single): Likewise.
16970 (output_condmove_single): Likewise.
16971 (frv_hard_regno_mode_ok): Likewise.
16972 (frv_matching_accg_mode): Likewise.
16973 * config/h8300/h8300.c (split_adds_subs): Likewise.
16974 (h8300_rtx_costs): Likewise.
16975 (h8300_print_operand): Likewise.
16976 (compute_mov_length): Likewise.
16977 (output_logical_op): Likewise.
16978 (compute_logical_op_length): Likewise.
16979 (compute_logical_op_cc): Likewise.
16980 (h8300_shift_needs_scratch_p): Likewise.
16981 (output_a_shift): Likewise.
16982 (compute_a_shift_length): Likewise.
16983 (compute_a_shift_cc): Likewise.
16984 (expand_a_rotate): Likewise.
16985 (output_a_rotate): Likewise.
16986 * config/i386/i386.c (classify_argument): Likewise.
16987 (function_arg_advance_32): Likewise.
16988 (function_arg_32): Likewise.
16989 (function_arg_64): Likewise.
16990 (function_value_64): Likewise.
16991 (ix86_gimplify_va_arg): Likewise.
16992 (ix86_legitimate_constant_p): Likewise.
16993 (put_condition_code): Likewise.
16994 (split_double_mode): Likewise.
16995 (ix86_avx256_split_vector_move_misalign): Likewise.
16996 (ix86_expand_vector_logical_operator): Likewise.
16997 (ix86_split_idivmod): Likewise.
16998 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
16999 (ix86_build_const_vector): Likewise.
17000 (ix86_build_signbit_mask): Likewise.
17001 (ix86_match_ccmode): Likewise.
17002 (ix86_cc_modes_compatible): Likewise.
17003 (ix86_expand_branch): Likewise.
17004 (ix86_expand_sse_cmp): Likewise.
17005 (ix86_expand_sse_movcc): Likewise.
17006 (ix86_expand_int_sse_cmp): Likewise.
17007 (ix86_expand_vec_perm_vpermi2): Likewise.
17008 (ix86_expand_vec_perm): Likewise.
17009 (ix86_expand_sse_unpack): Likewise.
17010 (ix86_expand_int_addcc): Likewise.
17011 (ix86_split_to_parts): Likewise.
17012 (ix86_vectorize_builtin_gather): Likewise.
17013 (ix86_vectorize_builtin_scatter): Likewise.
17014 (avx_vpermilp_parallel): Likewise.
17015 (inline_memory_move_cost): Likewise.
17016 (ix86_tieable_integer_mode_p): Likewise.
17017 (x86_maybe_negate_const_int): Likewise.
17018 (ix86_expand_vector_init_duplicate): Likewise.
17019 (ix86_expand_vector_init_one_nonzero): Likewise.
17020 (ix86_expand_vector_init_one_var): Likewise.
17021 (ix86_expand_vector_init_concat): Likewise.
17022 (ix86_expand_vector_init_interleave): Likewise.
17023 (ix86_expand_vector_init_general): Likewise.
17024 (ix86_expand_vector_set): Likewise.
17025 (ix86_expand_vector_extract): Likewise.
17026 (emit_reduc_half): Likewise.
17027 (ix86_emit_i387_round): Likewise.
17028 (ix86_mangle_type): Likewise.
17029 (ix86_expand_round_sse4): Likewise.
17030 (expand_vec_perm_blend): Likewise.
17031 (canonicalize_vector_int_perm): Likewise.
17032 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
17033 (expand_vec_perm_1): Likewise.
17034 (expand_vec_perm_interleave3): Likewise.
17035 (expand_vec_perm_even_odd_pack): Likewise.
17036 (expand_vec_perm_even_odd_1): Likewise.
17037 (expand_vec_perm_broadcast_1): Likewise.
17038 (ix86_vectorize_vec_perm_const_ok): Likewise.
17039 (ix86_expand_vecop_qihi): Likewise.
17040 (ix86_expand_mul_widen_hilo): Likewise.
17041 (ix86_expand_sse2_abs): Likewise.
17042 (ix86_expand_pextr): Likewise.
17043 (ix86_expand_pinsr): Likewise.
17044 (ix86_preferred_simd_mode): Likewise.
17045 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
17046 * config/i386/sse.md (*andnot<mode>3): Likewise.
17047 (<mask_codefor><code><mode>3<mask_name>): Likewise.
17048 (*<code><mode>3): Likewise.
17049 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
17050 (ia64_expand_atomic_op): Likewise.
17051 (ia64_arg_type): Likewise.
17052 (ia64_mode_to_int): Likewise.
17053 (ia64_scalar_mode_supported_p): Likewise.
17054 (ia64_vector_mode_supported_p): Likewise.
17055 (expand_vec_perm_broadcast): Likewise.
17056 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
17057 (iq2000_function_arg_advance): Likewise.
17058 (iq2000_function_arg): Likewise.
17059 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
17060 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
17061 (m68k_libcall_value): Likewise.
17062 (m68k_function_value): Likewise.
17063 (sched_attr_op_type): Likewise.
17064 * config/mcore/mcore.c (mcore_output_move): Likewise.
17065 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
17067 (microblaze_function_arg): Likewise.
17068 * config/mips/mips.c (mips16_build_call_stub): Likewise.
17069 (mips_print_operand): Likewise.
17070 (mips_mode_ok_for_mov_fmt_p): Likewise.
17071 (mips_vector_mode_supported_p): Likewise.
17072 (mips_preferred_simd_mode): Likewise.
17073 (mips_expand_vpc_loongson_even_odd): Likewise.
17074 (mips_expand_vec_unpack): Likewise.
17075 (mips_expand_vi_broadcast): Likewise.
17076 (mips_expand_vector_init): Likewise.
17077 (mips_expand_vec_reduc): Likewise.
17078 (mips_expand_msa_cmp): Likewise.
17079 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
17080 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
17081 (cc_flags_for_mode): Likewise.
17082 * config/msp430/msp430.c (msp430_print_operand): Likewise.
17083 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
17084 (nds32_output_casesi_pc_relative): Likewise.
17085 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17086 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
17087 (nvptx_gen_unpack): Likewise.
17088 (nvptx_gen_pack): Likewise.
17089 (nvptx_gen_shuffle): Likewise.
17090 (nvptx_gen_wcast): Likewise.
17091 (nvptx_preferred_simd_mode): Likewise.
17092 * config/pa/pa.c (pa_secondary_reload): Likewise.
17093 * config/pa/predicates.md (base14_operand): Likewise.
17094 * config/powerpcspe/powerpcspe-c.c
17095 (altivec_resolve_overloaded_builtin): Likewise.
17096 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
17098 (rs6000_preferred_simd_mode): Likewise.
17099 (output_vec_const_move): Likewise.
17100 (rs6000_expand_vector_extract): Likewise.
17101 (rs6000_split_vec_extract_var): Likewise.
17102 (reg_offset_addressing_ok_p): Likewise.
17103 (rs6000_legitimate_offset_address_p): Likewise.
17104 (rs6000_legitimize_address): Likewise.
17105 (rs6000_emit_set_const): Likewise.
17106 (rs6000_const_vec): Likewise.
17107 (rs6000_emit_move): Likewise.
17108 (spe_build_register_parallel): Likewise.
17109 (rs6000_darwin64_record_arg_recurse): Likewise.
17110 (swap_selector_for_mode): Likewise.
17111 (spe_init_builtins): Likewise.
17112 (paired_init_builtins): Likewise.
17113 (altivec_init_builtins): Likewise.
17114 (do_load_for_compare): Likewise.
17115 (rs6000_generate_compare): Likewise.
17116 (rs6000_expand_float128_convert): Likewise.
17117 (emit_load_locked): Likewise.
17118 (emit_store_conditional): Likewise.
17119 (rs6000_output_function_epilogue): Likewise.
17120 (rs6000_handle_altivec_attribute): Likewise.
17121 (rs6000_function_value): Likewise.
17122 (emit_fusion_gpr_load): Likewise.
17123 (emit_fusion_p9_load): Likewise.
17124 (emit_fusion_p9_store): Likewise.
17125 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
17126 (fusion_gpr_mem_load): Likewise.
17127 (fusion_addis_mem_combo_load): Likewise.
17128 (fusion_addis_mem_combo_store): Likewise.
17129 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
17130 (fusion_gpr_mem_load): Likewise.
17131 (fusion_addis_mem_combo_load): Likewise.
17132 (fusion_addis_mem_combo_store): Likewise.
17133 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17135 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
17136 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
17137 (rs6000_preferred_simd_mode): Likewise.
17138 (output_vec_const_move): Likewise.
17139 (rs6000_expand_vector_extract): Likewise.
17140 (rs6000_split_vec_extract_var): Likewise.
17141 (reg_offset_addressing_ok_p): Likewise.
17142 (rs6000_legitimate_offset_address_p): Likewise.
17143 (rs6000_legitimize_address): Likewise.
17144 (rs6000_emit_set_const): Likewise.
17145 (rs6000_const_vec): Likewise.
17146 (rs6000_emit_move): Likewise.
17147 (rs6000_darwin64_record_arg_recurse): Likewise.
17148 (swap_selector_for_mode): Likewise.
17149 (paired_init_builtins): Likewise.
17150 (altivec_init_builtins): Likewise.
17151 (rs6000_expand_float128_convert): Likewise.
17152 (emit_load_locked): Likewise.
17153 (emit_store_conditional): Likewise.
17154 (rs6000_output_function_epilogue): Likewise.
17155 (rs6000_handle_altivec_attribute): Likewise.
17156 (rs6000_function_value): Likewise.
17157 (emit_fusion_gpr_load): Likewise.
17158 (emit_fusion_p9_load): Likewise.
17159 (emit_fusion_p9_store): Likewise.
17160 * config/rx/rx.c (rx_gen_move_template): Likewise.
17161 (flags_from_mode): Likewise.
17162 * config/s390/predicates.md (s390_alc_comparison): Likewise.
17163 (s390_slb_comparison): Likewise.
17164 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
17165 (s390_vector_mode_supported_p): Likewise.
17166 (s390_cc_modes_compatible): Likewise.
17167 (s390_match_ccmode_set): Likewise.
17168 (s390_canonicalize_comparison): Likewise.
17169 (s390_emit_compare_and_swap): Likewise.
17170 (s390_branch_condition_mask): Likewise.
17171 (s390_rtx_costs): Likewise.
17172 (s390_secondary_reload): Likewise.
17173 (__SECONDARY_RELOAD_CASE): Likewise.
17174 (s390_expand_cs): Likewise.
17175 (s390_preferred_simd_mode): Likewise.
17176 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
17177 * config/sh/sh.c (sh_print_operand): Likewise.
17178 (dump_table): Likewise.
17179 (sh_secondary_reload): Likewise.
17180 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17181 * config/sh/sh.md (casesi_worker_1): Likewise.
17182 (casesi_worker_2): Likewise.
17183 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
17184 (fcc_comparison_operator): Likewise.
17185 * config/sparc/sparc.c (sparc_expand_move): Likewise.
17186 (emit_soft_tfmode_cvt): Likewise.
17187 (sparc_preferred_simd_mode): Likewise.
17188 (output_cbranch): Likewise.
17189 (sparc_print_operand): Likewise.
17190 (sparc_expand_vec_perm_bmask): Likewise.
17191 (vector_init_bshuffle): Likewise.
17192 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17193 (spu_vector_mode_supported_p): Likewise.
17194 (spu_expand_insv): Likewise.
17195 (spu_emit_branch_or_set): Likewise.
17196 (spu_handle_vector_attribute): Likewise.
17197 (spu_builtin_splats): Likewise.
17198 (spu_builtin_extract): Likewise.
17199 (spu_builtin_promote): Likewise.
17200 (spu_expand_sign_extend): Likewise.
17201 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17202 (tilegx_simd_int): Likewise.
17203 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
17204 (tilepro_simd_int): Likewise.
17205 * config/v850/v850.c (const_double_split): Likewise.
17206 (v850_print_operand): Likewise.
17207 (ep_memory_offset): Likewise.
17208 * config/vax/vax.c (vax_rtx_costs): Likewise.
17209 (vax_output_int_move): Likewise.
17210 (vax_output_int_add): Likewise.
17211 (vax_output_int_subtract): Likewise.
17212 * config/visium/predicates.md (visium_branch_operator): Likewise.
17213 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
17214 (visium_print_operand_address): Likewise.
17215 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17216 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
17217 (xtensa_expand_conditional_branch): Likewise.
17218 (xtensa_copy_incoming_a7): Likewise.
17219 (xtensa_output_literal): Likewise.
17220 * dfp.c (decimal_real_maxval): Likewise.
17221 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
17223 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17224 Alan Hayward <alan.hayward@arm.com>
17225 David Sherwood <david.sherwood@arm.com>
17227 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
17228 (mode_nunits_inline): Likewise.
17229 (mode_inner_inline): Likewise.
17230 (mode_unit_size_inline): Likewise.
17231 (mode_unit_precision_inline): Likewise.
17232 (emit_insn_modes_h): Likewise. Also emit a #define of the
17234 (emit_mode_wider): Add an E_ prefix to mode names.
17235 (emit_mode_complex): Likewise.
17236 (emit_mode_inner): Likewise.
17237 (emit_mode_adjustments): Likewise.
17238 (emit_mode_int_n): Likewise.
17239 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
17240 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
17241 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
17242 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
17243 (CRC32_BUILTIN, ENTRY): Likewise.
17244 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
17245 (aarch64_pop_regs): Likewise.
17246 (aarch64_process_components): Likewise.
17247 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
17248 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
17249 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
17250 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
17251 * config/arm/arm.c (arm_init_libfuncs): Likewise.
17252 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
17254 * config/i386/i386-builtin.def (pcmpestr): Likewise.
17255 (pcmpistr): Likewise.
17256 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
17257 * config/mmix/mmix.c (mmix_output_condition): Likewise.
17258 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17260 * config/rl78/rl78.c (mduc_regs): Likewise.
17261 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17262 (htm_expand_builtin): Likewise.
17263 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
17264 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
17265 * config/xtensa/xtensa.c (print_operand): Likewise.
17266 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
17267 (NUM_MODE_VECTOR_INT): Likewise.
17268 * genoutput.c (null_operand): Likewise.
17269 (output_operand_data): Likewise.
17270 * genrecog.c (print_parameter_value): Likewise.
17271 * lra.c (debug_operand_data): Likewise.
17273 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17275 * dwarf2out.c (scompare_loc_descriptor_wide)
17276 (scompare_loc_descriptor_narrow): New functions, split out from...
17277 (scompare_loc_descriptor): ...here.
17278 * expmed.c (emit_store_flag_int): New function, split out from...
17279 (emit_store_flag): ...here.
17281 2017-08-30 Richard Biener <rguenther@suse.de>
17283 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
17284 (dwarf2out_early_finish): Move setting of AT_pubnames from
17285 early debug output to early finish.
17287 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17289 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
17290 and -mdata-region to the assembler.
17292 2017-08-30 Richard Biener <rguenther@suse.de>
17294 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
17296 (gen_subprogram_die): Add DW_AT_object_pointer only early.
17297 (dwarf2out_early_global_decl): Only generate a DIE for the
17298 abstract origin if it doesn't already exist or is a declaration DIE.
17299 (resolve_addr): Do not add the linkage name twice when
17300 generating a stub DIE for the DW_TAG_GNU_call_site target.
17302 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17304 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17305 Use machine_mode rather than int for arg1_mode.
17307 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
17310 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
17311 that the second argument of the built-in functions to unpack
17312 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
17313 switch statement instead a lot of if statements.
17314 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
17315 Allow 64-bit values to be in Altivec registers as well as
17316 traditional floating point registers.
17317 (pack<mode>, FMOVE128_VSX iterator): Likewise.
17319 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
17321 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
17322 MAX_REGS_PER_ADDRESS == 1.
17324 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
17326 * config/i386/i386.opt (flag_fentry): Do not init to -1.
17327 * config/i386/i386.c (ix86_option_override_internal): Simplify
17328 setting of opts->x_flag_entry.
17330 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17331 Jakub Jelinek <jakub@redhat.com>
17332 Richard Biener <rguenther@suse.de>
17334 PR tree-optimization/81503
17335 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
17336 folded constant fits in the target type; reorder tests for clarity.
17338 2017-08-29 Martin Liska <mliska@suse.cz>
17340 * passes.def: Include pass_lower_switch.
17341 * stmt.c (dump_case_nodes): Remove and move to
17342 tree-switch-conversion.
17343 (case_values_threshold): Likewise.
17344 (expand_switch_as_decision_tree_p): Likewise.
17345 (emit_case_decision_tree): Likewise.
17346 (expand_case): Likewise.
17347 (balance_case_nodes): Likewise.
17348 (node_has_low_bound): Likewise.
17349 (node_has_high_bound): Likewise.
17350 (node_is_bounded): Likewise.
17351 (emit_case_nodes): Likewise.
17352 (struct simple_case_node): New struct.
17353 (add_case_node): Remove.
17354 (emit_case_dispatch_table): Use vector instead of case_list.
17355 (reset_out_edges_aux): Remove.
17356 (compute_cases_per_edge): Likewise.
17357 (expand_case): Build list of simple_case_node.
17358 (expand_sjlj_dispatch_table): Use it.
17359 * tree-switch-conversion.c (struct case_node): Moved from
17360 stmt.c and adjusted.
17361 (emit_case_nodes): Likewise.
17362 (node_has_low_bound): Likewise.
17363 (node_has_high_bound): Likewise.
17364 (node_is_bounded): Likewise.
17365 (case_values_threshold): Likewise.
17366 (reset_out_edges_aux): Likewise.
17367 (compute_cases_per_edge): Likewise.
17368 (add_case_node): Likewise.
17369 (dump_case_nodes): Likewise.
17370 (balance_case_nodes): Likewise.
17371 (expand_switch_as_decision_tree_p): Likewise.
17372 (emit_jump): Likewise.
17373 (emit_case_decision_tree): Likewise.
17374 (try_switch_expansion): Likewise.
17375 (do_jump_if_equal): Likewise.
17376 (emit_cmp_and_jump_insns): Likewise.
17377 (fix_phi_operands_for_edge): New function.
17378 (record_phi_operand_mapping): Likewise.
17379 (class pass_lower_switch): New pass.
17380 (pass_lower_switch::execute): New function.
17381 (make_pass_lower_switch): Likewise.
17382 (conditional_probability):
17383 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
17384 * tree-pass.h: Add make_pass_lower_switch.
17386 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17389 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
17392 2017-08-29 Richard Biener <rguenther@suse.de>
17394 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
17395 we do not add a DW_AT_inline attribute twice.
17396 (gen_subprogram_die): Remove code setting DW_AT_inline on
17397 DECL_ABSTRACT_P nodes.
17399 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
17401 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
17402 calls to internal functions.
17403 (gimplify_modify_expr): Likewise.
17404 * tree-call-cdce.c (use_internal_fn): Likewise.
17405 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
17406 (convert_to_divmod): Set the nothrow flag.
17407 * tree-if-conv.c (predicate_mem_writes): Likewise.
17408 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
17409 (vectorizable_call): Likewise.
17410 (vectorizable_store): Likewise.
17411 (vectorizable_load): Likewise.
17412 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
17413 (vect_recog_mask_conversion_pattern): Likewise.
17415 2017-08-29 Martin Liska <mliska@suse.cz>
17418 * gcc.c (driver_handle_option): Add new argument.
17419 * opts-common.c (handle_option): Pass
17420 target_option_override_hook.
17421 * opts-global.c (lang_handle_option): Add new option.
17422 (set_default_handlers): Add new argument.
17423 (decode_options): Likewise.
17424 * opts.c (target_handle_option): Likewise.
17425 (common_handle_option): Call target_option_override_hook.
17426 * opts.h (struct cl_option_handler_func): Add hook for
17427 target option override.
17428 (struct cl_option_handlers): Likewise.
17429 (set_default_handlers): Add new argument.
17430 (decode_options): Likewise.
17431 (common_handle_option): Likewise.
17432 (target_handle_option): Likewise.
17433 * toplev.c (toplev::main): Pass targetm.target_option.override
17436 2017-08-29 Richard Biener <rguenther@suse.de>
17437 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
17439 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
17440 life to the active subtree.
17442 2017-08-28 Jeff Law <law@redhat.com>
17444 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
17445 derive_equivalences.
17446 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
17447 Code moved into....
17448 (edge_info::derive_equivalences): New private member function
17450 * tree-ssa-dom.c (class edge_info): Changed from a struct
17451 to a class. Add ctor/dtor, methods and data members.
17452 (edge_info::edge_info): Renamed from allocate_edge_info.
17453 Initialize additional members.
17454 (edge_info::~edge_info): New.
17455 (free_dom_edge_info): Delete the edge info.
17456 (record_edge_info): Use new class & associated member functions.
17457 Tighten forms for testing for edge equivalences.
17458 (record_temporary_equivalences): Iterate over the simple
17459 equivalences rather than assuming there's only one per edge.
17460 (cprop_into_successor_phis): Iterate over the simple
17461 equivalences rather than assuming there's only one per edge.
17462 (optimize_stmt): Use operand_equal_p rather than pointer
17463 equality for mini-DSE code.
17465 2017-08-28 Nathan Sidwell <nathan@acm.org>
17467 * gcc.c (execute): Fold SIGPIPE handling into switch
17468 statement. Adjust internal error message.
17470 2017-08-28 Richard Biener <rguenther@suse.de>
17473 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
17474 Do nothing for removed DIEs.
17476 2017-08-28 Richard Biener <rguenther@suse.de>
17478 PR tree-optimization/81977
17479 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
17482 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
17485 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
17486 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
17487 using targetm.gen_mem_thread_fence.
17489 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
17492 * config/i386/i386.md (*<btsc><mode>): Change operand 2
17493 predicate to register_operand. Reorder operands.
17494 (*btr<mode>): Ditto.
17495 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
17496 (*btr<mode>_mask): Ditto.
17498 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
17500 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
17501 * config/rs6000/xmmintrin.h: New file.
17502 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
17504 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17507 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
17508 parameter and_insn and return it.
17509 (recombine_lvx_pattern): Insert a copy to ensure availability of
17510 the base register of the copied masking operation at the point of
17511 the instruction replacement.
17512 (recombine_stvx_pattern): Likewise.
17514 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
17516 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
17517 undocumented switches.
17518 (-mpower9-dform-vector): Likewise.
17519 (-mpower9-dform): Likewise.
17520 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
17521 comments to delete references to -mpower9-dform* switches.
17522 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
17523 Delete reference to -mpower9-dform* switches, test for
17524 -mpower9-vector instead.
17525 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
17526 (OTHER_P9_VECTOR_MASKS): Likewise.
17527 (POWERPC_MASKS): Likewise.
17528 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
17529 tests against -mpower9-dform* to -mpower9-vector. Delete code
17530 that checked for -mpower9-dform* consistancy with other options.
17531 Add test for -mpower9-misc to enable other power9 switches.
17532 (rs6000_init_hard_regno_mode_ok): Likewise.
17533 (rs6000_option_override_internal): Likewise.
17534 (rs6000_emit_prologue): Likewise.
17535 (rs6000_emit_epilogue): Likewise.
17536 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
17537 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
17538 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
17540 (emit_fusion_p9_store): Likewise.
17541 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
17542 resetting these macros if the assembler does not support ISA 3.0
17544 (TARGET_P9_DFORM_VECTOR): Likewise.
17545 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
17546 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
17548 2017-08-25 Alan Modra <amodra@gmail.com>
17550 PR rtl-optimization/81747
17551 * cse.c (cse_extended_basic_block): Don't attempt to record
17552 equivalences for degenerate conditional branches that branch
17553 to their fall-through.
17555 2017-08-24 Martin Sebor <msebor@redhat.com>
17557 PR middle-end/81908
17558 * gimple-fold.c (size_must_be_zero_p): New function.
17559 (gimple_fold_builtin_memory_op): Call it.
17561 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
17563 * config/rs6000/mm_malloc.h: New file.
17565 2017-08-24 Bin Cheng <bin.cheng@arm.com>
17567 PR tree-optimization/81913
17568 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
17569 analysis when either IVs in condition can wrap.
17571 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
17573 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
17574 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
17576 2017-08-24 Richard Biener <rguenther@suse.de>
17579 * targhooks.c (default_target_can_inline_p): Properly
17580 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
17581 is present and always compare.
17582 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
17583 infer -mfpmath=sse from TARGET_SSE_P.
17584 (ix86_can_inline_p): Properly use target_option_default_node when
17585 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
17587 2017-08-24 Richard Biener <rguenther@suse.de>
17590 * dwarf2out.c (output_die): Handle flag_generate_offload like
17592 (output_comp_unit): Likewise.
17593 (gen_array_type_die): Likewise.
17594 (dwarf2out_early_finish): Likewise.
17595 (note_variable_value_in_expr): Likewise.
17596 (dwarf2out_finish): Likewise. Adjust assert.
17597 * cgraphunit.c (symbol_table::compile): Move setting of
17598 flag_generate_offload earlier ...
17599 (symbol_table::finalize_compilation_unit): ... here, before
17600 early debug finalization.
17602 2017-08-24 Richard Biener <rguenther@suse.de>
17604 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
17605 and ipa-fnsummary.h.
17606 (ix86_can_inline_p): When ix86_fpmath flags do not match
17607 check whether the callee uses FP math at all.
17609 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
17611 PR middle-end/81931
17612 * tree-ssanames.c (get_nonzero_bits): Use element_precision
17613 instead of TYPE_PRECISION.
17615 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
17616 Alan Hayward <alan.hayward@arm.com>
17617 David Sherwood <david.sherwood@arm.com>
17619 * combine.c (make_extraction): Use subreg_offset_from_lsb.
17621 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
17623 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
17625 (ix86_frame::stack_realign_allocate): New field.
17626 (struct machine_frame_state): Modify comments.
17627 (machine_frame_state::sp_realigned_fp_end): New field.
17628 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
17629 layout calculation.
17630 (sp_valid_at): Add assertion to assure no attempt to access invalid
17631 offset of a realigned stack.
17632 (fp_valid_at): Likewise.
17633 (choose_baseaddr): Modify comments.
17634 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
17635 ix86_expand_prologue.
17636 (ix86_expand_prologue): Modify stack realignment and allocation.
17637 (ix86_expand_epilogue): Modify comments.
17638 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
17639 avx2_runtime, avx512f, and avx512f_runtime.
17641 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
17643 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
17644 (mstackrealign): Do not init to -1.
17645 * config/i386/i386.c (ix86_option_override_internal):
17646 Check opts_set, not opts when setting default value of
17647 opts->x_ix86_force_align_arg_pointer.
17649 2017-08-23 Richard Biener <rguenther@suse.de>
17651 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
17652 lang_hooks.decl_printable_name.
17653 * print-rtl-function.c (print_rtx_function): Likewise.
17654 * tree-pretty-print.c (dump_function_header): Likewise.
17656 2017-08-23 Richard Biener <rguenther@suse.de>
17659 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
17660 -g0 at compile-time.
17662 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17664 PR middle-end/19706
17665 * doc/sourcebuild.texi (Other hardware attributes):
17668 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17670 PR middle-end/19706
17671 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
17672 Fix single-use check.
17674 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
17676 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
17678 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17680 * doc/install.texi: Modify to add more details on running selected
17683 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17685 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
17686 is combined with -mabi=ms.
17687 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
17690 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17692 PR tree-optimization/81488
17693 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
17694 and cached_basis fields.
17695 (MAX_SPREAD): New constant.
17696 (alloc_cand_and_find_basis): Initialize new fields.
17697 (clear_visited): New function.
17698 (create_phi_basis_1): Rename from create_phi_basis, set visited
17699 and cached_basis fields.
17700 (create_phi_basis): New wrapper function.
17701 (phi_add_costs_1): Rename from phi_add_costs, add spread
17702 parameter, set visited field, short-circuit when limits reached.
17703 (phi_add_costs): New wrapper function.
17704 (record_phi_increments_1): Rename from record_phi_increments, set
17706 (record_phi_increments): New wrapper function.
17707 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
17708 (phi_incr_cost): New wrapper function.
17709 (all_phi_incrs_profitable_1): Rename from
17710 all_phi_incrs_profitable, set visited field.
17711 (all_phi_incrs_profitable): New wrapper function.
17713 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17714 Alan Hayward <alan.hayward@arm.com>
17715 David Sherwood <david.sherwood@arm.com>
17717 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
17718 that takes the outer and inner modes.
17719 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
17720 comparison as the canonical test for a paradoxical subreg.
17721 * combine.c (simplify_set): Use paradoxical_subreg_p.
17722 (make_extraction): Likewise.
17723 (force_to_mode): Likewise.
17724 (rtx_equal_for_field_assignment_p): Likewise.
17725 (gen_lowpart_for_combine): Likewise.
17726 (simplify_comparison): Likewise.
17727 * cse.c (equiv_constant): Likewise.
17728 * expmed.c (store_bit_field_1): Likewise.
17729 * final.c (alter_subreg): Likewise.
17730 * fwprop.c (propagate_rtx): Likewise.
17731 (forward_propagate_subreg): Likewise.
17732 * ira-conflicts.c (ira_build_conflicts): Likewise.
17733 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
17734 * lra-constraints.c (curr_insn_transform): Likewise.
17735 (split_reg): Likewise.
17736 * lra-eliminations.c (move_plus_up): Likewise.
17737 (lra_eliminate_regs_1): Likewise.
17738 * recog.c (general_operand): Likewise.
17739 * ree.c (combine_reaching_defs): Likewise.
17740 * reload.c (push_reload): Likewise.
17741 (find_reloads): Likewise.
17742 * reload1.c (elimination_effects): Likewise.
17743 (compute_reload_subreg_offset): Likewise.
17744 (choose_reload_regs): Likewise.
17745 * rtlanal.c (subreg_lsb_1): Likewise.
17746 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17747 (simplify_subreg): Likewise.
17748 * var-tracking.c (track_loc_p): Likewise.
17749 * emit-rtl.c (byte_lowpart_offset): Likewise.
17750 (paradoxical_subreg_p): Delete out-of-line definition.
17752 2017-08-22 Jeff Law <law@redhat.com>
17754 PR tree-optimization/81741
17755 PR tree-optimization/71947
17756 * tree-ssa-dom.c: Include tree-inline.h.
17757 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
17758 equivalences if one is more expensive to compute than the other.
17759 * tree-ssa-scopedtables.h (class const_or_copies): Make
17760 record_const_or_copy_raw method private.
17761 (class avail_exprs_stack): New method simplify_binary_operation.
17762 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
17763 avail_exprs_stack::simplify_binary_operation as needed.
17764 (avail_exprs_stack::simplify_binary_operation): New function.
17766 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
17768 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
17769 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
17770 (DOT_SYMBOLS): Likewise.
17771 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
17772 (RELOCATABLE_NEEDS_FIXUP): Likewise.
17773 (RS6000_ABI_NAME): Likewise.
17774 (TARGET_CMODEL): Likewise.
17775 (TOC_SECTION_ASM_OP): Likewise.
17776 (SET_CMODEL): New macro.
17777 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
17779 2017-08-22 Richard Biener <rguenther@suse.de>
17781 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
17782 to allow for free-lang-data replacements similar to verify_type_variant.
17784 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17785 Alan Hayward <alan.hayward@arm.com>
17786 David Sherwood <david.sherwood@arm.com>
17788 * config/aarch64/aarch64.md (casesi): Use DImode rather than
17789 VOIDmode for the LABEL_REF.
17791 2017-08-22 Richard Biener <rguenther@suse.de>
17793 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
17795 2017-08-22 Richard Biener <rguenther@suse.de>
17797 * common.opt (feliminate-dwarf2-dups): Ignore.
17798 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
17799 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
17800 same_die_p_wrap, compute_section_prefix,
17801 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
17802 (comdat_symbol_id, comdat_symbol_number): Likewise.
17803 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
17805 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
17806 (output_die): Mark unreachable path unreachable.
17807 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
17808 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
17809 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
17810 (dwarf2out_early_finish): Likewise.
17812 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
17814 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
17816 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
17819 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
17820 not VAR_P. Filter attribute warnings with OPT_Wattributes.
17821 (avr_attribute_table) <io, io_low, address>: Initialize
17822 .decl_required with true.
17824 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
17826 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
17827 undocumented debugging options.
17828 (-mvsx-scalar-double): Likewise.
17829 (-mallow-df-permute): Likewise.
17830 (-mvectorize-builtins): Likewise.
17831 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17832 (rs6000_builtin_vectorized_function): Likewise.
17833 (rs6000_builtin_md_vectorized_function): Likewise.
17834 (rs6000_opt_vars): Likewise.
17836 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
17839 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
17840 (*btrq_imm): Rename from *btrq.
17841 (*btcq_imm): Rename from *btcq.
17842 (btsc): New code attribute.
17843 (*<btsc><mode>): New insn pattern.
17844 (*btr<mode>): Ditto.
17845 (*<btsc><mode>_mask): New insn_and_split pattern.
17846 (*btr<mode>_mask): Ditto.
17848 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17849 Alan Hayward <alan.hayward@arm.com>
17850 David Sherwood <david.sherwood@arm.com>
17852 * function.c (pad_below): Simplify padding calculation.
17854 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17855 Alan Hayward <alan.hayward@arm.com>
17856 David Sherwood <david.sherwood@arm.com>
17858 * target.def (function_prologue): Remove frame size argument.
17859 (function_epilogue): Likewise.
17860 * doc/tm.texi: Regenerate.
17861 * final.c (final_start_function): Update call to function_prologue.
17862 (final_end_function): Update call to function_epilogue.
17863 (default_function_pro_epilogue): Remove frame size argument.
17864 * output.h (default_function_pro_epilogue): Likewise.
17865 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
17866 (arm_output_function_prologue): Likewise.
17867 * config/frv/frv.c (frv_function_prologue): Likewise.
17868 (frv_function_epilogue): Likewise.
17869 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
17870 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
17871 (ia64_output_function_epilogue): Likewise.
17872 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
17873 (m32r_output_function_epilogue): Likewise.
17874 * config/microblaze/microblaze.c (microblaze_function_prologue)
17875 (microblaze_function_epilogue): Likewise.
17876 * config/mips/mips.c (mips_output_function_prologue): Likewise.
17877 (mips_output_function_epilogue): Likewise.
17878 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
17879 (mmix_target_asm_function_epilogue): Likewise.
17880 * config/msp430/msp430.c (msp430_start_function): Likewise.
17881 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
17882 (nds32_asm_function_epilogue): Likewise.
17883 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
17884 * config/pa/pa.c (pa_output_function_prologue): Likewise.
17885 (pa_output_function_epilogue): Likewise.
17886 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
17887 (rs6000_output_function_epilogue): Likewise.
17888 * config/rl78/rl78.c (rl78_start_function): Likewise.
17889 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
17890 (rs6000_output_function_epilogue): Likewise.
17891 * config/rx/rx.c (rx_output_function_prologue): Likewise.
17892 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
17893 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
17894 (sparc_asm_function_epilogue): Likewise.
17896 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17898 * tree.h (type_has_mode_precision_p): New function.
17899 * convert.c (convert_to_integer_1): Use it.
17900 * expr.c (expand_expr_real_2): Likewise.
17901 (expand_expr_real_1): Likewise.
17902 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
17903 * match.pd: Likewise.
17904 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
17905 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
17906 * tree-tailcall.c (process_assignment): Likewise.
17907 * tree-vect-loop.c (vectorizable_reduction): Likewise.
17908 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
17909 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
17910 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17911 (vectorizable_assignment): Likewise.
17912 (vectorizable_shift): Likewise.
17913 (vectorizable_operation): Likewise.
17914 * tree-vrp.c (register_edge_assert_for_2): Likewise.
17916 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
17918 * match.pd: Add pow (C, x) simplification.
17920 2017-08-21 Richard Biener <rguenther@suse.de>
17922 PR tree-optimization/81900
17923 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
17924 for blocks with abnormal predecessors.
17925 (compute_antic): Do not set visited flag prematurely.
17927 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
17930 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
17932 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17934 * stor-layout.h (vector_type_mode): Move to...
17935 * tree.h (vector_type_mode): ...here.
17936 * stor-layout.c (vector_type_mode): Move to...
17937 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
17939 2017-08-21 Richard Biener <rguenther@suse.de>
17941 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
17942 register_external_die hooks.
17943 (debug_false_tree_charstarstar_uhwistar): Declare.
17944 (debug_nothing_tree_charstar_uhwi): Likewise.
17945 * debug.c (do_nothing_debug_hooks): Adjust.
17946 (debug_false_tree_charstarstar_uhwistar): New do nothing.
17947 (debug_nothing_tree_charstar_uhwi): Likewise.
17948 * dbxout.c (dbx_debug_hooks): Adjust.
17949 (xcoff_debug_hooks): Likewise.
17950 * sdbout.c (sdb_debug_hooks): Likewise.
17951 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
17952 * dwarf2out.c (macinfo_label_base): New global.
17953 (dwarf2out_register_external_die): New function for the
17954 register_external_die hook.
17955 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
17956 (dwarf2_debug_hooks): Use them.
17957 (dwarf2_lineno_debug_hooks): Adjust.
17958 (struct die_struct): Add with_offset flag.
17959 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
17960 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
17961 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
17962 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
17963 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
17964 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
17965 defining section names for the early LTO debug variants.
17966 (reset_indirect_string): New helper.
17967 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
17968 (print_dw_val): Add support for offsetted symbol references.
17969 (get_ultimate_context): Split out from is_cxx.
17970 (is_cxx): Use get_ultimate_context.
17971 (is_fortran): Add decl overload.
17972 (compute_comp_unit_symbol): Split out worker from
17973 compute_section_prefix.
17974 (compute_section_prefix): Call compute_comp_unit_symbol and
17975 set comdat_type_p here.
17976 (output_die): Skip DIE symbol output for the LTO added one.
17977 Handle DIE symbol references with offset.
17978 (output_comp_unit): Guard section name mangling properly.
17979 For LTO debug sections emit a symbol at the section beginning
17980 which we use to refer to its DIEs.
17981 (add_abstract_origin_attribute): For DIEs registered via
17982 dwarf2out_register_external_die directly refer to the early
17983 DIE rather than indirectly through the shadow one we created.
17984 Remove obsolete call to dwarf2out_abstract_function for
17985 non-function/block origins.
17986 (gen_array_type_die): When generating early LTO debug do
17987 not emit DW_AT_string_length.
17988 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
17989 late when in LTO. As suggested place a gcc_unreachable for
17990 the DECL_ABSTRACT_P case.
17991 (gen_subprogram_die): Avoid another specification DIE
17992 for early built declarations/definitions for the late LTO case.
17993 (gen_variable_die): Add type references for late duplicated VLA dies
17995 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
17996 we have the abstract instance already.
17997 (process_scope_var): Adjust decl DIE contexts in LTO which
17998 first puts them in limbo.
17999 (gen_decl_die): Do not generate type DIEs late apart from
18000 types for VLAs or for decls we do not yet have a DIE. Do not
18001 call dwarf2out_abstract_function late.
18002 (dwarf2out_early_global_decl): Make sure to create DIEs
18003 for abstract instances of a decl first.
18004 (dwarf2out_late_global_decl): Adjust comment.
18005 (output_macinfo_op): With multiple macro sections use
18006 macinfo_label_base to distinguish labels.
18007 (output_macinfo): Likewise. Update macinfo_label_base.
18008 Pass in the line info label.
18009 (note_variable_value_in_expr): When generating LTO resolve
18010 all variable values here by generating DIEs as needed.
18011 (init_sections_and_labels): Add early LTO debug flag parameter
18012 and generate different sections and names if set. Add generation
18013 counter for the labels so we can have multiple of them.
18014 (reset_dies): Helper to allow DIEs to be output multiple times.
18015 (dwarf2out_finish): When outputting DIEs to the fat part of an
18016 LTO object first reset DIEs.
18017 (dwarf2out_early_finish): Output early DIEs when generating LTO.
18018 (modified_type_die): Check for decl_ultimate_origin being self
18020 (gen_type_die_with_usage): Likewise.
18021 (gen_typedef_die): Allow decl_ultimate_origin being self.
18022 (set_decl_abstract_flags): Remove.
18023 (set_block_abstract_flags): Likewise.
18024 (dwarf2out_abstract_function): Treat the early generated DIEs
18025 as the abstract copy and only add DW_AT_inline and
18026 DW_AT_artificial here and call set_decl_origin_self.
18027 If the DIE has an abstract origin don't do anything.
18028 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
18029 if we have none yet (Go fails to build one, PR78628).
18030 (variably_modified_type_p): Prevent endless recursion for Ada
18031 cyclic pointer types.
18032 * lto-streamer-in.c: Include debug.h.
18033 (dref_queue): New global.
18034 (lto_read_tree_1): Stream in DIE references.
18035 (lto_input_tree): Register DIE references.
18036 (input_function): Stream DECL_DEBUG_ARGS.
18037 * lto-streamer-out.c: Include debug.h.
18038 (lto_write_tree_1): Output DIE references.
18039 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
18040 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
18041 (output_function): Stream DECL_DEBUG_ARGS.
18042 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
18043 Stream DECL_ABSTRACT_ORIGIN.
18044 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
18045 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
18046 DECL_CONTEXT for file-scope decls.
18047 * lto-streamer.h (struct dref_entry): Declare.
18048 (dref_queue): Likewise.
18049 * cfgexpand.c (pass_expand::execute): Do not call the
18050 outlining_inline_function hook here.
18051 * lto-wrapper.c (debug_obj): New global.
18052 (tool_cleanup): Unlink it if required.
18053 (debug_objcopy): New function.
18054 (run_gcc): Handle early debug sections in the IL files by
18055 extracting them to separate files, partially linkin them and
18056 feeding the result back as result to the linker.
18057 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
18058 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
18059 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
18060 sections into a separate segment.
18061 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
18063 (darwin_asm_dwarf_section): Likewise.
18064 (darwin_asm_output_dwarf_offset): Likewise.
18065 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
18067 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18068 Alan Hayward <alan.hayward@arm.com>
18069 David Sherwood <david.sherwood@arm.com>
18071 * read-md.h (md_reader::record_potential_iterator_use): Replace
18072 pointer argument with an rtx and an index.
18073 * read-rtl.c (iterator_group::apply_iterator): Likewise.
18074 (apply_mode_iterator): Likewise.
18075 (apply_code_iterator): Likewise.
18076 (apply_int_iterator): Likewise.
18077 (apply_subst_iterator): Likewise.
18078 (record_iterator_use): Likewise.
18079 (record_attribute_use): Likewise.
18080 (md_reader::record_potential_iterator_use): Likewise. Update calls
18081 to record_iterator_use and apply_iterator.
18082 (iterator_use): Replace ptr with x and index.
18083 (attribute_use): Likewise.
18084 (apply_attribute_uses): Update calls to apply_iterator.
18085 (apply_iterators): Likewise. Update initialization of iterator_use.
18086 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
18087 and record_potential_iterator_use.
18088 (rtx_reader::read_rtx_operand): Likewise.
18090 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18091 Alan Hayward <alan.hayward@arm.com>
18092 David Sherwood <david.sherwood@arm.com>
18094 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
18097 2017-08-21 Richard Biener <rguenther@suse.de>
18099 PR middle-end/81884
18100 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
18101 at struct end conservatively when comparing common bases.
18103 2017-08-21 Richard Biener <rguenther@suse.de>
18105 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
18106 (mem_ref_in_stmt): Remove.
18107 (determine_max_movement): Use ref index to get at the reference.
18108 (invariantness_dom_walker::before_dom_children): Deal with
18109 lim data already initialized.
18110 (gather_mem_refs_stmt): Initialize lim data and record ref index.
18112 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
18114 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
18115 (TARGET_ISA_ROUND): Ditto.
18116 (TARGET_ROUND): Ditto.
18117 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
18118 * config/i386/i386.md: Ditto.
18119 * config/i386/sse.md: Ditto.
18120 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
18121 with OPTION_MASK_ISA_SSE4_1.
18123 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
18126 * doc/extend.texi (x86 Built-in Functions): Correct the name of
18127 __builtin_ia32_lzcnt_u16.
18129 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
18132 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
18133 (rs6000_set_current_function): Rewrite function to use it.
18135 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
18138 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
18139 and TYPE_WARN_IF_NOT_ALIGN.
18140 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
18141 (handle_warn_if_not_align): New.
18142 (place_union_field): Call handle_warn_if_not_align.
18143 (place_field): Call handle_warn_if_not_align.
18144 Copy TYPE_WARN_IF_NOT_ALIGN.
18145 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
18146 (layout_type): Likewise.
18147 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
18149 (tree_decl_common): Add warn_if_not_align.
18150 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
18151 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
18152 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
18153 (DECL_WARN_IF_NOT_ALIGN): Likewise.
18154 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
18155 * doc/extend.texi: Document warn_if_not_aligned attribute.
18156 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
18158 2017-08-17 Martin Liska <mliska@suse.cz>
18161 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
18162 (get_data_dependence): Use it as pointer type.
18163 (distribute_loop): Likewise.
18165 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18167 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
18168 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
18169 (p8_vmrgow_<mode>_direct): New define_insn.
18170 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
18171 handle endianness for vmrgew and vmrgow permute patterns.
18173 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
18175 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
18176 * config/rs6000/rs6000-cpus.def: Remove comment.
18177 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
18178 (POWERPC_MASKS): Likewise.
18179 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
18180 use of TARGET_VSX_TIMODE.
18181 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
18182 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
18183 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
18184 (rs6000_option_override_internal): Remove dead code.
18185 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
18186 (rs6000_legitimize_reload_address): Likewise.
18187 (rs6000_legitimate_address_p): Likewise.
18188 (rs6000_opt_masks): Delete "vsx-timode".
18189 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
18190 from function comment.
18191 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
18192 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
18193 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
18195 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
18196 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
18197 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
18198 (VSX_TI): Likewise.
18200 (define_peephole2): Likewise.
18202 2017-08-17 Martin Sebor <msebor@redhat.com>
18205 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
18206 past the end of an array.
18207 (test_pp_format): Add test cases.
18209 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
18211 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
18212 missing ECF_NOTHROW flags.
18214 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
18217 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
18218 operands residing in integer registers.
18219 (*vsx_le_perm_load_<mode>): Likewise.
18220 (*vsx_le_perm_store_<mode>): Likewise.
18221 (define_peephole2): Add peepholes to optimize the above.
18223 2017-08-17 Marek Polacek <polacek@redhat.com>
18225 PR middle-end/81814
18226 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
18227 to mimic what shorten_compare did. Change the return type to bool.
18228 (fold_cond_expr_with_comparison): Update call to
18229 operand_equal_for_comparison_p.
18230 (fold_ternary_loc): Likewise.
18232 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
18234 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
18236 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
18238 2017-08-17 Richard Biener <rguenther@suse.de>
18240 * tree-ssa-structalias.c (solve_graph): When propagating
18241 to successors update the graphs succ edges and avoid duplicate work.
18243 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
18246 * config/i386/i386.c (ix86_option_override_internal): Save target
18247 specific options after ix86_stack_protector_guard_reg was changed.
18249 2017-08-17 Richard Biener <rguenther@suse.de>
18251 PR tree-optimization/81827
18252 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
18253 (new_var_info): Initialize it conservatively.
18254 (get_call_vi): Mark register vars.
18255 (new_scalar_tmp_constraint_exp): Likewise.
18256 (handle_rhs_call): Likewise.
18257 (handle_const_call): Likewise.
18258 (create_function_info_for): Likewise.
18259 (solve_constraints): Sort varinfos to separate register from
18260 non-register vars to pack points-to solution bitmaps during
18263 2017-08-17 Marek Polacek <polacek@redhat.com>
18265 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
18267 2017-08-17 Richard Biener <rguenther@suse.de>
18269 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
18270 to true when overflow is undefined and we saturated the result.
18272 2017-08-17 Alan Modra <amodra@gmail.com>
18275 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
18276 Don't use store multiple if only one reg needs saving.
18277 (interesting_frame_related_regno): New function.
18278 (rs6000_frame_related): Don't emit frame info for regs that
18280 (rs6000_emit_epilogue): Likewise.
18282 2017-08-16 Nathan Sidwell <nathan@acm.org>
18284 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
18285 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
18286 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
18287 * tree.c (free_lang_data_in_type): Use else-if chain. Always
18288 clear TYPE_LANG_1. Remove obsolete member-function stripping.
18289 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
18290 (verify_type): Adjust for TYPE_BINFO move.
18291 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
18292 process TYPE_BINFO directly.
18293 (hash_tree): Likewise.
18294 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
18296 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
18299 2017-08-16 David Malcolm <dmalcolm@redhat.com>
18301 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
18303 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
18306 * config/i386/i386.md (*anddi_1_btr): Change predicates of
18307 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
18308 Add ix86_binary_operator_ok to insn constraint.
18309 (*iordi_1_bts): Ditto.
18310 (*xordi_1_btc): Ditto.
18311 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
18312 Update corresponding peephole2 pattern.
18316 2017-08-16 Bin Cheng <bin.cheng@arm.com>
18318 PR tree-optimization/81832
18319 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
18320 copy loop header which has IFN_LOOP_DIST_ALIAS call.
18322 2017-08-16 Marek Polacek <polacek@redhat.com>
18325 * fold-const.c (fold_indirect_ref_1): Restore original behavior
18326 regarding size_zero_node.
18328 2017-08-16 Martin Liska <mliska@suse.cz>
18331 * config.gcc: Respect previously set extra_objs in case
18334 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
18336 PR tree-optimization/81835
18337 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
18338 the phi SSA_NAME. Check that the condition in a COND_EXPR does
18339 not depend on the phi.
18341 2017-08-16 Alan Modra <amodra@gmail.com>
18343 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
18346 2017-08-16 Alan Modra <amodra@gmail.com>
18348 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
18349 (save_reg_p): ..into this. Update all callers.
18350 (first_reg_to_save): Simplify.
18352 2017-08-16 Alan Modra <amodra@gmail.com>
18354 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
18357 2017-08-15 Joseph Myers <joseph@codesourcery.com>
18361 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
18362 constant count if that count is less than 32.
18364 2017-08-15 Nathan Sidwell <nathan@acm.org>
18366 * gcc.c (execute): Emit friendlier message if inferior is killed
18367 by an external cause.
18369 2017-08-15 Richard Biener <rguenther@suse.de>
18371 PR tree-optimization/81790
18372 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
18373 CONSTRUCTORs from simplifying and VN.
18375 2017-08-14 Martin Sebor <msebor@redhat.com>
18377 * builtin-attrs.def: Add comments.
18379 2017-08-14 Martin Sebor <msebor@redhat.com>
18382 * doc/extend.texi (attribute nonstring): Document new attribute.
18384 2017-08-14 Martin Sebor <msebor@redhat.com>
18387 * tree-diagnostic.c (default_tree_printer): Handle %G.
18388 * gimple-pretty-print.h (percent_G_format): Declare new function.
18389 * gimple-pretty-print.c (percent_G_format): Define.
18390 * tree-pretty-print.c (percent_K_format): Add argument.
18392 2017-08-14 Martin Sebor <msebor@redhat.com>
18394 PR translation/79998
18395 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
18396 Remove a stray space.
18398 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
18401 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
18402 (*iordi_1_bts): Ditto.
18403 (*xordi_1_btc): Ditto.
18405 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18408 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
18409 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18411 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
18412 quoted strings, and make more translator-friendly.
18413 (darwin_rs6000_override_options): Likewise.
18414 (rs6000_option_override_internal): Likewise.
18415 (rs6000_return_in_memory): Fix overlong line.
18416 (init_cmulative_args): Use quoted strings, and make more
18417 translator-friendly.
18418 (rs6000_pass_by_reference): Fix overlong line.
18419 (def_builtin): Use quoted strings.
18420 (altivec_expand_predicate_builtin): Use quoted strings, and make
18421 more translator-friendly.
18422 (htm_expand_builtin): Use quoted strings.
18423 (cpu_expand_builtin): Use quoted strings, and make more
18424 translator-friendly.
18425 (altivec_expand_builtin): Likewise.
18426 (paired_expand_predicate_builtin): Likewise.
18427 (rs6000_invalid_builtin): Likewise.
18428 (builtin_function_type): Use quoted strings.
18429 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
18430 more translator-friendly.
18431 (rs6000_trampoline_init): Likewise.
18432 (rs6000_handle_altivec_attribute): Likewise.
18433 (rs6000_inner_target_options): Use quoted strings.
18434 (rs6000_disable_incompatible_switches): Likewise.
18435 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
18436 strings, and make more translator-friendly.
18437 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
18439 2017-08-14 Bin Cheng <bin.cheng@arm.com>
18441 PR tree-optimization/81799
18442 * tree-loop-distribution.c (version_loop_by_alias_check): Force
18443 cond_expr to simple gimple operand.
18445 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
18447 PR middle-end/46932
18448 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
18450 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
18454 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
18455 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
18456 TARGET_GASISR_PROLOGUES.
18457 * config/avr/avr.c (avr_option_override): Same.
18458 (avr_pass_pre_proep::execute): Same.
18460 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
18463 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
18464 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
18466 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
18468 * config/i386/i386.md (*load_tp_<mode>): Redefine as
18469 define_insn_and_split. Split to a memory load from 0 in
18470 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
18471 using PTR mode iterator.
18472 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
18473 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
18474 (*add_tp_<mode>): Redefine as define_insn_and_split.
18475 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
18476 address space. Merge with *add_tp_x32 using PTR mode iterator.
18477 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
18478 Split to an add with a memory load from 0 in
18479 DEFAULT_TLS_SEG_REG address space.
18481 2017-08-12 Andrew Pinski <apinski@cavium.com>
18483 * config/aarch64/aarch64-option-extensions.def (rdma):
18484 Fix feature string to what Linux prints out in /proc/cpuinfo.
18486 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
18489 * dwarf2out.c (modified_type_die): For C typedef types that have
18490 an ultimate origin, process the ultimate origin instead of the
18492 (gen_typedef_die): Assert that input DECLs have no ultimate
18494 (gen_type_die_with_usage): For typedef variants that have an
18495 ultimate origin, just call gen_decl_die on the original DECL.
18496 (process_scope_var): Avoid creating DIEs for local typedefs and
18497 concrete static variables.
18499 2017-08-12 Alan Modra <amodra@gmail.com>
18503 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
18504 match gnu-user.h startfile.
18505 (ENDFILE_LINUX_SPEC): Similarly.
18507 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
18510 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
18512 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
18514 2017-08-11 Tamar Christina <tamar.christina@arm.com>
18515 * config/aarch64/aarch64.md (mov<mode>): Change.
18516 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
18517 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
18518 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
18520 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
18522 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
18523 for storage order barriers.
18525 2017-08-11 Martin Liska <mliska@suse.cz>
18528 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
18529 variables of void type.
18531 2017-08-11 Martin Liska <mliska@suse.cz>
18533 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
18534 TARGET_SUPPORTS_ALIASES.
18535 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
18536 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
18537 (optimize_weakref): Likewise.
18538 * symtab.c (symtab_node::noninterposable_alias): Likewise.
18539 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
18540 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
18542 2017-08-11 Martin Liska <mliska@suse.cz>
18545 * config/i386/i386.c (make_resolver_func): Do complete
18546 refactoring of the function.
18548 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
18551 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
18552 * config/i386/i386.c (ix86_stack_protect_guard): Use
18553 ix86_stack_protect_guard_symbol_str to generate varible declaration.
18554 * doc/invoke.texi (x86 Options): Document
18555 -mstack-protector-guard-symbol= option.
18557 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
18559 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
18560 * config/i386/i386.c (ix86_split_stack_guard): New function.
18561 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
18562 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
18563 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
18564 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
18565 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
18566 (split_stack_space_check): Call ix86_split_stack_guard.
18568 2017-08-10 Martin Sebor <msebor@redhat.com>
18570 * print-tree.c (print_node): Print location using the established
18572 Replace spaces with colons.
18573 (debug_raw, debug): Ditto.
18575 2017-08-10 Martin Sebor <msebor@redhat.com>
18578 * pretty-print.c (pp_format): Correct the handling of %s precision.
18580 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
18583 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18585 (ix86_finalize_stack_frame_flags): This. Also clear
18586 frame_pointer_needed if -fno-omit-frame-pointer is used without
18588 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18589 with ix86_finalize_stack_frame_flags.
18590 (ix86_expand_epilogue): Likewise.
18591 (ix86_expand_split_stack_prologue): Likewise.
18592 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
18594 2017-08-10 Martin Liska <mliska@suse.cz>
18597 * c-attribs.c (handle_target_attribute):
18598 Report warning for an empty string argument of target attribute.
18600 2017-08-09 Jakub Jelinek <jakub@redhat.com>
18603 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
18605 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
18606 or DECL_NONLOCAL labels.
18607 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
18608 or DECL_NONLOCAL labels here.
18610 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
18612 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
18613 to indicate when early gimple folding has been disabled.
18614 (rs6000_gimple_fold_builtin): Add debug content.
18615 (rs6000_invalid_builtin): Fix whitespace.
18616 (rs6000_expand_builtin): Fix whitespace.
18617 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
18619 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
18622 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
18623 SAVE_MULTIPLE if not all the registers that saves, should be saved.
18625 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
18627 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
18628 (qdf24xx): Likewise.
18629 * config/aarch64/aarch64.md: Include falkor.md.
18630 * config/aarch64/falkor.md: New.
18632 2017-08-09 Marek Polacek <polacek@redhat.com>
18635 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
18636 * diagnostic.c (emit_diagnostic): Add a comment.
18637 (emit_diagnostic_valist): New function.
18639 2017-08-09 Marek Polacek <polacek@redhat.com>
18642 * input.c (make_location): New overload.
18643 * input.h (make_location): Declare.
18645 2017-08-08 Alan Modra <amodra@gmail.com>
18646 H.J. Lu <hongjiu.lu@intel.com>
18649 * gcc.c (NO_PIE_SPEC): Delete.
18650 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
18652 (LINK_PIE_SPEC): ..to here.
18653 (LINK_COMMAND_SPEC): Support -no-pie.
18654 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
18655 chain of crtbegin*.o selection, update for PIE_SPEC changes and
18657 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
18658 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
18659 (ENDFILE_CRTEND_SPEC): Similarly.
18661 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
18664 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
18665 (mstack-protector-guard-offset=): Ditto.
18666 * config/i386/i386.c (ix86_option_override): Handle
18667 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
18669 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
18670 ix86_stack_protect_guard_offset variables.
18671 (TARGET_STACK_PROTECT_GUARD): Always define.
18672 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
18673 and -mstack-protector-guard-offset= options.
18675 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18677 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
18678 boundary case for the last candidate.
18680 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18682 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
18683 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
18685 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18687 PR middle-end/19706
18688 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
18689 * config/aarch64/aarch64-builtins.c
18690 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
18691 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
18692 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
18694 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18695 Andrew Pinski <pinskia@gmail.com>
18697 PR middle-end/19706
18698 * internal-fn.def (XORSIGN): New.
18699 * optabs.def (xorsign_optab): New.
18700 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
18701 (convert_expand_mult_copysign): New.
18702 (pass_optimize_widening_mul::execute): Call
18703 convert_expand_mult_copysign.
18705 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18707 PR tree-optimization/81354
18708 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
18709 Insert on edges rather than explicitly creating landing pads.
18710 (analyze_candidates_and_replace): Commit edge inserts.
18712 2017-08-08 Richard Biener <rguenther@suse.de>
18714 PR middle-end/81719
18715 * tree-ssa-loop-niter.c: Include tree-dfa.h.
18716 (expand_simple_operations): Also look through ADDR_EXPRs with
18717 MEM_REF bases treating them as POINTER_PLUS_EXPR.
18719 2017-08-08 Richard Biener <rguenther@suse.de>
18721 PR tree-optimization/81723
18722 * tree-vect-slp.c (struct bst_traits): New hash traits.
18723 (bst_fail): New global.
18724 (vect_build_slp_tree_2): New worker, split out from ...
18725 (vect_build_slp_tree): ... this now wrapping it with using
18726 bst_fail set to cache SLP tree build fails. Properly handle
18728 (vect_analyze_slp_instance): Allocate and free bst_fail.
18730 2017-08-08 Martin Liska <mliska@suse.cz>
18733 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
18734 LABEL_DECLs that can be from a different function.
18736 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18738 PR tree-optimization/81744
18739 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
18740 loop's number of iterations.
18742 2017-08-08 Martin Liska <mliska@suse.cz>
18744 * asan.c: Include header files.
18745 * attribs.c (build_decl_attribute_variant): New function moved
18747 (build_type_attribute_qual_variant): Likewise.
18748 (cmp_attrib_identifiers): Likewise.
18749 (simple_cst_list_equal): Likewise.
18750 (omp_declare_simd_clauses_equal): Likewise.
18751 (attribute_value_equal): Likewise.
18752 (comp_type_attributes): Likewise.
18753 (build_type_attribute_variant): Likewise.
18754 (lookup_ident_attribute): Likewise.
18755 (remove_attribute): Likewise.
18756 (merge_attributes): Likewise.
18757 (merge_type_attributes): Likewise.
18758 (merge_decl_attributes): Likewise.
18759 (merge_dllimport_decl_attributes): Likewise.
18760 (handle_dll_attribute): Likewise.
18761 (attribute_list_equal): Likewise.
18762 (attribute_list_contained): Likewise.
18763 * attribs.h (lookup_attribute): New function moved from tree.[ch].
18764 (lookup_attribute_by_prefix): Likewise.
18765 * bb-reorder.c: Include header files.
18766 * builtins.c: Likewise.
18767 * calls.c: Likewise.
18768 * cfgexpand.c: Likewise.
18769 * cgraph.c: Likewise.
18770 * cgraphunit.c: Likewise.
18771 * convert.c: Likewise.
18772 * dwarf2out.c: Likewise.
18773 * final.c: Likewise.
18774 * fold-const.c: Likewise.
18775 * function.c: Likewise.
18776 * gimple-expr.c: Likewise.
18777 * gimple-fold.c: Likewise.
18778 * gimple-pretty-print.c: Likewise.
18779 * gimple.c: Likewise.
18780 * gimplify.c: Likewise.
18781 * hsa-common.c: Likewise.
18782 * hsa-gen.c: Likewise.
18783 * internal-fn.c: Likewise.
18784 * ipa-chkp.c: Likewise.
18785 * ipa-cp.c: Likewise.
18786 * ipa-devirt.c: Likewise.
18787 * ipa-fnsummary.c: Likewise.
18788 * ipa-inline.c: Likewise.
18789 * ipa-visibility.c: Likewise.
18791 * lto-cgraph.c: Likewise.
18792 * omp-expand.c: Likewise.
18793 * omp-general.c: Likewise.
18794 * omp-low.c: Likewise.
18795 * omp-offload.c: Likewise.
18796 * omp-simd-clone.c: Likewise.
18797 * opts-global.c: Likewise.
18798 * passes.c: Likewise.
18799 * predict.c: Likewise.
18800 * sancov.c: Likewise.
18801 * sanopt.c: Likewise.
18802 * symtab.c: Likewise.
18803 * toplev.c: Likewise.
18804 * trans-mem.c: Likewise.
18805 * tree-chkp.c: Likewise.
18806 * tree-eh.c: Likewise.
18807 * tree-into-ssa.c: Likewise.
18808 * tree-object-size.c: Likewise.
18809 * tree-parloops.c: Likewise.
18810 * tree-profile.c: Likewise.
18811 * tree-ssa-ccp.c: Likewise.
18812 * tree-ssa-live.c: Likewise.
18813 * tree-ssa-loop.c: Likewise.
18814 * tree-ssa-sccvn.c: Likewise.
18815 * tree-ssa-structalias.c: Likewise.
18816 * tree-ssa.c: Likewise.
18817 * tree-streamer-in.c: Likewise.
18818 * tree-vectorizer.c: Likewise.
18819 * tree-vrp.c: Likewise.
18820 * tsan.c: Likewise.
18821 * ubsan.c: Likewise.
18822 * varasm.c: Likewise.
18823 * varpool.c: Likewise.
18824 * tree.c: Remove functions moved to attribs.[ch].
18825 * tree.h: Likewise.
18826 * config/aarch64/aarch64.c: Add attrs.h header file.
18827 * config/alpha/alpha.c: Likewise.
18828 * config/arc/arc.c: Likewise.
18829 * config/arm/arm.c: Likewise.
18830 * config/avr/avr.c: Likewise.
18831 * config/bfin/bfin.c: Likewise.
18832 * config/c6x/c6x.c: Likewise.
18833 * config/cr16/cr16.c: Likewise.
18834 * config/cris/cris.c: Likewise.
18835 * config/darwin.c: Likewise.
18836 * config/epiphany/epiphany.c: Likewise.
18837 * config/fr30/fr30.c: Likewise.
18838 * config/frv/frv.c: Likewise.
18839 * config/ft32/ft32.c: Likewise.
18840 * config/h8300/h8300.c: Likewise.
18841 * config/i386/winnt.c: Likewise.
18842 * config/ia64/ia64.c: Likewise.
18843 * config/iq2000/iq2000.c: Likewise.
18844 * config/lm32/lm32.c: Likewise.
18845 * config/m32c/m32c.c: Likewise.
18846 * config/m32r/m32r.c: Likewise.
18847 * config/m68k/m68k.c: Likewise.
18848 * config/mcore/mcore.c: Likewise.
18849 * config/microblaze/microblaze.c: Likewise.
18850 * config/mips/mips.c: Likewise.
18851 * config/mmix/mmix.c: Likewise.
18852 * config/mn10300/mn10300.c: Likewise.
18853 * config/moxie/moxie.c: Likewise.
18854 * config/msp430/msp430.c: Likewise.
18855 * config/nds32/nds32-isr.c: Likewise.
18856 * config/nds32/nds32.c: Likewise.
18857 * config/nios2/nios2.c: Likewise.
18858 * config/nvptx/nvptx.c: Likewise.
18859 * config/pa/pa.c: Likewise.
18860 * config/pdp11/pdp11.c: Likewise.
18861 * config/powerpcspe/powerpcspe.c: Likewise.
18862 * config/riscv/riscv.c: Likewise.
18863 * config/rl78/rl78.c: Likewise.
18864 * config/rx/rx.c: Likewise.
18865 * config/s390/s390.c: Likewise.
18866 * config/sh/sh.c: Likewise.
18867 * config/sol2.c: Likewise.
18868 * config/sparc/sparc.c: Likewise.
18869 * config/spu/spu.c: Likewise.
18870 * config/stormy16/stormy16.c: Likewise.
18871 * config/tilegx/tilegx.c: Likewise.
18872 * config/tilepro/tilepro.c: Likewise.
18873 * config/v850/v850.c: Likewise.
18874 * config/vax/vax.c: Likewise.
18875 * config/visium/visium.c: Likewise.
18876 * config/xtensa/xtensa.c: Likewise.
18878 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18881 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
18882 constraints since the -mupper-regs-* switches have been
18884 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
18885 into a vector from a double word element that was extracted from
18886 another vector, and eliminate extra XXPERMDI instructions.
18887 (vsx_concat_<mode>_2): Likewise.
18888 (vsx_concat_<mode>_3): Likewise.
18889 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
18890 concat to allow optimizing inserts from previous extracts.
18892 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
18894 * config/i386/i386.c (ix86_stack_protect_guard): Generate
18895 memory reference to a SSP offset in TLS address space.
18896 (ix86_print_operand) <case '@'>: Remove.
18897 (ix86_print_operand_punct_valid_p): Remove '@' code.
18898 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
18899 UNSPEC_SP_TLS_TEST.
18900 (stack_tls_protect_set_<mode>): Remove.
18901 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
18902 (stack_tls_protect_test_<mode>): Remove.
18903 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
18905 2017-08-07 Olivier Hainque <hainque@adacore.com>
18908 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
18910 2017-08-07 Douglas Rupp <rupp@adacore.com>
18912 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
18913 variable was referenced as multidir in command.
18915 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18918 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
18921 2017-08-07 Martin Liska <mliska@suse.cz>
18923 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
18924 * config/rl78/rl78.c: Add include of attribs.h.
18925 * config/sh/sh.c: Likewise.
18926 * config/v850/v850.c: Likewise.
18928 2017-08-07 Tom de Vries <tom@codesourcery.com>
18930 PR middle-end/78266
18931 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
18933 2017-08-07 Martin Liska <mliska@suse.cz>
18935 * config/mips/mips.c: Include attribs.h.
18937 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
18940 * doc/invoke.texi: Document change in behvaior for -Ofast for
18943 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
18945 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
18947 (aarch64_pop_regs): Likewise.
18948 (aarch64_gen_load_pair): Likewise.
18949 (aarch64_save_callee_saves): Likewise.
18950 (aarch64_restore_callee_saves): Likewise.
18952 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18954 * config/i386/i386.c: Revert the last change.
18956 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18959 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18961 (ix86_finalize_stack_frame_flags): This. Also clear
18962 frame_pointer_needed if -fno-omit-frame-pointer is used without
18964 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18965 with ix86_finalize_stack_frame_flags.
18966 (ix86_expand_epilogue): Likewise.
18967 (ix86_expand_split_stack_prologue): Likewise.
18969 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18972 * config/i386/i386.c (get_builtin_code_for_version): Set priority
18973 to P_AES for Westmere.
18975 2017-08-07 Jonathan Yong <10walls@gmail.com>
18977 * config/i386/mingw.opt (fset-stack-executable): Removed.
18978 * config/i386/cygming.opt (fset-stack-executable): Moved
18980 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
18982 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
18984 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
18986 2017-08-07 Marek Polacek <polacek@redhat.com>
18988 PR middle-end/81737
18989 * fold-const.c (fold_indirect_ref_1): Check type_domain.
18991 2017-08-07 Martin Liska <mliska@suse.cz>
18993 * attribs.h (canonicalize_attr_name): New function.
18994 (cmp_attribs): Move from c-format.c and adjusted.
18995 (is_attribute_p): Moved from tree.h.
18996 * tree-inline.c: Add new includes.
18997 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
18998 (private_is_attribute_p): Remove.
18999 (private_lookup_attribute): Likewise.
19000 (private_lookup_attribute_by_prefix): Simplify.
19001 (remove_attribute): Use is_attribute_p.
19002 * tree.h: Remove removed declarations.
19004 2017-08-07 Jakub Jelinek <jakub@redhat.com>
19006 PR middle-end/81698
19007 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
19008 instead of computing it in the function. Formatting fix.
19009 (expand_case): Don't rely on default_edge being the first edge,
19010 clear it if removing it, pass default_edge to
19011 emit_case_dispatch_table.
19012 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
19015 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
19017 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
19018 insn in the function, emit NOP after the insn.
19020 2017-08-06 Tom de Vries <tom@codesourcery.com>
19022 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
19025 2017-08-06 Tom de Vries <tom@codesourcery.com>
19027 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
19030 2017-08-04 Yury Gribov <tetra2005@gmail.com>
19032 PR tree-optimization/57371
19033 * match.pd: New pattern.
19035 2017-08-04 Marek Polacek <polacek@redhat.com>
19037 PR middle-end/81695
19038 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
19039 perform the computation in offset_int.
19041 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19043 PR tree-optimization/81136
19044 * tree-vectorizer.h: Include tree-hash-traits.h.
19045 (vec_base_alignments): New typedef.
19046 (vec_info): Add a base_alignments field.
19047 (vect_record_base_alignments): Declare.
19048 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
19050 (DR_IS_CONDITIONAL_IN_STMT): New macro.
19051 (create_data_ref): Add an is_conditional_in_stmt argument.
19052 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
19053 the is_conditional_in_stmt field.
19054 (data_ref_loc): Add an is_conditional_in_stmt field.
19055 (get_references_in_stmt): Set the is_conditional_in_stmt field.
19056 (find_data_references_in_stmt): Update call to create_data_ref.
19057 (graphite_find_data_references_in_stmt): Likewise.
19058 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
19059 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
19060 (vect_record_base_alignment): New function.
19061 (vect_record_base_alignments): Likewise.
19062 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
19063 for nested statements even if we fail to compute a misalignment.
19064 Use pooled base alignments for unconditional references.
19065 (vect_find_same_alignment_drs): Compare base addresses instead
19067 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
19068 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
19070 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19072 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
19073 Add an explicit name for the enum. Use auto_vec for slp_instances
19074 and grouped_stores.
19075 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
19077 (_bb_vec_info): Add a constructor and destructor.
19078 (vinfo_for_stmt): Return NULL for uids of -1 as well.
19079 (destroy_loop_vec_info): Delete.
19080 (vect_destroy_datarefs): Likewise.
19081 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
19082 (vec_info::vec_info): New function.
19083 (vec_info::~vec_info): Likewise.
19084 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
19085 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
19086 destroy_loop_vec_info.
19087 * tree-vect-loop.c (new_loop_vec_info): Replace with...
19088 (_loop_vec_info::_loop_vec_info): ...this.
19089 (destroy_loop_vec_info): Replace with...
19090 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
19091 the stmt_vec_infos. Leave handling of vec_info information to its
19092 destructor. Remove explicit vector releases.
19093 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
19094 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
19095 * tree-vect-slp.c (new_bb_vec_info): Replace with...
19096 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
19097 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
19098 (destroy_bb_vec_info): Replace with...
19099 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
19100 information to its destructor.
19101 (vect_slp_analyze_bb_1): Use new and delete instead of
19102 new_bb_vec_info and destroy_bb_vec_info.
19103 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
19106 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19108 * tree-data-ref.h (subscript): Add access_fn field.
19109 (data_dependence_relation): Add could_be_independent_p.
19110 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
19111 (same_access_functions): Move to tree-data-ref.c.
19112 * tree-data-ref.c (ref_contains_union_access_p): New function.
19113 (access_fn_component_p): Likewise.
19114 (access_fn_components_comparable_p): Likewise.
19115 (dr_analyze_indices): Add a reference to access_fn_component_p.
19116 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
19118 (constant_access_functions): Likewise.
19119 (add_other_self_distances): Likewise.
19120 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
19121 (initialize_data_dependence_relation): Use XCNEW and remove
19122 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
19123 of access functions that have the same type. Allow the
19124 subsequence to end with different bases in some circumstances.
19125 Record the chosen access functions in SUB_ACCESS_FN.
19126 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
19127 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
19128 (subscript_dependence_tester_1): Likewise dra and drb.
19129 (build_classic_dist_vector): Update calls accordingly.
19130 (subscript_dependence_tester): Likewise.
19131 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
19132 DDR_COULD_BE_INDEPENDENT_P.
19133 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
19134 comp_alias_ddrs instead of may_alias_ddrs.
19135 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
19137 (vect_analyze_data_ref_dependence): Use it if
19138 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
19139 distance vectors if that fails.
19140 (dependence_distance_ge_vf): New function.
19141 (vect_prune_runtime_alias_test_list): Use it. Don't clear
19142 LOOP_VINFO_MAY_ALIAS_DDRS.
19144 2017-08-04 Richard Biener <rguenther@suse.de>
19146 PR middle-end/81705
19147 * fold-const.c (fold_binary_loc): Properly restrict
19148 minus_var0 && minus_var1 case when associating undefined overflow
19151 2017-08-04 Tom de Vries <tom@codesourcery.com>
19153 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
19155 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19157 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19158 Don't start diagnostic messages with a capital letter.
19159 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19161 (rs6000_invalid_builtin): Likewise.
19162 (rs6000_trampoline_init): Likewise.
19164 2017-08-03 Jakub Jelinek <jakub@redhat.com>
19167 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
19168 after setting changeable df flags.
19170 2017-08-03 Richard Biener <rguenther@suse.de>
19172 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
19173 up if the use is in USE - X.
19175 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
19177 * toplev.c (dumpfile.h): New include.
19178 (internal_error_reentered): New static function. Use it...
19179 (internal_error_function): ...here to handle reentered internal_error.
19181 2017-08-03 Richard Biener <rguenther@suse.de>
19183 PR middle-end/81148
19184 * fold-const.c (split_tree): Add minus_var and minus_con
19185 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
19186 here but always use minus_*.
19187 (associate_trees): Assert we never associate with MINUS_EXPR
19188 and NULL first operand. Do not recurse for PLUS_EXPR operands
19189 when associating as MINUS_EXPR either.
19190 (fold_binary_loc): Track minus_var and minus_con.
19192 2017-08-03 Tom de Vries <tom@codesourcery.com>
19195 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
19196 ACCEL_COMPILER, apply finish_options on
19197 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
19199 2017-08-03 Tom de Vries <tom@codesourcery.com>
19202 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
19203 function_entry_patch_area_size > 0.
19205 2017-08-03 Jakub Jelinek <jakub@redhat.com>
19208 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
19209 instead of 10??LU, perform unit multiplication in wide_int,
19210 don't change alloc_object_size_limit if the limit is larger
19213 PR tree-optimization/81655
19214 PR tree-optimization/81588
19215 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
19216 the case when ranges[i].low and high are 1 for unsigned type with
19219 PR middle-end/81052
19220 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
19221 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
19223 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19225 * tree-vrp.h: Add include guard.
19227 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
19230 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
19231 (ud2): New insn pattern.
19232 * config/i386/i386.c (ix86_expand_epilogue):
19233 For naked functions, generate ud2 instead of trap insn.
19235 2017-08-02 Marek Polacek <polacek@redhat.com>
19238 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
19240 2017-08-02 Tom de Vries <tom@codesourcery.com>
19241 Cesar Philippidis <cesar@codesourcery.com>
19243 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
19244 Add missing edge probabilities.
19246 2017-08-02 Tamar Christina <tamar.christina@arm.com>
19248 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
19249 Correct endianness.
19251 2017-08-02 Jakub Jelinek <jakub@redhat.com>
19253 PR middle-end/79499
19254 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
19255 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
19256 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
19257 prologue_seq sequences - if any.
19259 2017-08-02 Richard Biener <rguenther@suse.de>
19261 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
19262 via vectors if supported, integer extracts via punning if supported
19263 or otherwise vector extracts.
19265 2017-08-02 Richard Biener <rguenther@suse.de>
19267 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
19269 (bitmap_insert_into_set): ... this.
19271 2017-08-02 Richard Biener <rguenther@suse.de>
19273 PR tree-optimization/81633
19275 2015-08-17 Alan Hayward <alan.hayward@arm.com>
19277 PR tree-optimization/71752
19278 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
19280 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
19282 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
19283 (machine_function::call_ms2sysv_pad_out): Remove field.
19284 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
19285 (ix86_compute_frame_layout): Likewise.
19287 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
19290 * config/i386/i386.c (ix86_set_func_type): Disallow naked
19291 attribute with interrupt attribute.
19293 2017-08-01 Andrew Pinski <apinski@cavium.com>
19295 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
19296 BIT_INSERT_EXPR's operand 1
19297 to see if the types precision matches.
19299 2017-08-01 Martin Liska <mliska@suse.cz>
19301 PR middle-end/70140
19302 * builtins.c (expand_builtin_memcpy_args): Remove.
19303 (expand_builtin_memcpy): Call newly added function
19304 expand_builtin_memory_copy_args.
19305 (expand_builtin_memcpy_with_bounds): Likewise.
19306 (expand_builtin_mempcpy): Remove last argument.
19307 (expand_builtin_mempcpy_with_bounds): Likewise.
19308 (expand_builtin_memory_copy_args): New function created from
19309 expand_builtin_mempcpy_args with small modifications.
19310 (expand_builtin_mempcpy_args): Remove.
19311 (expand_builtin_stpcpy): Remove unused argument.
19312 (expand_builtin): Likewise.
19313 (expand_builtin_with_bounds): Likewise.
19315 2017-08-01 Martin Liska <mliska@suse.cz>
19318 Make mempcpy more optimal (PR middle-end/70140).
19320 2017-08-01 Jakub Jelinek <jakub@redhat.com>
19323 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
19324 __builtin_vec_cmpne verify both arguments are compatible vectors
19325 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
19326 verify arg1_type is a pointer or array type. For __builtin_vec_st,
19327 move computation of aligned to after checking the argument types.
19331 * config/rs6000/vsx.md (vextract_fp_from_shorth,
19332 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
19335 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
19336 Doug Rupp <rupp@adacore.com>
19337 Olivier Hainque <hainque@adacore.com>
19339 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
19340 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
19342 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
19343 for TARGET_OS_CPP_BUILTIN.
19344 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
19345 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
19347 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
19348 passing required abi options to the assembler for EABI configurations.
19349 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
19350 of .text.hot and .text.unlikely sections for kernel modules when
19351 using ARM style exceptions.
19352 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
19353 options. Add EXTRA_CC1_SPEC.
19354 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
19356 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
19358 (ARM_TARGET2_DWARF_FORMAT): Define.
19359 * config/arm/t-vxworks: Adjust multilib control to removal of the
19360 Diab command line options.
19362 2017-08-01 Martin Liska <mliska@suse.cz>
19364 PR gcov-profile/81561
19365 * gcov.c (unblock): Make unblocking safe as we need to preserve
19366 index correspondence of blocks and block_lists.
19368 2017-08-01 Richard Biener <rguenther@suse.de>
19370 PR tree-optimization/81181
19371 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
19372 (compute_antic): ... end of iteration here.
19374 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
19376 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
19377 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
19378 (ftree-slp-vectorize): Likewise.
19379 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
19380 can no longer be set independent of flag_tree_loop_vectorize.
19381 * omp-general.c (emp_max_vf): Likewise.
19382 * opts.c (enable_fdo_optimizations): Remove references to
19383 flag_tree_vectorize, these are now implicit.
19384 (common_handle_option): Remove handling for OPT_ftree_vectorize,
19385 and leave it for the options machinery.
19387 2017-08-01 Martin Liska <mliska@suse.cz>
19389 PR middle-end/70140
19390 * builtins.c (expand_builtin_memcpy_args): Remove.
19391 (expand_builtin_memcpy): Call newly added function
19392 expand_builtin_memory_copy_args.
19393 (expand_builtin_memcpy_with_bounds): Likewise.
19394 (expand_builtin_mempcpy): Remove last argument.
19395 (expand_builtin_mempcpy_with_bounds): Likewise.
19396 (expand_builtin_memory_copy_args): New function created from
19397 expand_builtin_mempcpy_args with small modifications.
19398 (expand_builtin_mempcpy_args): Remove.
19399 (expand_builtin_stpcpy): Remove unused argument.
19400 (expand_builtin): Likewise.
19401 (expand_builtin_with_bounds): Likewise.
19403 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
19406 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
19407 print "ds:" only for immediates in generic address space.
19409 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
19412 * config/i386/i386.c (ix86_funciton_naked): New prototype.
19413 (ix86_function_ok_for_sibcall): Return false for naked functions.
19415 2017-08-01 Richard Biener <rguenther@suse.de>
19417 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
19418 (compute_antic): Seed worklist with exit block predecessors.
19419 * cfganal.c (dfs_find_deadend): For a cycle return the source
19420 of the edge closing it.
19422 2017-08-01 Tamar Christina <tamar.christina@arm.com>
19424 * config/aarch64/aarch64.c
19425 (aarch64_can_const_movi_rtx_p): Move 0 check.
19427 2017-08-01 Bin Cheng <bin.cheng@arm.com>
19429 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
19430 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
19432 * match.pd: Ditto in address comparison pattern.
19434 2017-08-01 Bin Cheng <bin.cheng@arm.com>
19436 PR tree-optimization/81627
19437 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
19438 closed ssa form for store-store chain.
19440 2017-08-01 Bin Cheng <bin.cheng@arm.com>
19442 PR tree-optimization/81620
19443 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
19444 for store-store chain.
19446 2017-08-01 Jakub Jelinek <jakub@redhat.com>
19448 PR tree-optimization/81588
19449 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
19450 ranges[i].in_p, invert comparison code ccode. For >/>=,
19451 swap rhs1 and rhs2 and comparison code unconditionally,
19452 for </<= don't do that. Don't swap rhs1/rhs2 again if
19453 ranges[i].in_p, instead invert comparison code ccode if
19454 opcode or oe->rank is BIT_IOR_EXPR.
19457 * optabs.def (vec_extract_optab, vec_init_optab): Change from
19458 a direct optab to conversion optab.
19459 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
19460 with GET_MODE_INNER as last argument instead of optab_handler.
19461 * expmed.c (extract_bit_field_1): Likewise. Use vector from
19462 vector extraction if possible and optab is available.
19463 * expr.c (store_constructor): Use convert_optab_handler instead
19464 of optab_handler. Use vector initialization from smaller
19465 vectors if possible and optab is available.
19466 * tree-vect-stmts.c (vectorizable_load): Likewise.
19467 * doc/md.texi (vec_extract, vec_init): Document that the optabs
19468 now have two modes.
19469 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
19470 of vec_init from half-sized vectors with the same element mode.
19471 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
19472 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
19473 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
19474 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
19475 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
19476 after mode in gen_vec_extract* calls.
19477 (vec_extract<mode>): Renamed to ...
19478 (vec_extract<mode><ssescalarmodelower>): ... this.
19479 (vec_extract<mode><ssehalfvecmodelower>): New expander.
19480 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
19481 element mode after mode in gen_vec_init* calls.
19482 (VEC_INIT_HALF_MODE): New mode iterator.
19483 (vec_init<mode>): Renamed to ...
19484 (vec_init<mode><ssescalarmodelower>): ... this.
19485 (vec_init<mode><ssehalfvecmodelower>): New expander.
19486 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
19487 (vec_extractv2sfsf): ... this.
19488 (vec_initv2sf): Renamed to ...
19489 (vec_initv2sfsf): ... this.
19490 (vec_extractv2si): Renamed to ...
19491 (vec_extractv2sisi): ... this.
19492 (vec_initv2si): Renamed to ...
19493 (vec_initv2sisi): ... this.
19494 (vec_extractv4hi): Renamed to ...
19495 (vec_extractv4hihi): ... this.
19496 (vec_initv4hi): Renamed to ...
19497 (vec_initv4hihi): ... this.
19498 (vec_extractv8qi): Renamed to ...
19499 (vec_extractv8qiqi): ... this.
19500 (vec_initv8qi): Renamed to ...
19501 (vec_initv8qiqi): ... this.
19502 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
19503 (vec_init<mode>): Renamed to ...
19504 (vec_init<mode><VEC_base_l>): ... this.
19505 (vec_extract<mode>): Renamed to ...
19506 (vec_extract<mode><VEC_base_l>): ... this.
19507 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
19508 (vec_initv2sfsf): ... this.
19509 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
19510 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
19511 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
19512 element mode after mode in gen_vec_init* calls.
19513 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
19514 (vec_init<mode><Vel>): ... this.
19515 (vec_extract<mode>): Renamed to ...
19516 (vec_extract<mode><Vel>): ... this.
19517 * config/aarch64/iterators.md (Vel): New mode attribute.
19518 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
19519 Add element mode after mode in gen_vec_extract* calls.
19520 * config/s390/vector.md (non_vec_l): New mode attribute.
19521 (vec_extract<mode>): Renamed to ...
19522 (vec_extract<mode><non_vec_l>): ... this.
19523 (vec_init<mode>): Renamed to ...
19524 (vec_init<mode><non_vec_l>): ... this.
19525 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
19526 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
19528 * config/arm/iterators.md (V_elem_l): New mode attribute.
19529 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
19530 (vec_extract<mode><V_elem_l>): ... this.
19531 (vec_extractv2di): Renamed to ...
19532 (vec_extractv2didi): ... this.
19533 (vec_init<mode>): Renamed to ...
19534 (vec_init<mode><V_elem_l>): ... this.
19535 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
19536 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
19537 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
19538 Add element mode after gen_vec_extract* calls.
19539 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
19540 (vec_init<mode><unitmode>): ... this.
19541 (vec_extract<mode>): Renamed to ...
19542 (vec_extract<mode><unitmode>): ... this.
19543 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
19544 (vec_init<mode><unitmode>): ... this.
19545 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
19546 (vec_initv2sfsf): ... this.
19547 (vec_extractv2sf): Renamed to ...
19548 (vec_extractv2sfsf): ... this.
19549 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
19550 Add element mode after gen_vec_extract* calls.
19551 * config/mips/mips.md (unitmode): New mode iterator.
19552 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
19553 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
19554 * config/spu/spu.md (inner_l): New mode attribute.
19555 (vec_init<mode>): Renamed to ...
19556 (vec_init<mode><inner_l>): ... this.
19557 (vec_extract<mode>): Renamed to ...
19558 (vec_extract<mode><inner_l>): ... this.
19559 * config/sparc/sparc.md (veltmode): New mode iterator.
19560 (vec_init<VMALL:mode>): Renamed to ...
19561 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
19562 * config/ia64/vect.md (vec_initv2si): Renamed to ...
19563 (vec_initv2sisi): ... this.
19564 (vec_initv2sf): Renamed to ...
19565 (vec_initv2sfsf): ... this.
19566 (vec_extractv2sf): Renamed to ...
19567 (vec_extractv2sfsf): ... this.
19568 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
19569 (vec_init<mode>): Renamed to ...
19570 (vec_init<mode><VEC_base_l>): ... this.
19571 (vec_extract<mode>): Renamed to ...
19572 (vec_extract<mode><VEC_base_l>): ... this.
19573 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
19574 (vec_initv2sfsf): ... this.
19575 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
19576 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
19577 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
19578 gen_vec_init* calls.
19580 2017-08-01 Richard Biener <rguenther@suse.de>
19582 PR tree-optimization/81297
19583 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
19584 TREE_OVERFLOW from INTEGER_CSTs.
19586 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
19588 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
19590 2017-07-31 Carl Love <cel@us.ibm.com>
19592 * config/rs6000/rs6000-c: Add support for built-in functions
19593 vector signed char vec_xl_be (signed long long, signed char *);
19594 vector unsigned char vec_xl_be (signed long long, unsigned char *);
19595 vector signed int vec_xl_be (signed long long, signed int *);
19596 vector unsigned int vec_xl_be (signed long long, unsigned int *);
19597 vector signed long long vec_xl_be (signed long long, signed long long *);
19598 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
19599 vector signed short vec_xl_be (signed long long, signed short *);
19600 vector unsigned short vec_xl_be (signed long long, unsigned short *);
19601 vector double vec_xl_be (signed long long, double *);
19602 vector float vec_xl_be (signed long long, float *);
19603 * config/rs6000/altivec.h (vec_xl_be): Add #define.
19604 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
19605 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
19607 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
19608 (altivec_expand_builtin): Add switch statement to call
19609 altivec_expand_xl_be for each builtin.
19610 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
19611 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
19612 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
19613 __builtin_vsx_le_be_v16qi.
19614 * doc/extend.texi: Update the built-in documentation file for the
19615 new built-in functions.
19617 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19620 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
19622 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
19624 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19626 * config.gcc: Add z14.
19627 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
19628 CPU model numbers for z13s and z14.
19629 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
19631 * config/s390/s390-opts.h (enum processor_type): Rename
19632 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19633 * config/s390/s390.c (processor_table): Add field for CPU name to
19634 be passed to Binutils.
19635 (s390_asm_output_machine_for_arch): Use the new field in
19636 processor_table for Binutils.
19637 (s390_expand_builtin): Replace arch12 with z14.
19638 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19639 (s390_get_sched_attrmask): Likewise.
19640 (s390_get_unit_mask): Likewise.
19641 * config/s390/s390.opt: Add z14 to processor_type enum.
19643 2017-07-31 Martin Jambor <mjambor@suse.cz>
19646 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
19647 regardless of optimization level.
19649 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
19650 Martin Liska <mliska@suse.cz>
19652 * predict.def: Remove old comment and adjust probability.
19653 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
19654 PREDICT statements.
19656 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19659 * config/i386/i386.c (ix86_function_naked): New function.
19660 (ix86_can_use_return_insn_p): Return false for naked functions.
19661 (ix86_expand_prologue): Skip prologue for naked functions.
19662 (ix86_expand_epilogue): Skip epilogue for naked functions
19663 and emit trap instruction.
19664 (ix86_warn_func_return): New function.
19665 (ix86_attribute_table): Add "naked" attribute specification.
19666 (TARGET_WARN_FUNC_RETURN): Define.
19667 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
19669 2017-07-31 Martin Liska <mliska@suse.cz>
19671 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
19672 (dump_gimple_bb_header): Always dump BB info.
19673 (pp_cfg_jump): Do not append info about BB when dumping a jump.
19675 2017-07-31 Martin Liska <mliska@suse.cz>
19678 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
19679 also with current_function_decl non-null equality.
19681 2017-07-31 Jakub Jelinek <jakub@redhat.com>
19684 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
19685 change type to the element type, instead add eltype variable and
19686 use it where we are interested in the element type.
19688 PR tree-optimization/81603
19689 * ipa-polymorphic-call.c
19690 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
19691 offset arithmetic in offset_int, bail out if the resulting bit offset
19692 doesn't fit into shwi.
19694 2017-07-31 Martin Liska <mliska@suse.cz>
19696 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
19697 (gimplify_save_expr): Fix comment.
19699 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
19702 * config/i386/i386.c (ix86_function_arg): Update arguments for
19704 (ix86_compute_frame_layout): Set the initial stack offset to
19705 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
19706 INCOMING_FRAME_SP_OFFSET.
19707 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
19708 stack before exception handler returns.
19709 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
19710 the 'ERROR_CODE' for exception handler.
19712 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
19714 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
19715 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
19716 (ASM_OUTPUT_REG_POP): Ditto.
19717 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
19718 instead of asm_fprintf to output pure string.
19720 2017-07-29 Jakub Jelinek <jakub@redhat.com>
19722 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
19723 to imported_module_or_decl hook.
19724 (debug_nothing_tree_tree_tree_bool): Remove.
19725 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
19726 * debug.c (do_nothing_debug_hooks): Use
19727 debug_nothing_tree_tree_tree_bool_bool instead of
19728 debug_nothing_tree_tree_tree_bool.
19729 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19730 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
19731 * sdbout.c (sdb_debug_hooks): Likewise.
19732 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
19733 (gen_namespace_die): Add DW_AT_export_symbols attribute if
19735 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
19736 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
19737 attribute, don't add anything.
19739 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19741 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
19742 (fold_build2_stat_loc): Likewise.
19743 (fold_build3_stat_loc): Likewise.
19744 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
19745 (fold_build1_loc): Remove macro.
19746 (fold_build2_loc): Likewise.
19747 (fold_build3_loc): Likewise.
19749 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19751 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
19752 (gimple_build_debug_bind_source_stat): Likewise.
19753 * gimple.h (gimple_build_debug_bind): Remove macro.
19754 (gimple_build_debug_bind_source): Likewise.
19756 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19758 * bitmap.c (bitmap_alloc): Adjust.
19759 (bitmap_gc_alloc): Likewise.
19760 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
19762 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19764 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
19765 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
19766 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
19767 (bitmap_gc_alloc_stat): Likewise.
19768 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
19770 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19772 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
19773 * rtl.h (shallow_copy_rtx): Remove macro.
19775 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19777 * emit-rtl.c (gen_raw_REG): Adjust.
19778 * gengenrtl.c (gendef): Likewise.
19779 * rtl.c (rtx_alloc_stat): Remove _stat from name.
19780 * rtl.h (rtx_alloc): Remove macro.
19782 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19784 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
19785 (build_tree_list_stat): Likewise.
19786 * tree.h (build_tree_list): Remove macro.
19787 (build_tree_list_vec): Likewise.
19789 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19791 * tree.c (make_vector_stat): Remove _stat from name.
19792 (build_vector_stat): Likewise.
19793 * tree.h (make_vector_stat): Remove macro.
19794 (build_vector_stat): Likewise.
19796 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19798 * tree.h (build_var_debug_value): Remove prototype.
19800 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19802 * tree.c (tree_cons_stat): Remove _stat from name.
19803 * tree.h (tree_cons): Remove macro.
19805 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19807 * tree.c (build_vl_exp_stat): Remove _stat from name.
19808 * tree.h (build_vl_exp): Remove macro.
19810 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19812 * tree.c (build_decl_stat): Remove _stat from name.
19813 * tree.h (build_decl): Remove macro.
19815 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19817 * gimple.c (gimple_build_with_ops_stat): Adjust.
19818 (gimple_alloc_stat): Remove _stat from name.
19819 * gimple.h (gimple_alloc): Remove macro.
19821 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19823 * tree.c (make_tree_vec_stat): Remove _stat from name.
19824 (grow_tree_vec_stat): Likewise.
19825 * tree.h (make_tree_vec_stat): Adjust prototype.
19826 (grow_tree_vec_stat): Likewise.
19827 (make_tree_vec): Remove macro.
19828 (grow_tree_vec): Likewise.
19830 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19832 * fold-const.c (fold_build1_stat_loc): Adjust.
19833 (fold_build2_stat_loc): Likewise.
19834 (fold_build3_stat_loc): Likewise.
19835 * tree.c (build0_stat): Remove _stat from name.
19836 (build1_stat): Likewise.
19837 (build2_stat): Likewise.
19838 (build3_stat): Likewise.
19839 (build4_stat): Likewise.
19840 (build5_stat): Likewise.
19841 * tree.h (build1_loc): Remove macro, and rename _stat function
19843 (build2_loc): Likewise.
19844 (build3_loc): Likewise.
19845 (build4_loc): Likewise.
19846 (build5_loc): Likewise.
19848 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19850 * tree.c (make_int_cst_stat): Remove _stat from name.
19851 * tree.h (make_int_cst_stat): Adjust prototype.
19852 (make_int_cst): Remove macro.
19854 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19856 * tree.c (make_tre_binfo_stat): Remove _stat from name.
19857 * tree.h (make_tree_binfo_stat): Adjust prototype.
19858 (make_tree_binfo): Remove.
19860 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19862 * tree.c (copy_node_stat): Rename to copy_node.
19863 (build_distinct_type_copy): Adjust.
19864 * tree.h (copy_node_stat): Adjust prototype.
19865 (copy_node): Remove macro.
19867 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19869 * tree.c (make_node_stat): rename to make_node.
19870 (build_tree_list_stat): Adjust.
19871 (build0_stat): Likewise.
19872 (build2_stat): Likewise.
19873 (build3_stat): Likewise.
19874 (build4_stat): Likewise.
19875 (build5_stat): Likewise.
19876 (build_decl_stat): Likewise.
19877 * tree.h (make_node_stat): Adjust prototype.
19878 (make_node): remove macro.
19880 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
19882 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
19883 (PPC_FEATURE2_SCV): Likewise.
19884 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
19886 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19888 * config/aarch64/aarch64.c
19889 (aarch64_internal_mov_immediate): Add new special pattern.
19890 * config/aarch64/aarch64.md (*movdi_aarch64):
19891 Add reg/32bit const mov case.
19893 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19894 Richard Sandiford <richard.sandiford@linaro.org>
19896 * config/aarch64/aarch64.md (mov<mode>): Generalize.
19897 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
19898 Add integer and movi cases.
19899 (movi-split-hf-df-sf split, fp16): New.
19900 (enabled): Added TARGET_FP_F16INST.
19901 * config/aarch64/iterators.md (GPF_HF): New.
19902 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
19904 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19906 * config/aarch64/aarch64.c
19907 (aarch64_simd_container_mode): Add prototype.
19908 (aarch64_expand_mov_immediate): Add HI support.
19909 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
19910 (aarch64_can_const_movi_rtx_p): New.
19911 (aarch64_preferred_reload_class):
19912 Remove restrictions of using FP registers for certain SIMD operations.
19913 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
19914 (aarch64_valid_floating_const): Add integer move validation.
19915 (aarch64_simd_imm_scalar_p): Remove.
19916 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
19917 (aarch64_legitimate_constant_p): Expand list of supported cases.
19918 * config/aarch64/aarch64-protos.h
19919 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
19920 (aarch64_reinterpret_float_as_int): New.
19921 (aarch64_simd_imm_scalar_p): Remove.
19922 * config/aarch64/constraints.md (Uvi): New.
19923 (Dd): Split into Ds and new Dd.
19924 * config/aarch64/aarch64.md (*movsi_aarch64):
19926 (*movdi_aarch64): Add SIMD mov case.
19928 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19930 * tree-predcom.c: (struct chain): Handle store-store chain in which
19931 stores for elimination only store loop invariant values.
19932 (execute_pred_commoning_chain): Ditto.
19933 (prepare_initializers_chain_store_elim): Ditto.
19934 (prepare_finalizers): Ditto.
19935 (is_inv_store_elimination_chain): New function.
19936 (initialize_root_vars_store_elim_1): New function.
19938 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19940 * tree-predcom.c: Revise general description of the pass.
19941 (enum chain_type): New enum type for store elimination.
19942 (struct chain): New field supporting store elimination.
19943 (struct component): Ditto.
19944 (dump_chain): Dump store-stores chain.
19945 (release_chain): Release resources.
19946 (split_data_refs_to_components): Compute and create component
19947 contains only stores for elimination.
19948 (get_chain_last_ref_at): New function.
19949 (make_invariant_chain): Initialization.
19950 (make_rooted_chain): Specify chain type in parameter and record it.
19951 (add_looparound_copies): Skip for store-stores chain.
19952 (determine_roots_comp): Compute type of chain and pass it to
19954 (initialize_root_vars_store_elim_2): New function.
19955 (finalize_eliminated_stores): New function.
19956 (remove_stmt): Handle store for elimination.
19957 (execute_pred_commoning_chain): Execute predictive commoning on
19958 store-store chains.
19959 (determine_unroll_factor): Skip unroll for store-stores chain.
19960 (prepare_initializers_chain_store_elim): New function.
19961 (prepare_initializers_chain): Hanlde store-store chain.
19962 (prepare_finalizers_chain, prepare_finalizers): New function.
19963 (tree_predictive_commoning_loop): Return integer value indicating
19964 if loop is unrolled or lcssa form is corrupted.
19965 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
19967 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19969 * tree-predcom.c (initialize_root): Delete.
19970 (execute_pred_commoning_chain): Initialize root vars and replace
19971 reference of non-combined chain directly, rather than call above
19974 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19976 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
19977 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
19979 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19981 * tree-predcom.c (struct chain): New field init_seq.
19982 (release_chain): Release init_seq.
19983 (prepare_initializers_chain): Record intialization stmts in above
19985 (insert_init_seqs): New function.
19986 (tree_predictive_commoning_loop): Call insert_init_seqs.
19988 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19990 * tree-predcom.c (determine_roots_comp): Skip trivial components.
19992 2017-07-28 Richard Biener <rguenther@suse.de>
19994 * match.pd: Remove superfluous :c.
19995 * genmatch.c (simplify::id): Add member.
19996 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
19998 (current_id): New global.
19999 (dt_node::parent): Move from ...
20000 (dt_operand::parent): ... here. Add for_id member.
20001 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
20002 (decision_tree::find_node): Relax order requirement when
20003 merging DT_TRUE nodes to ones inbetween the current simplify
20004 and the one we try to merge with. Add diagnostic whenever
20005 we need to enforce pattern order by not merging.
20006 (decision_tree::insert): Set current_id.
20007 (decision_tree::print_node): Dump parent node and for_id.
20008 (parser::last_id): Add member.
20009 (parser::push_simplify): Assign unique id.
20010 (parser::parser): Initialize last_id.
20012 2017-07-28 Martin Liska <mliska@suse.cz>
20015 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
20016 gimple_build_debug_bind.
20018 2017-07-28 Richard Biener <rguenther@suse.de>
20020 PR tree-optimization/81502
20021 * match.pd: Add pattern combining BIT_INSERT_EXPR with
20023 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
20025 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
20026 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
20027 for BIT_FIELD_REF args.
20028 * fold-const.c (make_bit_field_ref): Likewise.
20029 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
20031 2017-07-28 Jakub Jelinek <jakub@redhat.com>
20034 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
20035 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
20036 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
20037 Or it into SANITIZER_UNDEFINED.
20038 * ubsan.c: Include gimple-fold.h and varasm.h.
20039 (ubsan_expand_ptr_ifn): New function.
20040 (instrument_pointer_overflow): New function.
20041 (maybe_instrument_pointer_overflow): New function.
20042 (instrument_object_size): Formatting fix.
20043 (pass_ubsan::execute): Call instrument_pointer_overflow
20044 and maybe_instrument_pointer_overflow.
20045 * internal-fn.c (expand_UBSAN_PTR): New function.
20046 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
20047 * sanitizer.def (__ubsan_handle_pointer_overflow,
20048 __ubsan_handle_pointer_overflow_abort): New builtins.
20049 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
20050 * internal-fn.def (UBSAN_PTR): New internal function.
20051 * opts.c (sanitizer_opts): Add pointer-overflow.
20052 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
20053 * fold-const.c (build_range_check): Compute pointer range check in
20054 integral type if pointer arithmetics would be needed. Formatting
20057 2017-07-28 Martin Liska <mliska@suse.cz>
20060 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
20061 parameters that are of a variable-length.
20063 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
20065 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
20067 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
20068 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
20069 (CRT_CALL_STATIC_FUNCTION): Likewise.
20070 (ASM_DEFAULT_SPEC): New define.
20071 (ASM_SPEC32): Likewise.
20072 (ASM_SPEC64): Likewise.
20073 (ASM_SPEC_COMMON): Likewise.
20074 (ASM_SPEC): Likewise.
20075 (INVALID_64BIT): Likewise.
20076 (LINK_OS_DEFAULT_SPEC): Likewise.
20077 (LINK_OS_SPEC32): Likewise.
20078 (LINK_OS_SPEC64): Likewise.
20079 (POWERPC_LINUX): Likewise.
20080 (PTRDIFF_TYPE): Likewise.
20081 (RESTORE_FP_PREFIX): Likewise.
20082 (RESTORE_FP_SUFFIX): Likewise.
20083 (SAVE_FP_PREFIX): Likewise.
20084 (SAVE_FP_SUFFIX): Likewise.
20085 (SIZE_TYPE): Likewise.
20086 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
20087 (TARGET_64BIT): Likewise.
20088 (TARGET_64BIT): Likewise.
20089 (TARGET_AIX): Likewise.
20090 (WCHAR_TYPE_SIZE): Likewise.
20091 (WCHAR_TYPE): Undefine.
20092 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
20093 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
20094 (CPP_OS_RTEMS_SPEC): Delete.
20095 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
20096 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
20098 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
20100 2017-07-28 Jakub Jelinek <jakub@redhat.com>
20102 PR tree-optimization/81578
20103 * tree-parloops.c (build_new_reduction): Bail out if
20104 reduction_code isn't one of the standard OpenMP reductions.
20105 Move the details printing after that decision.
20107 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20109 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
20110 related to reload_in_progress.
20111 (splat_input_operand): Likewise.
20112 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
20114 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
20116 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
20117 (TARGET_INSTANTIATE_DECLS): Likewise.
20118 (legitimate_indexed_address_p): Delete reload_in_progress code.
20119 (rs6000_debug_legitimate_address_p): Likewise.
20120 (rs6000_eliminate_indexed_memrefs): Likewise.
20121 (rs6000_emit_le_vsx_store): Likewise.
20122 (rs6000_emit_move_si_sf_subreg): Likewise.
20123 (rs6000_emit_move): Likewise.
20124 (register_to_reg_type): Likewise.
20125 (rs6000_pre_atomic_barrier): Likewise.
20126 (rs6000_machopic_legitimize_pic_address): Likewise.
20127 (rs6000_allocate_stack_temp): Likewise.
20128 (rs6000_address_for_fpconvert): Likewise.
20129 (rs6000_address_for_altivec): Likewise.
20130 (rs6000_secondary_memory_needed_rtx): Delete function.
20131 (rs6000_check_sdmode): Likewise.
20132 (rs6000_alloc_sdmode_stack_slot): Likewise.
20133 (rs6000_instantiate_decls): Likewise.
20134 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
20135 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
20136 Delete reload_in_progress.
20137 (*vec_reload_and_plus_<mptrsize>): Likewise.
20138 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
20139 (vsx_div_v2di): Likewise.
20140 (vsx_udiv_v2di): Likewise.
20142 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20144 * config/rs6000/rs6000.opt (mlra): Replace with stub.
20145 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
20146 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
20147 (rs6000_debug_reg_global): Delete print of LRA status.
20148 (rs6000_option_override_internal): Delete dead LRA related code.
20149 (rs6000_lra_p): Delete function.
20150 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
20152 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
20154 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
20155 * config/riscv/rtems.h: New file.
20157 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20158 Sudakshina Das <sudi.das@arm.com>
20160 * config/aarch64/aarch64.md
20161 (define_split for and<mode>3nr_compare): Move
20162 non aarch64_logical_operand to a register.
20163 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
20164 register immediate operand to a register.
20165 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
20167 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20169 PR middle-end/81564
20170 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
20172 2017-07-27 Richard Biener <rguenther@suse.de>
20174 PR tree-optimization/81573
20175 PR tree-optimization/81494
20176 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
20177 multi defuse cycle case.
20179 2017-07-27 Richard Biener <rguenther@suse.de>
20181 PR tree-optimization/81571
20182 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
20185 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
20187 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
20188 earlier and only if MASK_FPU is set. Adjust formatting.
20190 2017-07-27 Martin Liska <mliska@suse.cz>
20192 * opt-functions.awk: Add validation of value of Init.
20193 * optc-gen.awk: Pass new argument.
20195 2017-07-27 Martin Liska <mliska@suse.cz>
20197 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
20198 Fix wrong condition.
20200 2017-07-27 Martin Liska <mliska@suse.cz>
20202 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
20203 BBs and edges seen by autoFDO.
20205 2017-07-27 Richard Biener <rguenther@suse.de>
20207 PR tree-optimization/81502
20208 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
20209 with incompatible but same sized type.
20210 (execute_update_addresses_taken): Likewise.
20212 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
20214 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
20215 flag_tree_loop_vectorize rather than flag_tree_vectorize.
20217 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20220 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
20221 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
20222 Change s_operand to memory_operand.
20224 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
20226 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
20227 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
20228 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
20229 Emit instructions rather than returning an expression. Handle TFmode
20230 and KFmode by casting to TImode.
20231 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
20232 (rs6000_emit_le_vsx_store): Likewise.
20233 * config/rs6000/vsx.md (VSX_TI): New iterator.
20234 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
20235 (*vsx_le_undo_permute_<mode>): Likewise.
20236 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
20237 emit the split sequence.
20238 (*vsx_le_perm_store_<mode>): Likewise.
20240 2017-07-27 Jakub Jelinek <jakub@redhat.com>
20242 PR tree-optimization/81555
20243 PR tree-optimization/81556
20244 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
20245 if true, force CHANGED for the recursive invocation.
20246 (reassociate_bb): Remember original length of ops array, pass
20247 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
20249 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
20250 attributes for noipa attribute. For naked attribute use
20251 lookup_attribute first before lookup_attribute_spec.
20252 * final.c (rest_of_handle_final): Disable IPA RA for functions with
20254 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
20255 for functions with noipa attribute.
20256 (cgraph_externally_visible_p): Return true for functions with noipa
20258 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
20259 for functions with noipa attribute.
20260 * doc/extend.texi: Document noipa function attribute.
20261 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
20262 also for functions with noipa attribute.
20263 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
20265 2017-07-26 Andrew Pinski <apinski@cavium.com>
20267 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
20268 vec_unalign_load_cost and vec_unalign_store_cost.
20270 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
20272 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
20273 -mvsx-small-integer option.
20274 (ISA_3_0_MASKS_IEEE): Likewise.
20275 (OTHER_VSX_VECTOR_MASKS): Likewise.
20276 (POWERPC_MASKS): Likewise.
20277 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
20278 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
20279 code, only testing for DImode being allowed in non-VSX floating
20281 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
20282 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
20284 (rs6000_option_override_internal): Delete -mvsx-small-integer.
20285 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
20286 TARGET_P8_VECTOR test.
20287 (rs6000_secondary_reload_simple_move): Likewise.
20288 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
20289 since TARGET_P9_VECTOR was already tested.
20290 (rs6000_opt_masks): Remove -mvsx-small-integer.
20291 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
20292 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
20294 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
20295 test for TARGET_VEXTRACTUB was used, and that uses
20297 (p9 extract splitter): Likewise.
20298 (vsx_extract_<mode>_di_p9): Likewise.
20299 (vsx_extract_<mode>_store_p9): Likewise.
20300 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
20301 for TARGET_P9_VECTOR was used. Delete code that is now dead with
20302 the elimination of TARGET_VSX_SMALL_INTEGER.
20303 (vsx_extract_<mode>_p8): Likewise.
20304 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
20305 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
20306 (vsx_set_<mode>_p9): Likewise.
20307 (vsx_set_v4sf_p9): Likewise.
20308 (vsx_set_v4sf_p9_zero): Likewise.
20309 (vsx_insert_extract_v4sf_p9): Likewise.
20310 (vsx_insert_extract_v4sf_p9_2): Likewise.
20311 * config/rs6000/rs6000.md (sign extend splitter): Change
20312 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
20313 (floatsi<mode>2_lfiwax_mem): Likewise.
20314 (floatunssi<mode>2_lfiwzx_mem): Likewise.
20315 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
20316 since a test for TARGET_P9_VECTOR was used.
20317 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
20318 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
20319 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
20320 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
20321 TARGET_P8_VECTOR test.
20322 (fix_trunc<mode>si2_stfiwx): Likewise.
20323 (fix_trunc<mode>si2_internal): Likewise.
20324 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
20325 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
20327 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
20328 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
20329 TARGET_P8_VECTOR test.
20330 (fixuns_trunc<mode>si2_stfiwx): Likewise.
20331 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
20332 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
20334 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
20335 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
20336 since a test for TARGET_P9_VECTOR was used.
20337 (splitter for loading small constants): Likewise.
20339 2017-07-26 Andrew Pinski <apinski@cavium.com>
20341 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
20344 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
20347 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
20348 (fp_valid_at): Likewise.
20350 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
20352 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
20353 (qdf24xx_addrcost_table): Likewise.
20354 (cortexa57_tunings): Update to use generic_branch_cost.
20355 (cortexa72_tunings): Likewise.
20356 (cortexa73_tunings): Likewise.
20357 (qdf24xx_tunings): Likewise.
20359 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
20361 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
20362 (thunderx2t99_branch_cost): Likewise.
20363 (cortexa35_tunings): Update to use generic_branch_cost.
20364 (cortexa53_tunings): Likewise.
20365 (cortexa57_tunings): Likewise.
20366 (cortexa72_tunings): Likewise.
20367 (cortexa73_tunings): Likewise.
20368 (thunderx2t99_tunings): Likewise.
20370 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
20372 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
20373 (sparc_option_override): Honour MASK_FSMULD.
20374 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
20375 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
20376 * config/sparc/sparc.opt (mfsmuld): New option.
20377 * doc/invoke.texi (mfsmuld): Document option.
20379 2017-07-26 Marek Polacek <polacek@redhat.com>
20381 PR middle-end/70992
20382 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
20384 2017-07-26 Richard Biener <rguenther@suse.de>
20386 * gimple-match-head.c (do_valueize): Return OP if valueize
20388 (get_def): New helper to get at the def stmt of a SSA name
20389 if valueize allows.
20390 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
20391 do_valueize to get at the def stmt.
20392 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
20394 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
20396 PR middle-end/46932
20397 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
20399 2017-07-26 Martin Liska <mliska@suse.cz>
20402 * function.c (expand_function_start): Make expansion of
20403 nonlocal_goto_save_area after parm_birth_insn.
20405 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
20407 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
20408 from all CPU target flags enable members.
20410 2017-07-26 Richard Biener <rguenther@suse.de>
20412 * genmatch.c (dt_simplify::gen): Make iterator vars const.
20413 (decision_tree::gen): Make 'type' const.
20414 (write_predicate): Likewise.
20416 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
20418 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
20419 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
20420 (rs6000_option_override_internal): Likewise.
20421 (rs6000_expand_vector_set): Likewise.
20422 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
20423 (TARGET_UPPER_REGS_SF): Likewise.
20424 (TARGET_UPPER_REGS_DI): Likewise.
20425 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
20426 (TARGET_DIRECT_MOVE_64BIT): Likewise.
20427 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
20428 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
20429 (Splitters for DI constants in Altivec registers): Likewise.
20430 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
20431 (vsx_set_v4sf_p9): Likewise.
20432 (vsx_set_v4sf_p9_zero): Likewise.
20433 (vsx_insert_extract_v4sf_p9): Likewise.
20434 (vsx_insert_extract_v4sf_p9_2): Likewise.
20436 2017-07-25 Carl Love <cel@us.ibm.com>
20438 * doc/extend.texi: Update the built-in documentation file for the
20439 existing built-in functions
20440 vector signed char vec_cnttz (vector signed char);
20441 vector unsigned char vec_cnttz (vector unsigned char);
20442 vector signed short vec_cnttz (vector signed short);
20443 vector unsigned short vec_cnttz (vector unsigned short);
20444 vector signed int vec_cnttz (vector signed int);
20445 vector unsigned int vec_cnttz (vector unsigned int);
20446 vector signed long long vec_cnttz (vector signed long long);
20447 vector unsigned long long vec_cnttz (vector unsigned long long);
20449 2017-07-25 Andrew Pinski <apinski@cavium.com>
20451 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
20452 accesses where the use is for the first operand of a BIT_INSERT.
20454 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
20457 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
20458 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
20460 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
20462 * config/i386/gstabs.h: Delete.
20463 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
20465 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
20467 * config/i386/i386.c (ix86_decompose_address): Do not check for
20468 register RTX when looking at index_reg or base_reg.
20469 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
20471 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
20473 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
20474 to update EH info here.
20476 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20478 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
20480 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20482 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
20484 2017-07-25 Torsten Duwe <duwe@suse.de>
20486 * common.opt: Introduce -fpatchable-function-entry
20487 command line option, and its variables function_entry_patch_area_size
20488 and function_entry_patch_area_start.
20489 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
20490 including a two-value parser.
20491 * target.def (print_patchable_function_entry): New target hook.
20492 * targhooks.h (default_print_patchable_function_entry): New function.
20493 * targhooks.c (default_print_patchable_function_entry): Likewise.
20494 * toplev.c (process_options): Switch off IPA-RA if
20495 patchable function entries are being generated.
20496 * varasm.c (assemble_start_function): Look at the
20497 patchable-function-entry command line switch and current
20498 function attributes and maybe generate NOP instructions by
20499 calling the print_patchable_function_entry hook.
20500 * doc/extend.texi: Document patchable_function_entry attribute.
20501 * doc/invoke.texi: Document -fpatchable_function_entry
20502 command line option.
20503 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
20505 * doc/tm.texi: Re-generate.
20507 2017-07-25 Jakub Jelinek <jakub@redhat.com>
20510 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
20511 TARGET_AVX512DQ rather than TARGET_AVX512BW.
20513 2017-07-25 Tamar Christina <tamar.christina@arm.com>
20515 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
20517 2017-07-25 Richard Biener <rguenther@suse.de>
20519 PR tree-optimization/81455
20520 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
20521 not walk in cycles when looking for guards.
20523 2017-07-25 Richard Biener <rguenther@suse.de>
20525 PR tree-optimization/81529
20526 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
20527 when optimizing backedge uses.
20529 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
20531 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
20533 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
20534 to dl_section_ref. On AIX, append an expression to subtract
20535 the size of the section length to dl_section_ref.
20537 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
20539 * configure.ac: If any of the config.* scripts fail, exit 1.
20540 * configure: Regenerate.
20542 2017-07-25 Richard Biener <rguenther@suse.de>
20544 PR middle-end/81546
20545 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
20546 of immediate uses, be more verbose on errors.
20548 2017-07-25 Richard Biener <rguenther@suse.de>
20550 PR tree-optimization/81510
20551 * tree-vect-loop.c (vect_is_simple_reduction): When the
20552 reduction stmt is not inside the loop bail out.
20554 2017-07-25 Richard Biener <rguenther@suse.de>
20556 PR tree-optimization/81303
20557 * tree-vect-loop-manip.c (vect_loop_versioning): Build
20558 profitability check against LOOP_VINFO_NITERSM1.
20560 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20562 * domwalk.c (cmp_bb_postorder): Simplify.
20563 (sort_bbs_postorder): New function. Use it...
20564 (dom_walker::walk): ...here to optimize common cases.
20566 2017-07-25 Martin Liska <mliska@suse.cz>
20569 * ipa-visibility.c (function_and_variable_visibility): Make the
20570 redirection just on target that supports aliasing.
20571 Fix GNU coding style.
20573 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
20576 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
20578 * config/arm/rtems.h: Likewise.
20579 * config/bfin/rtems.h: Likewise.
20580 * config/i386/rtemself.h: Likewise.
20581 * config/lm32/rtems.h: Likewise.
20582 * config/m32c/rtems.h: Likewise.
20583 * config/m68k/rtemself.h: Likewise.
20584 * config/microblaze/rtems.h: Likewise.
20585 * config/mips/rtems.h: Likewise.
20586 * config/moxie/rtems.h: Likewise.
20587 * config/nios2/rtems.h: Likewise.
20588 * config/powerpcspe/rtems.h: Likewise.
20589 * config/rs6000/rtems.h: Likewise.
20590 * config/rtems.h: Likewise.
20591 * config/sh/rtems.h: Likewise.
20592 * config/sh/rtemself.h: Likewise.
20593 * config/sparc/rtemself.h: Likewise.
20595 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20598 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
20599 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
20600 * tree-ssa-structalias.c (alias_get_name): Same.
20602 2017-07-25 Bin Cheng <bin.cheng@arm.com>
20605 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
20606 instructions if no du chain is found.
20608 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20610 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
20612 2017-07-25 Richard Biener <rguenther@suse.de>
20614 PR middle-end/81505
20615 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
20618 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
20620 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
20621 upper-regs options.
20622 (ISA_2_7_MASKS_SERVER): Likewise.
20623 (ISA_3_0_MASKS_IEEE): Likewise.
20624 (OTHER_P8_VECTOR_MASKS): Likewise.
20625 (OTHER_VSX_VECTOR_MASKS): Likewise.
20626 (POWERPC_MASKS): Likewise.
20627 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
20628 duplicate list of options.
20629 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
20630 explicit -mupper-regs options.
20631 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
20632 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
20633 alias for -mupper-regs-df.
20634 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20635 (rs6000_init_hard_regno_mode_ok): Likewise.
20636 (rs6000_option_override_internal): Likewise.
20637 (rs6000_opt_masks): Likewise.
20638 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
20639 options in terms of whether -mvsx or -mpower8-vector was used.
20640 (TARGET_UPPER_REGS_DI): Likewise.
20641 (TARGET_UPPER_REGS_SF): Likewise.
20642 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
20643 -mupper-regs-* options.
20645 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20647 * passes.c (emergency_dump_function): Print some empty lines and a
20648 header before the RTL dump.
20650 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20652 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
20654 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
20657 * config/aarch64/aarch64.c (aarch64_classify_symbol):
20658 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
20660 2017-07-24 Carl Love <cel@us.ibm.com>
20662 * config/rs6000/rs6000-c.c: Add support for built-in functions
20663 vector float vec_extract_fp32_from_shorth (vector unsigned short);
20664 vector float vec_extract_fp32_from_shortl (vector unsigned short);
20665 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
20666 vec_extract_fp_from_shortl): Add defines for the two builtins.
20667 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
20668 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
20670 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
20671 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
20672 * doc/extend.texi: Update the built-in documentation file for the
20673 new built-in function.
20675 2017-07-24 Jakub Jelinek <jakub@redhat.com>
20678 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
20680 * doc/generic.texi: Likewise.
20681 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
20682 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
20684 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
20686 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
20687 (aarch64_mls_elt_merge<mode>): Likewise.
20689 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
20691 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
20692 having __cxa_atexit.
20694 2017-07-23 Michael Collison <michael.collison@arm.com>
20696 * config/arm/arm.c (arm_option_override): Deprecate
20697 use of -mstructure-size-boundary.
20698 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
20699 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
20701 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20704 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
20705 Reduce cost estimate for direct moves.
20707 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
20710 * config/i386/i386.c (ix86_option_override_internal): Disable
20711 BMI, BMI2 and TBM instructions for -m16.
20713 2017-07-21 Carl Love <cel@us.ibm.com>
20715 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20716 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20717 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20718 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20719 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20720 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20721 VMULOSW): New enum "unspec" values.
20722 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20723 altivec_vmulosw): New patterns.
20724 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20725 VMULOSW): Add definitions.
20727 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
20729 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
20730 (qdf24xx): Likewise.
20731 * config/aarch64/aarch64-options-extensions.def (rdma); New.
20732 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
20733 (AARCH64_FL_V8_1): Renumber.
20734 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
20735 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
20736 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
20737 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
20738 rdma to feature modifiers list.
20740 2017-07-21 Yury Gribov <tetra2005@gmail.com>
20742 PR middle-end/56727
20743 * ipa-visibility (function_and_variable_visibility): Convert
20744 recursive PLT call to direct call if appropriate.
20746 2017-07-21 Andrew Pinski <apinski@cavium.com>
20748 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
20749 operand 1 to see if the types precision matches.
20750 * fold-const.c (operand_equal_p): Likewise.
20752 2017-07-21 Richard Biener <rguenther@suse.de>
20754 PR tree-optimization/81303
20755 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
20756 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
20757 (vect_peeling_hash_get_lowest_cost): Adjust.
20758 (vect_enhance_data_refs_alignment): Likewise. Use
20759 vect_get_peeling_costs_all_drs to compute the penalty for no
20760 peeling to match up costs.
20762 2017-07-21 Richard Biener <rguenther@suse.de>
20764 PR tree-optimization/81500
20765 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
20766 we didn't identify a reduction path.
20768 2017-07-21 Tom de Vries <tom@codesourcery.com>
20769 Cesar Philippidis <cesar@codesourcery.com>
20771 PR gcov-profile/81442
20772 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
20775 2017-07-21 Tom de Vries <tom@codesourcery.com>
20778 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
20780 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
20781 nvptx_override_options_after_change.
20783 2017-07-21 Ulrich Drepper <drepper@redhat.com>
20785 * dwarf2out.c (output_file_names): Avoid double testing for
20786 dwarf_version >= 5.
20788 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
20790 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
20792 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20794 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
20796 (try_crossjump_to_edge): Do not punt on partitioned functions.
20798 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20800 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
20801 Put all BBs reachable only via paths crossing cold region to cold
20803 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
20805 2016-07-21 Richard Biener <rguenther@suse.de>
20807 PR tree-optimization/81303
20808 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
20809 into account prologue and epilogue iterations when raising
20810 min_profitable_iters to sth at least covering one vector iteration.
20812 2017-07-21 Tamar Christina <tamar.christina@arm.com>
20814 * config/arm/arm.c (arm_test_cpu_arch_dat):
20817 2017-07-20 Nathan Sidwell <nathan@acm.org>
20819 Remove TYPE_METHODS.
20820 * tree.h (TYPE_METHODS): Delete.
20821 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
20822 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
20823 (dbxout_type_methods): Scan TYPE_FIELDS.
20824 (dbxout_type): Don't check TYPE_METHODS here.
20825 * function.c (use_register_for_decl): Always ignore register for
20826 class types when not optimizing.
20827 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
20828 * tree.c (free_lang_data_in_type): Stitch out member functions and
20829 templates from TYPE_FIELDS.
20830 (build_distinct_type_copy, verify_type_variant,
20831 verify_type): Member fns are on TYPE_FIELDS.
20832 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
20833 * tree-pretty-print.c (dump_generic_node): Likewise.
20835 2017-07-20 Jakub Jelinek <jakub@redhat.com>
20838 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
20839 V2TImode and V4TImode.
20840 (ix86_expand_vector_extract): Likewise.
20841 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
20842 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
20843 (ssescalarmode): Handle V4TImode and V2TImode.
20844 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
20845 (*vec_extractv2ti, *vec_extractv4ti): New insns.
20846 (VEXTRACTI128_MODE): New mode iterator.
20847 (splitter for *vec_extractv?ti first element): New.
20848 (VEC_INIT_MODE): New mode iterator.
20849 (vec_init<mode>): Consolidate 3 expanders into one using
20850 VEC_INIT_MODE mode iterator.
20852 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20854 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
20855 non_spilled_static_chain_regno_p.
20857 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20859 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
20861 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
20863 * bb-reorder.c (connect_traces): Allow copying of blocks within
20866 2017-07-20 Richard Biener <rguenther@suse.de>
20868 * gimple.h (gimple_phi_result): Add gphi * overload.
20869 (gimple_phi_result_ptr): Likewise.
20870 (gimple_phi_arg): Likewise. Adjust index assert to only
20871 allow actual argument accesses rather than all slots available
20873 (gimple_phi_arg_def): Add gphi * overload.
20874 * tree-phinodes.c (make_phi_node): Initialize only actual
20876 (resize_phi_node): Clear memory not covered by old node,
20877 do not initialize excess argument slots.
20878 (reserve_phi_args_for_new_edge): Initialize new argument slot
20881 2017-07-20 Bin Cheng <bin.cheng@arm.com>
20883 PR tree-optimization/81388
20885 2016-07-21 Bin Cheng <bin.cheng@arm.com>
20887 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
20888 by removing computation of may_be_zero.
20890 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20891 Tom de Vries <tom@codesourcery.com>
20893 PR middle-end/81030
20894 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
20895 when gimple level profile disagrees with what RTL expander did.
20897 2017-07-20 Richard Biener <rguenther@suse.de>
20899 PR tree-optimization/61171
20900 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
20901 (vect_analyze_stmt): Add slp instance parameter.
20902 (vectorizable_reduction): Likewise.
20903 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
20904 (vect_is_simple_reduction): Deal with chains not detected
20905 as SLP reduction chain, specifically not properly associated
20906 chains containing a mix of plus/minus.
20907 (get_reduction_op): Remove.
20908 (get_initial_defs_for_reduction): Simplify, pass in whether
20909 this is a reduction chain, pass in the SLP node for the PHIs.
20910 (vect_create_epilog_for_reduction): Get the SLP instance as
20912 (vectorizable_reduction): Get the SLP instance as arg.
20913 During analysis remember the SLP node with the PHIs in the
20914 instance. Simplify getting at the vectorized reduction PHIs.
20915 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
20916 through SLP instance.
20917 (vect_slp_analyze_operations): Likewise.
20918 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
20919 (vect_transform_stmt): Likewise.
20921 2017-07-20 Tom de Vries <tom@codesourcery.com>
20923 PR tree-optimization/81489
20924 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
20925 read of phi arg location to before loop that modifies phi.
20927 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20929 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
20932 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20934 PR middle-end/81331
20935 * except.c (execute): Fix ordering issue.
20937 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20939 PR rtl-optimization/81423
20940 * combine.c (make_compound_operation_int): Don't try to optimize
20941 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
20943 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20945 PR rtl-optimization/81423
20946 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
20947 with a constant that is -1 in the truncated to mode.
20949 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20951 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
20952 (determine_unlikely_bbs): ... here.
20953 * predict.h (propagate_unlikely_bbs_forward): Declare.
20954 * cfgexpand.c (pass_expand::execute): Use it.
20955 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
20957 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
20958 propagate_unlikely_bbs_forward.
20960 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20962 PR middle-end/81331
20963 * except.c (maybe_add_nop_after_section_switch): New function.
20966 2017-07-19 Tom de Vries <tom@codesourcery.com>
20968 * gimple.h (gimple_phi_set_arg): Make assert more strict.
20970 2017-07-19 Tom de Vries <tom@codesourcery.com>
20972 * gimple.h (gimple_phi_arg): Make assert more strict.
20974 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
20976 * config.gcc (powerpc*-*-*): Add mmintrin.h.
20977 * config/rs6000/mmintrin.h: New file.
20978 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
20980 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20982 PR tree-optimization/81346
20983 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
20985 2017-07-19 Tom de Vries <tom@codesourcery.com>
20987 * config/nvptx/nvptx.md (VECIM): Add V2DI.
20989 2017-07-19 Tom de Vries <tom@codesourcery.com>
20991 * config/nvptx/nvptx-modes.def: Add V2DImode.
20992 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
20993 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
20994 (nvptx_output_mov_insn): Handle lack of mov.b128.
20995 (nvptx_print_operand): Handle 'H' and 'L' codes.
20996 (nvptx_vector_mode_supported): Allow V2DImode.
20997 (nvptx_preferred_simd_mode): New function.
20998 (nvptx_data_alignment): New function.
20999 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
21000 nvptx_preferred_simd_mode.
21001 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
21003 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
21005 2017-07-19 Tom de Vries <tom@codesourcery.com>
21007 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
21008 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
21009 (nvptx_vector_mode_supported): New function. Allow V2SImode.
21010 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
21011 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
21012 (mov<VECIM>_insn): New define_insn.
21013 (define_expand "mov<VECIM>): New define_expand.
21015 2017-07-19 Tom de Vries <tom@codesourcery.com>
21017 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
21019 2017-07-19 Jakub Jelinek <jakub@redhat.com>
21021 PR tree-optimization/81346
21022 * fold-const.h (fold_div_compare, range_check_type): Declare.
21023 * fold-const.c (range_check_type): New function.
21024 (build_range_check): Use range_check_type.
21025 (fold_div_compare): No longer static, rewritten into
21026 a match.pd helper function.
21027 (fold_comparison): Don't call fold_div_compare here.
21028 * match.pd (X / C1 op C2): New optimization using fold_div_compare
21029 as helper function.
21031 2017-07-19 Nathan Sidwell <nathan@acm.org>
21033 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
21034 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
21035 * tree.c (find_decls_types_r, verify_type): Use
21036 TYPE_{MIN,MAX}_VALUE_RAW.
21037 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
21038 (hash_tree): Likewise.
21039 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
21041 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
21044 2017-07-18 Tom de Vries <tom@codesourcery.com>
21046 PR middle-end/81464
21047 * omp-expand.c (expand_omp_for_static_chunk): Handle
21048 equal-argument loop exit phi.
21050 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
21053 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
21054 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
21055 operand 2 predicate.
21056 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
21057 operand 2 predicate.
21058 (ror,rol -> rorx splitters): Use const_int_operand as
21059 operand 2 predicate.
21061 2017-06-18 Richard Biener <rguenther@suse.de>
21063 PR tree-optimization/81410
21064 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
21065 the gap in the ! slp_perm SLP case after each group.
21067 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
21069 PR middle-end/81463
21070 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
21073 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
21075 PR middle-end/81462
21076 * predict.c (set_even_probabilities): Cleanup; do not affect
21077 probabilities that are already known.
21078 (combine_predictions_for_bb): Call even when count is set.
21080 2017-07-18 Nathan Sidwell <nathan@acm.org>
21082 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
21085 2017-07-18 Bin Cheng <bin.cheng@arm.com>
21088 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
21089 optimization for loop niter analysis.
21091 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
21094 * config/avr/avr.c (avr_optimize_casesi): Don't use
21095 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
21097 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
21099 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
21100 body_cost_vec from _vect_peel_extended_info.
21101 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
21102 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
21105 2017-07-18 Bin Cheng <bin.cheng@arm.com>
21107 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
21109 2017-07-18 Richard Biener <rguenther@suse.de>
21111 PR tree-optimization/80620
21112 PR tree-optimization/81403
21113 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
21114 info when re-using a VN table entry.
21116 2017-07-18 Richard Biener <rguenther@suse.de>
21118 PR tree-optimization/81418
21119 * tree-vect-loop.c (vectorizable_reduction): Properly compute
21120 vectype_in. Verify that with lane-reducing reduction operations
21121 we have a single def-use cycle.
21123 2017-07-17 Carl Love <cel@us.ibm.com>
21125 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
21127 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21128 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21129 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21130 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21131 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21132 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21133 VMULOSW): New enum "unspec" values.
21134 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
21135 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
21136 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21137 altivec_vmulosw): New patterns.
21138 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21139 VMULOSW): Add definitions.
21141 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
21143 * config/alpha/alpha.c: Include predict.h.
21145 2017-07-17 Yury Gribov <tetra2005@gmail.com>
21147 * tree-vrp.c (compare_assert_loc): Fix comparison function
21148 to return predictable results.
21150 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21152 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
21154 (subdi3): Likewise.
21155 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
21156 * doc/invoke.texi (mexpand-adddi): Update text.
21158 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21160 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
21161 that also clobbers the CC register. The old expand code is moved
21163 (*arc_clzsi2): ... here.
21164 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
21165 the CC register. The old expand code is moved to ...
21166 (arc_ctzsi2): ... here.
21168 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21170 * config/arc/arc.opt (mindexed-loads): Use initial value
21171 TARGET_INDEXED_LOADS_DEFAULT.
21172 (mauto-modify-reg): Use initial value
21173 TARGET_AUTO_MODIFY_REG_DEFAULT.
21174 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
21175 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
21176 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
21177 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
21179 2017-07-17 Martin Liska <mliska@suse.cz>
21182 * opts.c (finish_options): Do not allow -fgnu-tm
21183 w/ -fsanitize={kernel-,}address. Say sorry.
21185 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21188 * tree-loop-distribution.c (classify_partition): Only assert on
21189 numer of iterations.
21190 (merge_dep_scc_partitions): Delete prameter. Update function call.
21191 (distribute_loop): Remove code handling loop with unknown niters.
21192 (pass_loop_distribution::execute): Skip loop with unknown niters.
21194 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21197 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
21198 function sort_partitions_by_post_order.
21200 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21202 PR tree-optimization/81374
21203 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
21204 the max index of basic blocks, rather than number of basic blocks.
21206 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21208 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
21210 (arc_legitimate_pic_operand_p): Likewise.
21211 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
21213 (arc_needs_pcl_p): Likewise.
21214 (arc_legitimate_pc_offset_p): Likewise.
21215 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
21216 function is also used in constrains.md.
21217 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
21218 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
21219 PLUS. Only return true/false in known cases, otherwise assert.
21220 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
21221 is already called in arc_legitimate_constant_p.
21222 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
21224 (LEGITIMATE_PIC_OPERAND_P): Use
21225 arc_raw_symbolic_reference_mentioned_p function.
21226 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
21231 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21232 Andrew Burgess <andrew.burgess@embecosm.com>
21234 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
21235 (arc_return_address_register): New function.
21236 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
21237 (arc_handle_fndecl_attribute): Add naked attribute.
21238 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21239 (TARGET_WARN_FUNC_RETURN): Likewise.
21240 (arc_allocate_stack_slots_for_args): New function.
21241 (arc_warn_func_return): Likewise.
21242 (machine_function): Change type fn_type.
21243 (arc_compute_function_type): Consider new naked function type,
21244 change function return type.
21245 (arc_must_save_register): Adapt to handle new
21246 arc_compute_function_type's return type.
21247 (arc_expand_prologue): Likewise.
21248 (arc_expand_epilogue): Likewise.
21249 (arc_return_address_regs): Delete.
21250 (arc_return_address_register): New function.
21251 (arc_epilogue_uses): Use above function.
21252 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
21253 (arc_function_type): Change encoding, add naked type.
21254 (ARC_INTERRUPT_P): Change to handle the new encoding.
21255 (ARC_FAST_INTERRUPT_P): Likewise.
21256 (ARC_NORMAL_P): Define.
21257 (ARC_NAKED_P): Likewise.
21258 (arc_compute_function_type): Delete prototype.
21259 * config/arc/arc.md (in_ret_delay_slot): Use
21260 arc_return_address_register function.
21261 (simple_return): Likewise.
21262 (p_return_i): Likewise.
21264 2017-07-17 Jakub Jelinek <jakub@redhat.com>
21266 PR tree-optimization/81428
21267 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
21268 can't be built for those types.
21270 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
21272 Remove stuff dead since r239246.
21274 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
21275 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
21276 (avr_inform_devices): Remove dead stuff.
21278 2017-07-17 Tamar Christina <tamar.christina@arm.com>
21280 * config/arm/arm_neon.h: Fix softp typo.
21282 2017-07-17 Jakub Jelinek <jakub@redhat.com>
21284 PR tree-optimization/81365
21285 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
21286 aggregate moves onto bb predecessor edges, make sure there are no
21287 loads that could alias the lhs in between the start of bb and the
21290 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
21293 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
21294 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
21295 [LSHIFTRT, outer_code = TRUNCATE]: Same.
21297 2017-07-17 Jakub Jelinek <jakub@redhat.com>
21299 PR tree-optimization/81396
21300 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
21301 (init_symbolic_number): Initialize it to 1.
21302 (perform_symbolic_merge): Add n_ops from both operands into the new
21304 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
21305 without base_addr as useless if they need more than one operation.
21306 (bswap_replace): Handle !bswap case for NULL base_addr.
21308 2017-07-17 Tom de Vries <tom@codesourcery.com>
21311 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
21314 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
21316 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
21317 conditional builtin define __FIX_LEON3FT_B2BST.
21319 2017-07-17 Daniel Cederman <cederman@gaisler.com>
21321 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
21322 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
21325 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
21327 PR rtl-optimization/81424
21328 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
21329 to remove potential trapping from operands if -fnon-call-exceptions.
21331 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21333 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
21334 profile_proability for scalling.
21335 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
21337 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21339 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
21341 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21343 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
21344 fixpoint arithmetics.
21346 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21348 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
21349 fixpoint arithmetics.
21351 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21353 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
21354 fixpoint arithmetics.
21356 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21358 * profile-count.h (profile_probability::from_reg_br_prob_note,
21359 profile_probability::to_reg_br_prob_note): New functions.
21360 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
21361 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
21362 * predict.c (probability_reliable_p): Update.
21363 (edge_probability_reliable_p): Update.
21364 (br_prob_note_reliable_p): Update.
21365 (invert_br_probabilities): Update.
21366 (add_reg_br_prob_note): New function.
21367 (combine_predictions_for_insn): Update.
21368 * asan.c (asan_clear_shadow): Update.
21369 * cfgbuild.c (compute_outgoing_frequencies): Update.
21370 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21371 (update_br_prob_note): Update.
21372 (rtl_verify_edges): Update.
21373 (purge_dead_edges): Update.
21374 (fixup_reorder_chain): Update.
21375 * emit-rtl.c (try_split): Update.
21376 * ifcvt.c (cond_exec_process_insns): Update.
21377 (cond_exec_process_if_block): Update.
21378 (dead_or_predicable): Update.
21379 * internal-fn.c (expand_addsub_overflow): Update.
21380 (expand_neg_overflow): Update.
21381 (expand_mul_overflow): Update.
21382 * loop-doloop.c (doloop_modify): Update.
21383 * loop-unroll.c (compare_and_jump_seq): Update.
21384 * optabs.c (emit_cmp_and_jump_insn_1): Update.
21385 * predict.h: Update.
21386 * reorg.c (mostly_true_jump): Update.
21388 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
21389 * config/alpha/alpha.c (emit_unlikely_jump): Update.
21390 * config/arc/arc.c: (emit_unlikely_jump): Update.
21391 * config/arm/arm.c: (emit_unlikely_jump): Update.
21392 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
21393 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
21394 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
21395 (ix86_print_operand): Update.
21396 (ix86_split_fp_branch): Update.
21397 (predict_jump): Update.
21398 * config/ia64/ia64.c (ia64_print_operand): Update.
21399 * config/mmix/mmix.c (mmix_print_operand): Update.
21400 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
21401 (rs6000_expand_split_stack_prologue): Update.
21402 * config/rs6000/rs6000.c: Update.
21403 * config/s390/s390.c (s390_expand_vec_strlen): Update.
21404 (s390_expand_vec_movstr): Update.
21405 (s390_expand_cs_tdsi): Update.
21406 (s390_expand_split_stack_prologue): Update.
21407 * config/sh/sh.c (sh_print_operand): Update.
21408 (expand_cbranchsi4): Update.
21409 (expand_cbranchdi4): Update.
21410 * config/sparc/sparc.c (output_v9branch): Update.
21411 * config/spu/spu.c (get_branch_target): Update.
21412 (ea_load_store_inline): Update.
21413 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
21414 * config/tilepro/tilepro.c: Update.
21416 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
21418 * gimplify.c (mostly_copy_tree_r): Revert latest change.
21419 (gimplify_save_expr): Likewise.
21421 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21423 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
21425 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21427 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
21429 * timevar.def (TV_IPA_FNSUMMARY): Define.
21431 2017-07-16 Daniel Cederman <cederman@gaisler.com>
21433 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
21434 to back store errata sensitive sequence from being generated.
21435 (sqrtdf2_fix): Likewise.
21437 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21439 * tree-ssa-threadupdate.c (compute_path_counts,
21440 update_joiner_offpath_counts): Use profile_probability.
21442 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
21445 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21447 * config/arm/arm-c.c (arm_cpu_builtins): Define
21448 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
21450 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
21452 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21453 array entries to represent __ieee128 versions of the
21454 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
21455 scalar_extract_sig, and scalar_insert_exp built-in functions.
21456 (altivec_resolve_overloaded_builtin): Add special case handling
21457 for the __builtin_scalar_insert_exp function, as represented by
21458 the P9V_BUILTIN_VEC_VSIEDP constant.
21459 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
21460 exponent support for __ieee128 argument.
21461 (VSESQP): Add scalar extract signature support for __ieee128
21463 (VSTDCNQP): Add scalar test negative support for __ieee128
21465 (VSIEQP): Add scalar insert exponent support for __int128 argument
21466 with __ieee128 result.
21467 (VSIEQPF): Add scalar insert exponent support for __ieee128
21468 argument with __ieee128 result.
21469 (VSTDCQP): Add scalar test data class support for __ieee128
21471 (VSTDCNQP): Add overload support for scalar test negative with
21472 __ieee128 argument.
21473 (VSTDCQP): Add overload support for scalar test data class
21474 __ieee128 argument.
21475 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
21476 UNSPEC_VSX_SXSIGDP.
21477 (UNSPEC_VSX_SIEXPQP): New constant.
21478 (xsxexpqp): New insn for VSX scalar extract exponent quad
21480 (xsxsigqp): New insn for VSX scalar extract significand quad
21482 (xsiexpqpf): New insn for VSX scalar insert exponent quad
21483 precision with floating point argument.
21484 (xststdcqp): New expand for VSX scalar test data class quad
21486 (xststdcnegqp): New expand for VSX scalar test negative quad
21488 (xststdcqp): New insn to match expansions for VSX scalar test data
21489 class quad precision and VSX scalar test negative quad precision.
21490 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
21491 special case operand checking to enforce that second operand of
21492 VSX scalar test data class with quad precision argument is a 7-bit
21494 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21495 prototypes and descriptions of __ieee128 versions of
21496 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
21497 scalar_test_data_class, and scalar_test_neg built-in functions.
21499 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21501 PR tree-optimization/81162
21502 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
21503 replace a negate with an add.
21505 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
21507 * doc/invoke.texi (arm/-mcpu): Document +crypto.
21509 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21511 * config/arm/arm-c.c (arm_cpu_builtins): Define
21512 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
21514 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21516 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
21517 (armv8-r): Set ARM Cortex-R52 as default CPU.
21518 * config/arm/arm-tables.opt: Regenerate.
21519 * config/arm/arm-tune.md: Regenerate.
21520 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
21522 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
21523 extension for -mcpu=cortex-r52.
21525 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21527 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
21528 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
21529 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
21530 (fp-armv8): Define it as FP_ARMv8 only.
21531 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
21532 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
21534 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
21535 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
21536 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
21537 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
21538 than TARGET_FPU_ARMV8.
21539 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
21540 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
21541 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
21543 * config/arm/neon.md (neon_vrint): Likewise.
21544 (neon_vcvt): Likewise.
21545 (neon_<fmaxmin_op><mode>): Likewise.
21546 (<fmaxmin><mode>3): Likewise.
21547 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
21548 * config/arm/predicates.md (arm_cond_move_operator): Check against
21549 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
21551 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
21553 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
21554 to top of function.
21556 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21558 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
21559 loop in comment with memset.
21561 2017-07-14 Martin Liska <mliska@suse.cz>
21563 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
21564 * dwarf2out.c (is_java): Remove the function.
21565 (output_pubname): Remove usage of the function.
21566 (lower_bound_default): Remove usage of DW_LANG_Java.
21567 (gen_compile_unit_die): Likewise.
21568 * gcc.c: Remove compiler defaults for .java and .zip files.
21569 * gimple-expr.c (remove_suffix): Change as there's no longer
21570 extension than 4-letter one.
21571 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
21572 (gimplify_save_expr): Likewise.
21573 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
21574 as it's possible even for other languages than Java.
21575 * langhooks.h (struct lang_hooks): Remove Java from a comment.
21576 * lto-opts.c (lto_write_options): Remove reference to Java.
21577 * opts.c (strip_off_ending): Update file extension handling.
21578 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
21579 * tree-eh.c (lower_resx): Likewise.
21580 * tree.c (free_lang_data_in_type): Remove dead code.
21581 (find_decls_types_r): Likewise.
21582 (build_common_builtin_nodes): Remove Java from a comment.
21583 (verify_type): Remove dead code.
21584 * varasm.c (assemble_external): Remove Java from a comment.
21586 2017-07-14 Martin Liska <mliska@suse.cz>
21588 * opts.c (finish_options): Add quotes.
21589 (common_handle_option): Likewise.
21591 2017-07-14 Martin Liska <mliska@suse.cz>
21593 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
21594 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
21595 Remove N_SO_PASCAL.
21596 * dwarf2out.c (lower_bound_default): Do not handle
21598 (gen_compile_unit_die): Likewise.
21599 * gcc.c: Remove default extension binding for GNU Pascal.
21600 * stmt.c: Remove Pascal language from a comment.
21601 * xcoffout.c: Likewise.
21603 2017-07-13 David Malcolm <dmalcolm@redhat.com>
21606 * diagnostic-show-locus.c (fixit_cmp): New function.
21607 (layout::layout): Sort m_fixit_hints.
21608 (column_range::column_range): Assert that the values are valid.
21609 (struct char_span): New struct.
21610 (correction::overwrite): New method.
21611 (struct source_line): New struct.
21612 (line_corrections::add_hint): Add assertions. Reimplement memcpy
21613 calls in terms of classes source_line and char_span, and
21614 correction::overwrite.
21615 (selftest::test_overlapped_fixit_printing_2): New function.
21616 (selftest::diagnostic_show_locus_c_tests): Call it.
21618 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
21620 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
21621 early if there is no lhs.
21623 2017-07-13 Martin Liska <mliska@suse.cz>
21625 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
21626 (gen_reference_type_die): Likewise.
21627 * stor-layout.c: Remove Pascal-related comment.
21629 2017-07-13 Martin Liska <mliska@suse.cz>
21631 * opts.c (finish_options): Add quotes to error messages.
21632 (parse_sanitizer_options): Likewise.
21634 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
21636 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
21638 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
21640 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
21642 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
21644 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
21646 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
21648 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
21651 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
21652 provides the hardware capability bits, define the macro
21653 __BUILTIN_CPU_SUPPORTS__.
21654 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
21655 if GLIBC does not provide the hardware capability bits. Add a
21656 gcc_unreachable call if the built-in cpu function is neither
21657 __builtin_cpu_is nor __builtin_cpu_supports.
21658 (rs6000_get_function_versions_dispatcher): Change the warning
21659 that an old GLIBC is used which does not export the capability
21660 bits to be an error.
21661 * doc/extend.texi (target_clones attribute): Document the
21662 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
21663 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
21664 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
21665 the macros defined by GCC if the newer GLIBC is available.
21667 2017-07-12 Jeff Law <law@redhat.com>
21669 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
21670 remaining includes slightly.
21671 * config/riscv/riscv-builtins.c: Include profile-count.h.
21673 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21676 * config/avr/avr.c (avr_set_current_function): In diagnostic
21677 messages: Quote keywords and (parts of) identifiers.
21678 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
21681 2017-07-12 Carl Love <cel@us.ibm.com>
21683 * config/rs6000/rs6000-c.c: Add support for built-in functions
21684 vector bool char vec_revb (vector bool char);
21685 vector bool short vec_revb (vector short char);
21686 vector bool int vec_revb (vector bool int);
21687 vector bool long long vec_revb (vector bool long long);
21688 * doc/extend.texi: Update the built-in documentation file for the
21689 new built-in functions.
21691 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21693 * config/s390/s390.md: Remove movcc splitter.
21695 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21697 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
21698 load/store on condition.
21700 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21703 * config/avr/avr.c (avr_encode_section_info)
21704 [progmem && !TREE_READONLY]: Error if progmem object needs
21707 2017-07-11 Michael Collison <michael.collison@arm.com>
21709 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
21712 2017-07-11 Carl Love <cel@us.ibm.com>
21714 * config/rs6000/rs6000-c.c: Add support for builtins
21715 vector unsigned int vec_parity_lsbb (vector signed int);
21716 vector unsigned int vec_parity_lsbb (vector unsigned int);
21717 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
21718 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
21719 vector unsigned long long vec_parity_lsbb (vector signed long long);
21720 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
21721 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
21722 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
21723 * doc/extend.texi: Update the built-in documentation file for the
21724 new built-in functions.
21726 2017-07-11 David Malcolm <dmalcolm@redhat.com>
21728 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
21729 (layout::m_primary_loc): New field.
21730 (layout::layout): Initialize new field. Move location filtering
21731 logic from here to...
21732 (layout::maybe_add_location_range): ...this new method. Add
21733 support for filtering to just the lines already specified by other
21735 (layout::will_show_line_p): New method.
21736 (gcc_rich_location::add_location_if_nearby): New method.
21737 (selftest::test_add_location_if_nearby): New test function.
21738 (selftest::diagnostic_show_locus_c_tests): Call it.
21739 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
21742 2017-07-11 Tom de Vries <tom@codesourcery.com>
21744 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
21745 (bb_first_real_insn): New function.
21746 (nvptx_single): Add extra initialization of broadcasted condition
21749 2017-07-11 Nathan Sidwell <nathan@acm.org>
21751 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
21753 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
21755 * doc/extend.texi (AVR Function Attributes): Remove weblink to
21756 Binutils doc as TEXI will mess them up.
21757 * doc/invoke.texi (AVR Options): Same here.
21759 2017-07-11 Daniel Cederman <cederman@gaisler.com>
21761 * config/sparc/sparc.opt (mfix-ut700): New option.
21762 (mfix-gr712rc): Likewise.
21763 (sparc_fix_b2bst): New variable.
21764 * doc/invoke.texi (SPARC options): Document them.
21765 (ARM options): Fix warnings.
21766 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
21767 instructions to prevent sequences that can trigger the store-store
21768 errata for certain LEON3FT processors.
21769 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
21770 (sparc_option_override): Set sparc_fix_b2bst appropriately.
21771 * config/sparc/sparc.md (fix_b2bst): New attribute.
21772 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
21774 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
21777 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
21779 (*rsqrtsf2_sse): Ditto.
21781 (div<mode>3): Macroize insn from divdf3 and divsf3
21782 using MODEF mode iterator.
21784 2017-07-10 Martin Sebor <msebor@redhat.com>
21786 PR tree-optimization/80397
21787 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
21788 instead of testing for equality to INTEGER_TYPE.
21790 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
21792 * config.gcc: Remove uclibc from arc target spec.
21794 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
21796 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
21798 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21801 * lto-wrapper.c (remove_option): New function.
21802 (merge_and_complain): Merge PIC/PIE options more realistically.
21804 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21806 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
21810 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
21811 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
21812 * config.in: Regenerate.
21813 * configure: Regenerate.
21814 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
21815 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
21816 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
21817 (TARGET_GASISR_PROLOGUES): ...target mask.
21818 * common/config/avr/avr-common.c
21819 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
21820 Set -mgas-isr-prologues.
21821 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
21822 INSERT_PASS_BEFORE for it.
21823 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
21824 * config/avr/avr.c (avr_option_override)
21825 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
21826 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
21827 (avr_attribute_table) <no_gccisr>: Add new function attribute.
21828 (avr_set_current_function) <is_no_gccisr>: Init machine field.
21829 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
21831 (make_avr_pass_pre_proep): New function.
21832 (emit_push_sfr) <treg>: Add argument to function and use it
21833 instead of TMP_REG.
21834 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
21835 and set machine->gasisr.yes.
21836 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
21837 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
21838 __gcc_isr.n_pushed to .L__stack_usage.
21839 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
21840 (avr_asm_final_postscan_insn): ...this new static function.
21841 * config/avr/avr.h (machine_function)
21842 <is_no_gccisr, use_L__stack_usage>: New fields.
21843 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
21844 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
21845 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
21846 (gasisr, *gasisr): New expander and insn.
21847 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21848 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
21849 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
21851 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
21853 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
21856 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21858 Move jump-tables out of .text again.
21861 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
21862 (ASM_OUTPUT_ADDR_VEC): New function.
21863 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
21864 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
21865 INSN_ADDRESSes as asm comment.
21866 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
21867 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
21868 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
21869 * config/avr/avr.md (*tablejump): Adjust comment.
21870 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
21871 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
21873 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
21874 (avr_output_addr_vec): New proto.
21875 (avr_log_t) <insn_addresses>: New field.
21877 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
21880 * config/i386/i386.c (ix86_function_arg_advance): Set
21881 outgoing_args_on_stack to true if there are outgoing arguments
21883 (ix86_function_arg): Likewise.
21884 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
21885 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
21886 * config/i386/i386.h (machine_function): Add
21887 outgoing_args_on_stack.
21889 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
21891 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
21892 supporting pthreds.
21893 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
21895 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
21897 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
21898 (REAL_H): Remove $(MACHMODE_H).
21899 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
21901 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
21902 $(MACHMODE_H) and double-int.h.
21903 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
21905 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
21908 2017-07-07 Andrew Pinski <apinski@cavium.com>
21910 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
21911 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
21913 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21915 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
21916 Add warning if GCC was not configured to link against a GLIBC that
21917 exports the hardware capability bits.
21918 (make_resolver_func): Make resolver function private and not a
21919 COMDAT function. Create the name with clone_function_name instead
21920 of make_unique_name.
21923 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
21924 correct operand in doing the split.
21926 2017-07-07 Carl Love <cel@us.ibm.com>
21928 * config/rs6000/rs6000-c: Add support for built-in function
21929 vector unsigned short vec_pack_to_short_fp32 (vector float,
21931 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
21932 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
21933 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
21934 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
21935 (convert_4f32_8i16): Add define_expand.
21936 * doc/extend.texi: Update the built-in documentation file for the
21937 new built-in function.
21939 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21941 * config/sparc/m8.md: New file.
21942 * config/sparc/sparc.md: Include m8.md.
21944 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21946 * config/sparc/sparc.opt: New option -mvis4b.
21947 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
21948 (sparc_option_override): Handle VIS4B.
21949 (enum sparc_builtins): Define
21950 SPARC_BUILTIN_DICTUNPACK{8,16,32},
21951 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
21952 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
21953 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
21954 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
21955 (check_constant_argument): New function.
21956 (sparc_vis_init_builtins): Define builtins
21957 __builtin_vis_dictunpack{8,16,32},
21958 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
21959 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
21960 __builtin_vis_fpcmpde{8,16,32}shl and
21961 __builtin_vis_fpcmpur{8,16,32}shl.
21962 (sparc_expand_builtin): Check that the constant operands to
21963 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
21964 constant and in range.
21965 * config/sparc/sparc-c.c (sparc_target_macros): Handle
21967 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
21968 (SPARC_IMM5_P): Likewise.
21969 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
21970 (enabled): Handle vis4b.
21971 (UNSPEC_DICTUNPACK): New unspec.
21972 (UNSPEC_FPCMPSHL): Likewise.
21973 (UNSPEC_FPUCMPSHL): Likewise.
21974 (UNSPEC_FPCMPDESHL): Likewise.
21975 (UNSPEC_FPCMPURSHL): Likewise.
21976 (cpu_feature): New CPU feature `vis4b'.
21977 (dictunpack{8,16,32}): New insns.
21978 (FPCSMODE): New mode iterator.
21979 (fpcscond): New code iterator.
21980 (fpcsucond): Likewise.
21981 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
21982 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
21983 (fpcmpde{8,16,32}{si,di}shl): Likewise.
21984 (fpcmpur{8,16,32}{si,di}shl): Likewise.
21985 * config/sparc/constraints.md: Define constraints `q' for unsigned
21986 2-bit integer constants and `t' for unsigned 5-bit integer
21988 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
21990 (imm5_operand_dictunpack16): Likewise.
21991 (imm5_operand_dictunpack32): Likewise.
21992 (imm2_operand): Likewise.
21993 * doc/invoke.texi (SPARC Options): Document -mvis4b.
21994 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
21995 ditunpack* and fpcmp*shl builtins.
21997 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21999 * config.gcc: Handle m8 in --with-{cpu,tune} options.
22000 * config.in: Add HAVE_AS_SPARC6 define.
22001 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
22003 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
22005 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
22006 (CPP_CPU_SPEC): Handle m8.
22007 (ASM_CPU_SPEC): Likewise.
22008 * config/sparc/sparc-opts.h (enum processor_type): Add
22010 * config/sparc/sparc.c (m8_costs): New struct.
22011 (sparc_option_override): Handle TARGET_CPU_m8.
22012 (sparc32_initialize_trampoline): Likewise.
22013 (sparc64_initialize_trampoline): Likewise.
22014 (sparc_issue_rate): Likewise.
22015 (sparc_register_move_cost): Likewise.
22016 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
22017 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
22018 (ASM_CPU64_DEFAULT_SPEC): Likewise.
22019 (CPP_CPU_SPEC): Handle M8.
22020 (ASM_CPU_SPEC): Likewise.
22021 (AS_M8_FLAG): Define.
22022 * config/sparc/sparc.md: Add m8 to the cpu attribute.
22023 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
22024 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
22026 * configure: Regenerate.
22027 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
22030 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22032 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
22034 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
22035 ("*movdi_insn_sp32"): Do not set v3pipe.
22036 ("*movsi_insn"): Likewise.
22037 ("*movdi_insn_sp64"): Likewise.
22038 ("*movsf_insn"): Likewise.
22039 ("*movdf_insn_sp32"): Likewise.
22040 ("*movdf_insn_sp64"): Likewise.
22041 ("*zero_extendsidi2_insn_sp64"): Likewise.
22042 ("*sign_extendsidi2_insn"): Likewise.
22043 ("*mov<VM32:mode>_insn"): Likewise.
22044 ("*mov<VM64:mode>_insn_sp64"): Likewise.
22045 ("*mov<VM64:mode>_insn_sp32"): Likewise.
22046 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
22047 ("<vlop:code><VL:mode>3"): Likewise.
22048 ("*not_<vlop:code><VL:mode>3"): Likewise.
22049 ("*nand<VL:mode>_vis"): Likewise.
22050 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
22051 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
22052 ("one_cmpl<VL:mode>2"): Likewise.
22053 ("faligndata<VM64:mode>_vis"): Likewise.
22054 ("alignaddrsi_vis"): Likewise.
22055 ("alignaddrdi_vis"): Likweise.
22056 ("alignaddrlsi_vis"): Likewise.
22057 ("alignaddrldi_vis"): Likewise.
22058 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
22059 ("bmaskdi_vis"): Likewise.
22060 ("bmasksi_vis"): Likewise.
22061 ("bshuffle<VM64:mode>_vis"): Likewise.
22062 ("cmask8<P:mode>_vis"): Likewise.
22063 ("cmask16<P:mode>_vis"): Likewise.
22064 ("cmask32<P:mode>_vis"): Likewise.
22065 ("pdistn<P:mode>_vis"): Likewise.
22066 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
22068 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22070 * config/sparc/sparc.md ("subtype"): New insn attribute.
22071 ("*wrgsr_sp64"): Set insn subtype.
22072 ("*rdgsr_sp64"): Likewise.
22073 ("alignaddrsi_vis"): Likewise.
22074 ("alignaddrdi_vis"): Likewise.
22075 ("alignaddrlsi_vis"): Likewise.
22076 ("alignaddrldi_vis"): Likewise.
22077 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
22078 ("fexpand_vis"): Likewise.
22079 ("fpmerge_vis"): Likewise.
22080 ("faligndata<VM64:mode>_vis"): Likewise.
22081 ("bshuffle<VM64:mode>_vis"): Likewise.
22082 ("cmask8<P:mode>_vis"): Likewise.
22083 ("cmask16<P:mode>_vis"): Likewise.
22084 ("cmask32<P:mode>_vis"): Likewise.
22085 ("fchksm16_vis"): Likewise.
22086 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
22087 ("fmean16_vis"): Likewise.
22088 ("fp<plusminus_insn>64_vis"): Likewise.
22089 ("<plusminus_insn>v8qi3"): Likewise.
22090 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
22091 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
22092 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
22093 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
22094 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
22095 ("*movqi_insn"): Likewise.
22096 ("*movhi_insn"): Likewise.
22097 ("*movsi_insn"): Likewise.
22098 ("movsi_pic_gotdata_op"): Likewise.
22099 ("*movdi_insn_sp32"): Likewise.
22100 ("*movdi_insn_sp64"): Likewise.
22101 ("movdi_pic_gotdata_op"): Likewise.
22102 ("*movsf_insn"): Likewise.
22103 ("*movdf_insn_sp32"): Likewise.
22104 ("*movdf_insn_sp64"): Likewise.
22105 ("*zero_extendhisi2_insn"): Likewise.
22106 ("*zero_extendqihi2_insn"): Likewise.
22107 ("*zero_extendqisi2_insn"): Likewise.
22108 ("*zero_extendqidi2_insn"): Likewise.
22109 ("*zero_extendhidi2_insn"): Likewise.
22110 ("*zero_extendsidi2_insn_sp64"): Likewise.
22111 ("ldfsr"): Likewise.
22112 ("prefetch_64"): Likewise.
22113 ("prefetch_32"): Likewise.
22114 ("tie_ld32"): Likewise.
22115 ("tie_ld64"): Likewise.
22116 ("*tldo_ldub_sp32"): Likewise.
22117 ("*tldo_ldub1_sp32"): Likewise.
22118 ("*tldo_ldub2_sp32"): Likewise.
22119 ("*tldo_ldub_sp64"): Likewise.
22120 ("*tldo_ldub1_sp64"): Likewise.
22121 ("*tldo_ldub2_sp64"): Likewise.
22122 ("*tldo_ldub3_sp64"): Likewise.
22123 ("*tldo_lduh_sp32"): Likewise.
22124 ("*tldo_lduh1_sp32"): Likewise.
22125 ("*tldo_lduh_sp64"): Likewise.
22126 ("*tldo_lduh1_sp64"): Likewise.
22127 ("*tldo_lduh2_sp64"): Likewise.
22128 ("*tldo_lduw_sp32"): Likewise.
22129 ("*tldo_lduw_sp64"): Likewise.
22130 ("*tldo_lduw1_sp64"): Likewise.
22131 ("*tldo_ldx_sp64"): Likewise.
22132 ("*mov<VM32:mode>_insn"): Likewise.
22133 ("*mov<VM64:mode>_insn_sp64"): Likewise.
22134 ("*mov<VM64:mode>_insn_sp32"): Likewise.
22136 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22138 * config/sparc/sparc.md ("type"): New insn type viscmp.
22139 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
22141 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
22142 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
22143 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
22144 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
22146 ("n7_vis_logical_11cycle"): Likewise.
22147 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
22148 * config/sparc/niagara2.md ("niag3_vis": Likewise.
22149 * config/sparc/niagara.md ("niag_vis"): Likewise.
22150 * config/sparc/ultra3.md ("us3_fga"): Likewise.
22151 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
22153 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22155 * config/sparc/sparc.md: New instruction type `bmask'.
22156 (bmaskdi_vis): Use the `bmask' type.
22157 (bmasksi_vis): Likewise.
22158 * config/sparc/ultra3.md (us3_array): Likewise.
22159 * config/sparc/niagara7.md (n7_array): Likewise.
22160 * config/sparc/niagara4.md (n4_array): Likewise.
22161 * config/sparc/niagara2.md (niag2_vis): Likewise.
22162 (niag3_vis): Likewise.
22163 * config/sparc/niagara.md (niag_vis): Likewise.
22165 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22167 * ipa-comdats.c: Remove optimize check from gate.
22168 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
22169 for functions not optimized.
22170 (ipa_fn_summary_read): Skip optimize check.
22171 (ipa_fn_summary_write): Likewise.
22172 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
22174 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
22176 (can_inline_edge_p): Check flag_pcc_struct_return for match.
22177 (check_callers): Give up on caller which is not optimized.
22178 (inline_small_functions): Likewise.
22179 (ipa_inline): Do not give up when not optimizing.
22180 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
22181 away unoptimizes cdtors.
22182 (whole_program_function_and_variable_visibility): Do
22183 ipa_discover_readonly_nonaddressable_vars in LTO mode.
22184 * ipa.c (process_references): Do not check optimize.
22185 (symbol_table::remove_unreachable_nodes): Update optimize check.
22186 (set_writeonly_bit): Update optimize check.
22187 (pass_ipa_cdtor_merge::gate): Do not check optimize.
22188 (pass_ipa_single_use::gate): Remove.
22190 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22192 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
22193 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
22194 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
22195 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
22196 permute_load, permute_store, adjust_extract, adjust_splat,
22197 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
22198 replace_swap_with_copy, dump_swap_insn_table,
22199 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
22200 recombine_lvx_pattern, recombine_stvx_pattern,
22201 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
22202 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
22203 to file rs6000-p8swap.c.
22204 * config/rs6000/rs6000-p8swap.c: New file.
22205 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
22206 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
22207 and rs6000*-*-* targets.
22209 2017-07-06 David Malcolm <dmalcolm@redhat.com>
22211 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
22213 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22215 * lto-wrapper.c (merge_and_complain): Do not merge
22216 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
22217 fsigned_zeros, ftrapping_math, fwrapv.
22218 (append_compiler_options): Do not track these options.
22219 (append_linker_options): Likewie
22221 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22223 * cgraphunit.c (cgraph_node::finalize_function): When
22224 !flag_toplevel_reorde set no_reorder flag.
22225 (varpool_node::finalize_decl): Likewise.
22226 (symbol_table::compile): Drop no toplevel reorder path.
22228 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22230 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
22231 edges; zero probability is not better than uninitialized.
22233 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
22235 * asan.h (asan_sanitize_allocas_p): Declare.
22236 * asan.c (asan_sanitize_allocas_p): New function.
22237 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
22238 (handle_builtin_alloca): Likewise.
22239 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
22240 if !asan_sanitize_allocas_p.
22241 * params.def (asan-instrument-allocas): Add new option.
22242 * params.h (ASAN_PROTECT_ALLOCAS): Define.
22243 * opts.c (common_handle_option): Disable allocas sanitization for
22246 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
22248 * asan.c: Include gimple-fold.h.
22249 (get_last_alloca_addr): New function.
22250 (handle_builtin_stackrestore): Likewise.
22251 (handle_builtin_alloca): Likewise.
22252 (asan_emit_allocas_unpoison): Likewise.
22253 (get_mem_refs_of_builtin_call): Add new parameter, remove const
22254 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
22255 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
22256 (instrument_builtin_call): Pass gimple iterator to
22257 get_mem_refs_of_builtin_call.
22258 (last_alloca_addr): New global.
22259 * asan.h (asan_emit_allocas_unpoison): Declare.
22260 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
22261 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
22262 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
22263 if function calls alloca.
22264 * gimple-fold.c (replace_call_with_value): Remove static keyword.
22265 * gimple-fold.h (replace_call_with_value): Declare.
22266 * internal-fn.c: Include asan.h.
22267 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
22268 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
22270 2017-07-06 David Malcolm <dmalcolm@redhat.com>
22272 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
22273 (C_SELFTEST_FLAGS): New.
22274 (CPP_SELFTEST_FLAGS): New.
22275 (SELFTEST_DEPS): New, from deps of s-selftest.
22276 (C_SELFTEST_DEPS): New, from deps of s-selftest.
22277 (CPP_SELFTEST_DEPS): New.
22278 (selftest): Add dependency on s-selftest-c++.
22279 (s-selftest): Rename to...
22280 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
22281 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
22282 than SELFTEST_FLAGS.
22283 (selftest-gdb): Rename to...
22284 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
22286 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
22287 (selftest-valgrind): Rename to...
22288 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
22290 (selftest-valgrind): Reintroduce as an alias for
22291 selftest-c-valgrind.
22292 (s-selftest-c++): New.
22293 (selftest-c++-gdb): New.
22294 (selftest-c++-valgrind): New.
22296 2017-07-06 Olivier Hainque <hainque@adacore.com>
22298 * gcc.c (process_command): When deciding if undefined variables
22299 should be ignored when processing specs, accept "gcc -v" as well.
22301 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22303 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
22304 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
22306 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22308 * config/arm/arm-cpus.in (armv8-r): Add new entry.
22309 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
22310 * config/arm/arm-tables.opt: Regenerate.
22311 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
22313 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
22315 2017-07-06 Carl Love <cel@us.ibm.com>
22317 * ChangeLog: Clean up from mid air collision
22319 2017-07-06 Carl Love <cel@us.ibm.com>
22321 * config/rs6000/rs6000-c.c: Add support for built-in functions
22322 vector signed int vec_subc (vector signed int, vector signed int);
22323 vector signed __int128 vec_subc (vector signed __int128,
22324 vector signed __int128);
22325 vector unsigned __int128 vec_subc (vector unsigned __int128,
22326 vector unsigned __int128);
22327 vector signed int vec_sube (vector signed int, vector signed int,
22328 vector signed int);
22329 vector unsigned int vec_sube (vector unsigned int,
22330 vector unsigned int,
22331 vector unsigned int);
22332 vector signed __int128 vec_sube (vector signed __int128,
22333 vector signed __int128,
22334 vector signed__int128);
22335 vector unsigned __int128 vec_sube (vector unsigned __int128,
22336 vector unsigned __int128,
22337 vector unsigned __int128);
22338 vector signed int vec_subec (vector signed int, vector signed int,
22339 vector signed int);
22340 vector unsigned int vec_subec (vector unsigned int,
22341 vector unsigned int,
22342 vector unsigned int);
22343 vector signed __int128 vec_subec (vector signed __int128,
22344 vector signed __int128,
22345 vector signed__int128);
22346 vector unsigned __int128 vec_subec (vector unsigned __int128,
22347 vector unsigned __int128,
22348 vector unsigned __int128);
22349 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
22350 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
22351 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
22352 BU_ALTIVEC_OVERLOAD_X definitions.
22353 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
22354 * doc/extend.texi: Update the built-in documentation file for the new
22355 built-in functions.
22357 2017-07-06 David Malcolm <dmalcolm@redhat.com>
22360 * diagnostic-show-locus.c (layout::layout): Use start and finish
22361 spelling location for the start and finish of each range.
22362 * genmatch.c (linemap_client_expand_location_to_spelling_point):
22363 Add unused aspect param.
22364 * input.c (expand_location_1): Add "aspect" param, and use it
22365 to access the correct part of the location.
22366 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
22368 (expand_location_to_spelling_point): Likewise.
22369 (linemap_client_expand_location_to_spelling_point): Add "aspect"
22370 param, and pass it to expand_location_1.
22372 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
22374 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
22375 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
22376 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
22377 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
22378 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
22379 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
22380 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
22381 _mm_maskz_getmant_ss): New intrinsics.
22382 (__builtin_ia32_getexpss128_mask): Changed to ...
22383 __builtin_ia32_getexpss128_round ... this.
22384 (__builtin_ia32_getexpsd128_mask): Changed to ...
22385 __builtin_ia32_getexpsd128_round ... this.
22386 * config/i386/i386-builtin-types.def
22387 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
22388 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
22389 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
22390 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
22391 __builtin_ia32_getmantss_mask_round): New builtins.
22392 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
22393 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
22394 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
22395 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
22396 * config/i386/sse.md
22397 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
22398 avx512f_sgetexp<mode><mask_scalar_name>
22399 <round_saeonly_scalar_name> ... this.
22400 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
22401 %0, %1, %2<round_saeonly_op3>}): Changed to ...
22402 vgetexp<ssescalarmodesuffix>
22403 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22404 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
22405 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
22406 avx512f_vgetmant<mode><mask_scalar_name>
22407 <round_saeonly_scalar_name> ... this.
22408 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
22409 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
22410 vgetmant<ssescalarmodesuffix>
22411 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
22412 %0<mask_scalar_operand4>, %1, %2
22413 <round_saeonly_scalar_mask_op4>, %3} ... this.
22414 * config/i386/subst.md (mask_scalar_operand4,
22415 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
22416 round_saeonly_scalar_nimm_predicate): New subst attributes.
22418 2017-07-06 Julia Koval <julia.koval@intel.com>
22420 * config/i386/i386.c (ix86_erase_embedded_rounding):
22421 Remove code for old rounding pattern.
22423 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
22425 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
22427 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
22429 * doc/sourcebuild.texi (Test Directives, Variants of
22430 dg-require-support): Add documentation for dg-require-stack-check.
22432 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
22434 * config/i386/subst.md (mask_scalar, round_scalar,
22435 round_saeonly_scalar): New meta-templates.
22436 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
22437 round_scalar_mask_operand3, round_scalar_mask_op3,
22438 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
22439 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
22440 round_saeonly_scalar_constraint,
22441 round_saeonly_scalar_prefix): New subst attribute.
22442 * config/i386/sse.md
22443 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
22444 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
22445 <round_scalar_name> ... this.
22446 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
22447 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
22448 <round_scalar_name> ... this.
22449 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
22450 <sse>_vm<code><mode>3<mask_scalar_name>
22451 <round_saeonly_scalar_name> ... this.
22452 (v<plusminus_mnemonic><ssescalarmodesuffix>
22453 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
22454 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
22455 v<plusminus_mnemonic><ssescalarmodesuffix>
22456 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22457 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
22458 (v<multdiv_mnemonic><ssescalarmodesuffix>
22459 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
22460 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
22461 v<multdiv_mnemonic><ssescalarmodesuffix>
22462 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22463 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
22464 (v<maxmin_float><ssescalarmodesuffix>
22465 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
22466 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
22467 v<maxmin_float><ssescalarmodesuffix>
22468 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22469 %0<mask_scalar_operand3>, %1, %<iptr>2
22470 <round_saeonly_scalar_mask_op3>} ... this.
22472 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
22474 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
22475 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
22477 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
22478 Alan Hayward <alan.hayward@arm.com>
22479 David Sherwood <david.sherwood@arm.com>
22481 * combine.c (simplify_if_then_else): Remove "enum" before
22483 * compare-elim.c (can_eliminate_compare): Likewise.
22484 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
22486 (aarch64_lookup_simd_builtin_type): Likewise.
22487 (aarch64_simd_builtin_type): Likewise.
22488 (aarch64_init_simd_builtin_types): Likewise.
22489 (aarch64_simd_expand_args): Likewise.
22490 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
22492 (aarch64_reverse_mask): Likewise.
22493 (aarch64_simd_emit_reg_reg_move): Likewise.
22494 (aarch64_gen_adjusted_ldpstp): Likewise.
22495 (aarch64_ccmp_mode_to_code): Likewise.
22496 (aarch64_operands_ok_for_ldpstp): Likewise.
22497 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
22498 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
22500 (aarch64_min_divisions_for_recip_mul): Likewise.
22501 (aarch64_reassociation_width): Likewise.
22502 (aarch64_get_condition_code_1): Likewise.
22503 (aarch64_simd_emit_reg_reg_move): Likewise.
22504 (aarch64_simd_attr_length_rglist): Likewise.
22505 (aarch64_reverse_mask): Likewise.
22506 (aarch64_operands_ok_for_ldpstp): Likewise.
22507 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
22508 (aarch64_gen_adjusted_ldpstp): Likewise.
22509 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
22511 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
22512 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
22513 (arm_lookup_simd_builtin_type): Likewise.
22514 (arm_simd_builtin_type): Likewise.
22515 (arm_init_simd_builtin_types): Likewise.
22516 (arm_expand_builtin_args): Likewise.
22517 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
22518 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
22519 (ft32_setup_incoming_varargs): Likewise.
22520 (ft32_function_arg): Likewise.
22521 (ft32_function_arg_advance): Likewise.
22522 (ft32_pass_by_reference): Likewise.
22523 (ft32_arg_partial_bytes): Likewise.
22524 (ft32_valid_pointer_mode): Likewise.
22525 (ft32_addr_space_pointer_mode): Likewise.
22526 (ft32_addr_space_legitimate_address_p): Likewise.
22527 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
22529 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
22530 (ix86_emit_outlined_ms2sysv_restore): Likewise.
22531 (iamcu_alignment): Likewise.
22532 (canonicalize_vector_int_perm): Likewise.
22533 (ix86_noce_conversion_profitable_p): Likewise.
22534 (ix86_mpx_bound_mode): Likewise.
22535 (ix86_operands_ok_for_move_multiple): Likewise.
22536 * config/microblaze/microblaze-protos.h
22537 (microblaze_expand_conditional_branch_reg): Likewise.
22538 * config/microblaze/microblaze.c
22539 (microblaze_expand_conditional_branch_reg): Likewise.
22540 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
22542 (rs6000_reassociation_width): Likewise.
22543 (rs6000_invalid_binary_op): Likewise.
22544 (fusion_p9_p): Likewise.
22545 (emit_fusion_p9_load): Likewise.
22546 (emit_fusion_p9_store): Likewise.
22547 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
22549 (riscv_hard_regno_mode_ok_p): Likewise.
22550 (riscv_address_insns): Likewise.
22551 (riscv_split_symbol): Likewise.
22552 (riscv_legitimize_move): Likewise.
22553 (riscv_function_value): Likewise.
22554 (riscv_hard_regno_nregs): Likewise.
22555 (riscv_expand_builtin): Likewise.
22556 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
22557 (riscv_build_integer): Likewise.
22558 (riscv_split_integer): Likewise.
22559 (riscv_legitimate_constant_p): Likewise.
22560 (riscv_cannot_force_const_mem): Likewise.
22561 (riscv_regno_mode_ok_for_base_p): Likewise.
22562 (riscv_valid_base_register_p): Likewise.
22563 (riscv_valid_offset_p): Likewise.
22564 (riscv_valid_lo_sum_p): Likewise.
22565 (riscv_classify_address): Likewise.
22566 (riscv_legitimate_address_p): Likewise.
22567 (riscv_address_insns): Likewise.
22568 (riscv_load_store_insns): Likewise.
22569 (riscv_force_binary): Likewise.
22570 (riscv_split_symbol): Likewise.
22571 (riscv_force_address): Likewise.
22572 (riscv_legitimize_address): Likewise.
22573 (riscv_move_integer): Likewise.
22574 (riscv_legitimize_const_move): Likewise.
22575 (riscv_legitimize_move): Likewise.
22576 (riscv_address_cost): Likewise.
22577 (riscv_subword): Likewise.
22578 (riscv_output_move): Likewise.
22579 (riscv_canonicalize_int_order_test): Likewise.
22580 (riscv_emit_int_order_test): Likewise.
22581 (riscv_function_arg_boundary): Likewise.
22582 (riscv_pass_mode_in_fpr_p): Likewise.
22583 (riscv_pass_fpr_single): Likewise.
22584 (riscv_pass_fpr_pair): Likewise.
22585 (riscv_get_arg_info): Likewise.
22586 (riscv_function_arg): Likewise.
22587 (riscv_function_arg_advance): Likewise.
22588 (riscv_arg_partial_bytes): Likewise.
22589 (riscv_function_value): Likewise.
22590 (riscv_pass_by_reference): Likewise.
22591 (riscv_setup_incoming_varargs): Likewise.
22592 (riscv_print_operand): Likewise.
22593 (riscv_elf_select_rtx_section): Likewise.
22594 (riscv_save_restore_reg): Likewise.
22595 (riscv_for_each_saved_reg): Likewise.
22596 (riscv_register_move_cost): Likewise.
22597 (riscv_hard_regno_mode_ok_p): Likewise.
22598 (riscv_hard_regno_nregs): Likewise.
22599 (riscv_class_max_nregs): Likewise.
22600 (riscv_memory_move_cost): Likewise.
22601 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
22602 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
22603 (rl78_addr_space_address_mode): Likewise.
22604 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22606 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
22607 (rs6000_reassociation_width): Likewise.
22608 (rs6000_invalid_binary_op): Likewise.
22609 (fusion_p9_p): Likewise.
22610 (emit_fusion_p9_load): Likewise.
22611 (emit_fusion_p9_store): Likewise.
22612 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
22613 (ok_for_simple_move_operands): Likewise.
22614 (ok_for_simple_move_strict_operands): Likewise.
22615 (ok_for_simple_arith_logic_operands): Likewise.
22616 (visium_legitimize_reload_address): Likewise.
22617 (visium_select_cc_mode): Likewise.
22618 (output_cbranch): Likewise.
22619 (visium_split_double_move): Likewise.
22620 (visium_expand_copysign): Likewise.
22621 (visium_expand_int_cstore): Likewise.
22622 (visium_expand_fp_cstore): Likewise.
22623 * config/visium/visium.c (visium_pass_by_reference): Likewise.
22624 (visium_function_arg): Likewise.
22625 (visium_function_arg_advance): Likewise.
22626 (visium_libcall_value): Likewise.
22627 (visium_setup_incoming_varargs): Likewise.
22628 (visium_legitimate_constant_p): Likewise.
22629 (visium_legitimate_address_p): Likewise.
22630 (visium_legitimize_address): Likewise.
22631 (visium_secondary_reload): Likewise.
22632 (visium_register_move_cost): Likewise.
22633 (visium_memory_move_cost): Likewise.
22634 (prepare_move_operands): Likewise.
22635 (ok_for_simple_move_operands): Likewise.
22636 (ok_for_simple_move_strict_operands): Likewise.
22637 (ok_for_simple_arith_logic_operands): Likewise.
22638 (visium_function_value_1): Likewise.
22639 (rtx_ok_for_offset_p): Likewise.
22640 (visium_legitimize_reload_address): Likewise.
22641 (visium_split_double_move): Likewise.
22642 (visium_expand_copysign): Likewise.
22643 (visium_expand_int_cstore): Likewise.
22644 (visium_expand_fp_cstore): Likewise.
22645 (visium_split_cstore): Likewise.
22646 (visium_select_cc_mode): Likewise.
22647 (visium_split_cbranch): Likewise.
22648 (output_cbranch): Likewise.
22649 (visium_print_operand_address): Likewise.
22650 * expmed.c (flip_storage_order): Likewise.
22651 * expmed.h (emit_cstore): Likewise.
22652 (flip_storage_order): Likewise.
22653 * genrecog.c (validate_pattern): Likewise.
22654 * hsa-gen.c (gen_hsa_addr): Likewise.
22655 * internal-fn.c (expand_arith_overflow): Likewise.
22656 * ira-color.c (allocno_copy_cost_saving): Likewise.
22657 * lra-assigns.c (find_hard_regno_for_1): Likewise.
22658 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
22659 (process_invariant_for_inheritance): Likewise.
22660 * lra-eliminations.c (move_plus_up): Likewise.
22661 * omp-low.c (lower_oacc_reductions): Likewise.
22662 * simplify-rtx.c (simplify_subreg): Likewise.
22663 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
22664 (TARGET_CHKP_BOUND_MODE): Likewise..
22665 * targhooks.c (default_chkp_bound_mode): Likewise.
22666 (default_setup_incoming_vararg_bounds): Likewise.
22667 * targhooks.h (default_chkp_bound_mode): Likewise.
22668 (default_setup_incoming_vararg_bounds): Likewise.
22669 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
22670 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
22671 (have_whole_vector_shift): Likewise.
22672 * tree-vect-stmts.c (vectorizable_load): Likewise.
22673 * doc/tm.texi: Regenerate.
22675 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22677 Graceful degrade if Binutils PR21472 is not available.
22680 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
22681 .rodata in flash test fails.
22682 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
22683 * confgure: Regenerate.
22684 * config.in: Regenerate.
22685 * config/avr/avr.c (avr_asm_named_section)
22686 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
22687 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
22688 (avr_asm_init_sections): Same.
22690 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22692 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
22693 (fma<VH:mode>4_intrinsic): Likewise.
22694 (*fmsub<VCVTF:mode>4): Likewise.
22695 (*fmsub<VH:mode>4_intrinsic): Likewise.
22697 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22700 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
22701 Don't depend on "optimize > 0".
22702 (out_movhi_r_mr, out_movqi_mr_r): Same.
22703 (out_movhi_mr_r, out_movqi_r_mr): Same.
22704 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
22705 io_address_operand on "optimize > 0".
22707 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22709 * tree-loop-distribution.c: Add general explanantion on the pass.
22710 (generate_loops_for_partition): Mark distributed loop.
22711 (pg_add_dependence_edges): New parameter. Handle alias data
22712 dependence specially and record it in the parameter if asked.
22713 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
22714 (init_partition_graph_vertices, add_partition_graph_edge): New.
22715 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
22716 (free_partition_graph_vdata, build_partition_graph): New.
22717 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
22718 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
22719 (data_ref_segment_size, latch_dominated_by_data_ref): New.
22720 (compute_alias_check_pairs, version_loop_by_alias_check): New.
22721 (version_for_distribution_p, finalize_partitions): New.
22722 (distribute_loop): Handle alias data dependence specially. Factor
22723 out loop fusion code as functions and call these functions.
22725 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22727 * tree-loop-distribution.c (classify_partition): New parameter and
22728 better handle reduction statement.
22729 (rdg_build_partitions): Revise comment.
22730 (distribute_loop): Compute statements in all partitions and pass it
22731 to classify_partition.
22733 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22735 * tree-loop-distribution.c (enum partition_type): New.
22736 (struct partition): New field type.
22737 (partition_merge_into): Add parameter. Update partition type.
22738 (data_dep_in_cycle_p, update_type_for_merge): New functions.
22739 (build_rdg_partition_for_vertex): Compute partition type.
22740 (rdg_build_partitions): Dump partition type.
22741 (distribute_loop): Update calls to partition_merge_into.
22743 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22745 * tree-loop-distribution.c (struct ddr_hasher): New.
22746 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
22748 (classify_partition): Call get_data_dependence.
22749 (pg_add_dependence_edges): Ditto.
22750 (distribute_loop): Release data dependence hash table.
22752 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22754 * tree-loop-distribution.c (ref_base_address): Delete.
22755 (similar_memory_accesses): Rename ...
22756 (share_memory_accesses): ... to this. Check if partitions access
22757 the same memory reference.
22758 (distribute_loop): Call share_memory_accesses.
22760 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22762 * tree-loop-distribution.c (struct partition): New field recording
22763 its data reference.
22764 (partition_alloc, partition_free): Init and release data refs.
22765 (partition_merge_into): Merge data refs.
22766 (build_rdg_partition_for_vertex): Collect data refs for partition.
22767 (pg_add_dependence_edges): Change parameters from vector to bitmap.
22769 (distribute_loop): Remve data refs from vertice data of partition
22772 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22774 * tree-loop-distribution.c (params.h): Include header file.
22775 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
22776 (datarefs_vec): New global var.
22777 (create_rdg_vertices): Use datarefs_vec directly.
22778 (free_rdg): Don't free data references.
22779 (build_rdg): Update use. Don't free data references.
22780 (distribute_loop): Compute global variable for data references.
22781 Bail out if there are too many data references.
22783 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22785 * tree-loop-distribution.c (loop_nest): New global var.
22786 (build_rdg): Use loop directly, rather than loop nest.
22787 (pg_add_dependence_edges): Remove loop nest parameter. Use global
22789 (distribute_loop): Compute global variable loop nest. Update use.
22791 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22793 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
22794 (partition_merge_into): New parameter. Dump reason for fusion.
22795 (distribute_loop): Update use of partition_merge_into.
22797 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22799 * tree-loop-distribution.c (bb_top_order_index): New.
22800 (bb_top_order_index_size, bb_top_order_cmp): New.
22801 (stmts_from_loop): Use topological order.
22802 (pass_loop_distribution::execute): Compute and release topological
22803 order for basic blocks.
22805 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22807 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
22810 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22812 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
22813 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
22814 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
22815 * internal-fn.def (LOOP_DIST_ALIAS): New.
22816 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
22817 (fold_loop_internal_call): ... this.
22818 (vect_loop_dist_alias_call): New function.
22819 (set_uid_loop_bbs): Call fold_loop_internal_call.
22820 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
22823 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22826 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
22827 Require dead FLAGS_REG at the beginning of a peephole.
22829 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22832 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
22833 arguments in the call to __builtin_ia32_sbb_u32.
22834 (_subborrow_u64): Swap _X and _Y arguments in the call to
22835 __builtin_ia32_sbb_u64.
22837 2017-07-04 Jakub Jelinek <jakub@redhat.com>
22840 * tree-vrp.c (compare_assert_loc): Turn into a function template
22841 with stable template parameter. Only test if a->e is NULL,
22842 !a->e == !b->e has been verified already. Use e == NULL or
22843 e != NULL instead of e or ! e tests. If stable is true, don't use
22844 iterative_hash_expr, on the other side allow a or b or both NULL
22845 and sort the NULLs last.
22846 (process_assert_insertions): Sort using compare_assert_loc<false>
22847 instead of compare_assert_loc, later sort using
22848 compare_assert_loc<true> before calling process_assert_insertions_for
22849 in a loop. Use break instead of continue once seen NULL pointer.
22851 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22853 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
22854 Cortex-R7 and Cortex-R8 processors.
22856 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22858 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
22859 uninitialized while src is not.
22861 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
22863 * common/config/arm/arm-common.c: Adjust include path for
22865 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
22866 (arm-cpu.h): Create in build directory. Adjust dependency rules.
22867 (arm-cpu-data.h): Likewise.
22868 (arm-cpu-cdata.h): Likewise.
22869 * config/arm/arm-cpu.h: Delete.
22870 * config/arm/arm-cpu-cdata.h: Delete.
22871 * config/arm/arm-cpu-data.h: Delete.
22873 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
22875 * config/arm/arm-cpus.in (cortex-a55): New.
22876 (cortex-a75): Likewise.
22877 (cortex-a75.cortex-a55): Likewise.
22878 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
22880 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
22881 * config/arm/arm-cpu-cdata.h: Regenerate.
22882 * config/arm/arm-cpu-data.h: Regenerate.
22883 * config/arm/arm-cpu.h: Regenerate.
22884 * config/arm/arm-tables.opt: Regenerate.
22885 * config/arm/arm-tune.md: Regenerate.
22887 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22889 * haifa-sched.c (sched_create_recovery_edges): Update profile.
22891 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22893 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
22896 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
22898 PR tree-optimization/81292
22899 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
22900 full_string_p, also call adjust_related_strinfos if the adjustment
22901 is simple, otherwise invalidate related strinfos.
22903 2017-07-04 Martin Liska <mliska@suse.cz>
22906 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
22907 newly created variable as DECL_IGNORED_P.
22909 2017-07-04 Martin Liska <mliska@suse.cz>
22912 * ipa-inline.c (inline_small_functions):
22913 Use xstrdup_for_dump.
22915 2017-07-04 Tom de Vries <tom@codesourcery.com>
22917 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
22919 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22922 * config/darwin.c (darwin_function_switched_text_sections):
22925 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
22927 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
22929 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22931 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
22933 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22935 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
22936 min_profitable_iters, and th as inclusive lower bounds.
22937 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
22938 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
22939 for min_profitable_iters and min_profitable_estimate.
22940 (vect_transform_loop): Treat th as an inclusive lower bound.
22941 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22943 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22946 * config/darwin.c (darwin_function_switched_text_sections):
22947 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
22948 in two pieces, and suppress the use of buf.
22950 2017-07-03 Nathan Sidwell <nathan@acm.org>
22952 * hash-table.h (hash_table_mod1): Fix indentation.
22954 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22956 PR middle-end/81290
22957 * predict.c (force_edge_cold): Be more careful about propagation
22959 * profile-count.h (profile_probability::guessed,
22960 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
22962 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
22964 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
22966 * doc/invoke.texi (rcpc architecture extension): Document it.
22968 2017-07-03 Richard Biener <rguenther@suse.de>
22970 PR tree-optimization/60510
22971 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
22972 the scalar reduction PHI and use it.
22973 (vectorizable_reduction): Properly guard the single_defuse_cycle
22974 path for non-SLP reduction chains where we cannot use it.
22975 Rework reduc_def/index and vector type deduction. Rework
22976 vector operand gathering during reduction op code-gen.
22977 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
22978 chains dissolve the chain and leave it to non-SLP reduction
22981 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22983 * tree-data-ref.h (dr_alignment): Declare.
22984 * tree-data-ref.c (dr_alignment): New function.
22985 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
22986 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
22988 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
22990 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22992 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
22993 and base_misalignment fields.
22994 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
22995 * tree-data-ref.c: Include builtins.h.
22996 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
22997 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
22998 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
22999 * tree-vect-data-refs.c: Include tree-cfg.h.
23000 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
23001 fields instead of calculating an alignment here.
23002 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
23003 innermost_loop_behavior fields.
23005 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23007 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
23009 (DR_STEP_ALIGNMENT): New macro.
23010 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
23011 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
23012 (create_data_ref): Print it.
23013 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
23014 to tell whether the step preserves vector (mis)alignment.
23015 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
23016 Move the check for an integer step and generalise to all INTEGER_CST.
23017 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
23018 Print the outer step alignment.
23020 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23022 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
23023 with offset_alignment.
23024 (DR_ALIGNED_TO): Delete.
23025 (DR_OFFSET_ALIGNMENT): New macro.
23026 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
23027 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
23028 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
23029 (create_data_ref): Likewise.
23030 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
23031 (vect_analyze_data_refs): Likewise.
23032 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
23033 creating dummy innermost behavior.
23035 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23037 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
23038 with a "innermost_loop_behavior *" and refeence tree.
23039 * tree-data-ref.c (dr_analyze_innermost): Likewise.
23040 (create_data_ref): Update call accordingly.
23041 * tree-predcom.c (find_looparound_phi): Likewise.
23043 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23045 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
23046 fields with dr_wrt_vec_loop.
23047 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
23048 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
23049 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
23050 (vect_dr_behavior): New function.
23051 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
23052 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
23053 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
23054 track whether the step preserves the misalignment.
23055 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
23056 Use vect_dr_behavior.
23057 (vect_setup_realignment): Update call accordingly.
23058 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
23059 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
23060 call to vect_create_addr_base_for_vector_ref.
23061 (vect_create_cond_for_align_checks): Likewise.
23062 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
23063 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
23064 (vect_recog_mask_conversion_pattern): Likewise.
23065 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
23066 (new_stmt_vec_info): Remove redundant zeroing.
23068 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
23070 * common/config/arm/arm-common.c (arm_be8_option): New function.
23071 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
23072 (ISA_ARMv6): Add isa_bit_be8.
23073 * config/arm/arm.h (arm_be8_option): Add prototype.
23074 (BE8_SPEC_FUNCTION): New define.
23075 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
23076 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
23077 (mlittle-endian): Similarly.
23078 (mbe8, mbe32): New options.
23079 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
23080 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
23082 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23084 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
23086 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23088 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
23089 (cleanup_tree_cfg_bb): Use it.
23090 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
23092 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
23094 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23097 * loop-doloop.c (add_test): Update profile.
23099 2017-07-03 Martin Liska <mliska@suse.cz>
23102 * sanopt.c (rewrite_usage_of_param): New function.
23103 (sanitize_rewrite_addressable_params): Likewise.
23104 (pass_sanopt::execute): Call rewrite_usage_of_param.
23106 2017-07-03 Richard Biener <rguenther@suse.de>
23108 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
23109 back to using VIEW_CONVERT_EXPR.
23111 2017-07-03 Martin Liska <mliska@suse.cz>
23114 * doc/extend.texi: Document when a resolver function is
23115 generated for target_clones.
23117 2017-07-03 Martin Liska <mliska@suse.cz>
23119 * asan.c (asan_emit_stack_protection): Unpoison just red zones
23120 and shadow memory of auto variables which are subject of
23121 use-after-scope sanitization.
23122 (asan_expand_mark_ifn): Add do set only when is_poison.
23124 2016-07-03 Richard Biener <rguenther@suse.de>
23126 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
23128 (vect_force_simple_reduction): Record reduction def -> phi mapping.
23129 (vectorizable_reduction): Perform reduction PHI creation when
23130 visiting a reduction PHI and adjust and simplify code generation
23131 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
23132 (vect_transform_loop): Visit reduction PHIs.
23133 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
23134 defs into the SLP tree.
23135 (vect_build_slp_tree): Reduction defs terminate the recursion.
23136 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
23138 (vect_get_vec_defs_for_stmt_copy): Export.
23139 (vect_get_vec_defs): Likewise.
23140 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
23142 (vect_get_vec_defs_for_stmt_copy): Declare.
23143 (vect_get_vec_defs): Likewise.
23145 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23147 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
23148 parameter with a "loop" parameter and use it instead of the
23149 loop containing DR_STMT. Don't check simple_iv when doing
23150 BB analysis. Describe the two analysis modes in the comment.
23152 2017-07-03 Tom de Vries <tom@codesourcery.com>
23154 PR tree-optimization/69468
23155 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
23156 (find_same_succ_bb): Handle ignore_edge_flags.
23158 2017-07-03 Tom de Vries <tom@codesourcery.com>
23160 PR tree-optimization/81192
23161 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
23163 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
23165 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
23167 2017-07-03 Tom de Vries <tom@codesourcery.com>
23169 PR tree-optimization/81192
23170 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
23171 BB_SAME_SUCC (bb) == NULL.
23173 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23175 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
23178 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23180 * dumpfile.c: Include profile-count.h
23181 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
23183 (insert_cond_bb): Update profile.
23184 * tree-cfg.h (insert_cond_bb): Update prototype.
23185 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
23186 * tree-dump.c: Do not include tree-cfg.
23188 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23190 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
23192 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23194 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
23197 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23199 * tree-complex.c (expand_complex_div_wide): update profile.
23201 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23202 Alan Hayward <alan.hayward@arm.com>
23203 David Sherwood <david.sherwood@arm.com>
23205 * Makefile.in (MACHMODE_H): Remove insn-modes.h
23206 (CORETYPES_H): New define.
23207 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
23208 (insn-modes-inline.h, s-modes-inline-h): New rules.
23209 (generated_files): Add insn-modes-inline.h.
23210 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
23211 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
23212 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
23213 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
23214 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
23215 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
23216 (build/gencodes.o, build/genconditions.o): Likewise.
23217 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
23218 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
23219 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
23220 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
23221 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
23222 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
23223 * coretypes.h: Include everything up to real.h for generators.
23224 Include insn-modes.h first. Include wide-int-print.h after
23225 wide-int.h. Include insn-modes-inline.h and then machmode.h.
23226 * machmode.h: Don't include insn-modes.h here.
23227 * function-tests.c: Remove includes of signop.h, machmode.h,
23228 double-int.h and wide-int.h.
23230 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
23232 * optc-save-gen.awk: Likewise.
23233 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
23234 * godump.c: Remove include of wide-int-print.h.
23235 * pretty-print.h: Likewise.
23236 * wide-int-print.cc: Likewise.
23237 * wide-int.cc: Likewise.
23238 * hash-map-tests.c: Remove include of signop.h.
23239 * hash-set-tests.c: Likewise.
23240 * rtl-tests.c: Likewise.
23241 * mkconfig.sh: Remove include of machmode.h.
23242 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
23244 (emit_insn_modes_inline_h): ...this new function. Emit the code
23245 into an insn-modes-inline.h header file, adding appropriate
23246 include guards and end comments.
23247 (emit_insn_modes_c_header): Remove include of machmode.h.
23248 (emit_min_insn_modes_c_header): Include coretypes.h rather than
23250 (main): Handle -i flag and call emit_insn_modes_inline_h when
23253 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23255 * tree-ssa-strlen.c (strinfo): Rename the length field to
23256 nonzero_chars. Add a full_string_p field.
23257 (compare_nonzero_chars, zero_length_string_p): New functions.
23258 (get_addr_stridx): Add an offset_out parameter.
23259 Use compare_nonzero_chars.
23260 (get_stridx): Update accordingly. Use compare_nonzero_chars.
23261 (new_strinfo): Update after above changes to strinfo.
23262 (set_endptr_and_length): Set full_string_p.
23263 (get_string_length): Update after above changes to strinfo.
23264 (unshare_strinfo): Update call to new_strinfo.
23265 (maybe_invalidate): Likewise.
23266 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
23267 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
23268 as a uhwi instead of an shwi. Update after above changes to
23269 strinfo and new_strinfo.
23270 (zero_length_string): Assert that chainsi contains full strings.
23271 Use zero_length_string_p. Update call to new_strinfo.
23272 (adjust_related_strinfos): Update after above changes to strinfo.
23273 Copy full_string_p from origsi.
23274 (adjust_last_stmt): Use zero_length_string_p.
23275 (handle_builtin_strlen): Update after above changes to strinfo and
23276 new_strinfo. Install the lhs as the string length if the previous
23277 entry didn't describe a full string.
23278 (handle_builtin_strchr): Update after above changes to strinfo
23280 (handle_builtin_strcpy): Likewise.
23281 (handle_builtin_strcat): Likewise.
23282 (handle_builtin_malloc): Likewise.
23283 (handle_pointer_plus): Likewise.
23284 (handle_builtin_memcpy): Likewise. Track nonzero characters
23285 that aren't necessarily followed by a nul terminator.
23286 (handle_char_store): Likewise.
23288 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23290 PR tree-optimization/80769
23291 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
23292 for malloc and calloc. Document the new invariant that all related
23293 strinfos have delayed lengths or none do.
23294 (verify_related_strinfos): Move earlier in file.
23295 (set_endptr_and_length): New function, split out from...
23296 (get_string_length): ...here. Also set the lengths of related
23298 (zero_length_string): Assert that chainsi has known (rather than
23300 (adjust_related_strinfos): Likewise.
23302 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23304 PR tree-optimization/81136
23305 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
23306 assert that two references with the same misalignment have the same
23307 compile-time misalignment if those compile-time misalignments
23310 2017-07-01 Andi Kleen <ak@linux.intel.com>
23312 * print-tree.c (print_node): Print all attributes.
23314 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23316 * cfg.c (scale_bbs_frequencies): New function.
23317 * cfg.h (scale_bbs_frequencies): Declare it.
23318 * cfgloopanal.c (single_likely_exit): Cleanup.
23319 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
23321 (scale_loop_profile): Likewise.
23322 (loop_version): Likewise.
23323 (create_empty_loop_on_edge): Update.
23324 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
23325 scale_loop_frequencies, scale_loop_profile, loopify,
23326 loop_version): Update prototypes.
23327 * modulo-sched.c (sms_schedule): Update.
23328 * predict.c (unlikely_executed_edge_p): Also check probability.
23329 (probably_never_executed_edge_p): Fix typo.
23330 * tree-if-conv.c (version_loop_for_if_conversion): Update.
23331 * tree-parloops.c (gen_parallel_loop): Update.
23332 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
23333 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23334 * tree-ssa-loop-split.c (split_loop): Update.
23335 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
23336 * tree-vect-loop-manip.c (vect_do_peeling): Update.
23337 (vect_loop_versioning): Update.
23338 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23340 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23342 * trans-mem.c (split_bb_make_tm_edge): Update profile.
23344 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23346 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
23347 to keep profile consistent.
23349 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23351 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
23352 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
23353 * profile-count.h (max_safe_multiplier): Make unsigned.
23354 (profile_count::guessed_zero): New.
23356 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23358 * bb-reorder.c (fix_up_crossing_landing_pad,
23359 fix_crossing_conditional_branches): Use make_single_succ_edge
23360 to keep profile consistent.
23362 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23364 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
23367 2017-07-01 Jakub Jelinek <jakub@redhat.com>
23370 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
23371 the right scopes, make sure cond_jump isn't preserved between multiple
23372 iterations. Search for fallthru edge whenever there are 3+ edges and
23373 use find_fallthru_edge for it.
23375 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23377 Patch by Alexander Monakov <amonakov@ispras.ru>
23378 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
23379 probabilities consistently.
23381 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23383 * pa.c (pa_expand_compare_and_swap_loop): Update call of
23384 emit_cmp_and_jump_insns.
23386 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23389 * tree-inline.c (expand_call_inline): Combine profile statuses.
23391 2017-06-30 Andrew Pinski <apinski@cavium.com>
23393 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
23394 fold_stmt returned true.
23396 2017-06-30 Nathan Sidwell <nathan@acm.org>
23398 * ggc.h (empty_string): Delete.
23399 * cfgexpand.c (expand_asm_stmt): Use plain "".
23400 * optabs.c (expand_asm_memory_barrier): Likewise.
23401 * stringpool.c (empty_string): Delete.
23402 (digit_vector, digit_string): Delete.
23403 (ggc_alloc_string): Use plain "", don't optimize single digit
23404 strings. Use ggc_alloc_atomic.
23406 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
23408 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
23409 comparison set and one other set, use the cost of the non-comparison
23412 2017-06-30 Nathan Sidwell <nathan@acm.org>
23414 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
23417 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
23419 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
23420 loops. Remove now unneeded calls to gimple_switch_set_label() that
23421 just set removed labels to NULL_TREE.
23423 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
23425 * tree-ssanames.c (set_range_info_raw): Abstract from ...
23426 (set_range_info): ...here. Only call set_range_info_raw if domain
23428 (set_nonzero_bits): Call set_range_info_raw.
23429 * tree-ssanames.h (set_range_info_raw): New.
23431 2017-06-30 Jakub Jelinek <jakub@redhat.com>
23434 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
23435 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
23436 of nonimmediate_operand and <store_mask_constraint> instead of m
23437 for the input operand. For V8FI iterator, always split if input
23438 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
23439 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
23440 <store_mask_predicate> instead of register_operand and
23441 <store_mask_constraint> instead of v for the input operand. Make
23442 sure both operands aren't MEMs for if not <mask_applied>.
23444 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
23446 * lto-wrapper.c (copy_file) Close both file descriptors before
23449 2017-06-30 Martin Liska <mliska@suse.cz>
23452 * multiple_target.c (create_dispatcher_calls): Make ifunc
23453 also for function that don't have calls or are not referenced.
23455 2017-06-30 Richard Biener <rguenther@suse.de>
23457 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
23458 analyze the first scalar stmt. Move vector type computation
23459 for the BB case here from ...
23460 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
23461 live operation processing in the SLP case properly.
23463 2017-06-30 Richard Biener <rguenther@suse.de>
23465 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
23467 2017-06-30 Martin Liska <mliska@suse.cz>
23470 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
23471 before BUILT_IN_UNWIND_RESUME when ASAN is used.
23473 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
23475 * doc/invoke.texi (AArch64): Add missing options and remove redundant
23478 2017-06-30 Richard Biener <rguenther@suse.de>
23480 PR tree-optimization/81249
23481 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
23482 condition reduction result to original scalar type.
23484 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23486 * profile-count.h (enum profile_quality): Fix typos and whitespace
23489 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23491 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
23492 type for branch probabilities.
23494 2017-06-29 Julian Brown <julian@codesourcery.com>
23495 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23497 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
23498 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
23499 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
23500 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
23502 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23504 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
23505 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
23506 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
23507 CC usage from generic code to here.
23508 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
23509 CC usage into the target macros.
23511 2017-06-29 Maya Rashish <coypu@sdf.org>
23513 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
23516 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23518 * arm/arm-builtins.c: Include profile-count.h
23519 * except.c (sjlj_emit_function_enter): Use
23520 profile_probability::unlikely.
23522 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23524 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
23525 and tocrel_offset be pointer args rather than implicitly using
23527 (legitimate_constant_pool_address_p, rs6000_emit_move,
23528 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
23529 tocrel_offset and use in toc_relative_expr_p call.
23530 (print_operand, print_operand_address): Use static tocrel_base_oac
23531 and tocrel_offset_oac.
23532 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
23535 2017-06-29 Maya Rashish <coypu@sdf.org>
23537 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
23539 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
23541 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
23542 objects, take into account only the alignment of 'op0' and 'mode1' if
23545 2017-06-29 Steve Ellcey <sellcey@cavium.com>
23547 * ccmp.c (ccmp_tree_comparison_p): New function.
23548 (ccmp_candidate_p): Update to use above function.
23549 (get_compare_parts): New function.
23550 (expand_ccmp_next): Update to use new functions.
23551 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
23553 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
23554 take mode as argument.
23555 * ccmp.h (expand_ccmp_expr): Add mode as argument.
23556 * expr.c (expand_expr_real_1): Pass mode as argument.
23558 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
23560 * combine.c (combine_instructions): Print insns to dump_file, together
23563 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23565 * asan.c (asan_emit_stack_protection): Update.
23566 (create_cond_insert_point): Update.
23567 * auto-profile.c (afdo_propagate_circuit): Update.
23568 * basic-block.h (struct edge_def): Turn probability to
23569 profile_probability.
23570 (EDGE_FREQUENCY): Update.
23571 * bb-reorder.c (find_traces_1_round): Update.
23572 (better_edge_p): Update.
23573 (sanitize_hot_paths): Update.
23574 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
23575 (make_single_succ_edge): Update.
23576 (check_bb_profile): Update.
23577 (dump_edge_info): Update.
23578 (update_bb_profile_for_threading): Update.
23579 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
23580 probabilitycount to 0.
23581 * cfgbuild.c (compute_outgoing_frequencies): Update.
23582 * cfgcleanup.c (try_forward_edges): Update.
23583 (outgoing_edges_match): Update.
23584 (try_crossjump_to_edge): Update.
23585 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
23586 (expand_gimple_tailcall): Update.
23587 (construct_init_block): Use make_single_succ_edge.
23588 (construct_exit_block): Use make_single_succ_edge.
23589 * cfghooks.c (verify_flow_info): Update.
23590 (redirect_edge_succ_nodup): Update.
23591 (split_edge): Update.
23592 (account_profile_record): Update.
23593 * cfgloopanal.c (single_likely_exit): Update.
23594 * cfgloopmanip.c (scale_loop_profile): Update.
23595 (set_zero_probability): Remove.
23596 (duplicate_loop_to_header_edge): Update.
23597 * cfgloopmanip.h (loop_version): Update prototype.
23598 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
23599 (force_nonfallthru_and_redirect): Update.
23600 (update_br_prob_note): Update.
23601 (rtl_verify_edges): Update.
23602 (purge_dead_edges): Update.
23603 (rtl_lv_add_condition_to_bb): Update.
23604 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
23605 * cgraphunit.c (init_lowered_empty_function): Update.
23606 (cgraph_node::expand_thunk): Update.
23607 * cilk-common.c: Include profile-count.h
23608 * dojump.c (inv): Remove.
23609 (jumpifnot): Update.
23610 (jumpifnot_1): Update.
23611 (do_jump_1): Update.
23613 (do_jump_by_parts_greater_rtx): Update.
23614 (do_compare_rtx_and_jump): Update.
23615 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
23616 do_jump_1. do_compare_rtx_and_jump): Update prototype.
23617 * dwarf2cfi.c: Include profile-count.h
23618 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
23619 (sjlj_emit_dispatch_table): Likewise.
23620 * explow.c: Include profile-count.h
23621 * expmed.c (emit_store_flag_force): Update.
23622 (do_cmp_and_jump): Update.
23623 * expr.c (compare_by_pieces_d::generate): Update.
23624 (compare_by_pieces_d::finish_mode): Update.
23625 (emit_block_move_via_loop): Update.
23626 (store_expr_with_bounds): Update.
23627 (store_constructor): Update.
23628 (expand_expr_real_2): Update.
23629 (expand_expr_real_1): Update.
23630 * expr.h (try_casesi, try_tablejump): Update prototypes.
23631 * gimple-pretty-print.c (dump_probability): Update.
23632 (dump_profile): New.
23633 (dump_gimple_label): Update.
23634 (dump_gimple_bb_header): Update.
23635 * graph.c (draw_cfg_node_succ_edges): Update.
23636 * hsa-gen.c (convert_switch_statements): Update.
23637 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
23638 (find_if_case_1): Update.
23639 (find_if_case_2): Update.
23640 * internal-fn.c (expand_arith_overflow_result_store): Update.
23641 (expand_addsub_overflow): Update.
23642 (expand_neg_overflow): Update.
23643 (expand_mul_overflow): Update.
23644 (expand_vector_ubsan_overflow): Update.
23645 * ipa-cp.c (good_cloning_opportunity_p): Update.
23646 * ipa-split.c (split_function): Use make_single_succ_edge.
23647 * ipa-utils.c (ipa_merge_profiles): Update.
23648 * loop-doloop.c (add_test): Update.
23649 (doloop_modify): Update.
23650 * loop-unroll.c (compare_and_jump_seq): Update.
23651 (unroll_loop_runtime_iterations): Update.
23652 * lra-constraints.c (lra_inheritance): Update.
23653 * lto-streamer-in.c (input_cfg): Update.
23654 * lto-streamer-out.c (output_cfg): Update.
23655 * mcf.c (adjust_cfg_counts): Update.
23656 * modulo-sched.c (sms_schedule): Update.
23657 * omp-expand.c (expand_omp_for_init_counts): Update.
23658 (extract_omp_for_update_vars): Update.
23659 (expand_omp_ordered_sink): Update.
23660 (expand_omp_for_ordered_loops): Update.
23661 (expand_omp_for_generic): Update.
23662 (expand_omp_for_static_nochunk): Update.
23663 (expand_omp_for_static_chunk): Update.
23664 (expand_cilk_for): Update.
23665 (expand_omp_simd): Update.
23666 (expand_omp_taskloop_for_outer): Update.
23667 (expand_omp_taskloop_for_inner): Update.
23668 * omp-simd-clone.c (simd_clone_adjust): Update.
23669 * optabs.c (expand_doubleword_shift): Update.
23670 (expand_abs): Update.
23671 (emit_cmp_and_jump_insn_1): Update.
23672 (expand_compare_and_swap_loop): Update.
23673 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
23674 * predict.c (predictable_edge_p): Update.
23675 (edge_probability_reliable_p): Update.
23676 (set_even_probabilities): Update.
23677 (combine_predictions_for_insn): Update.
23678 (combine_predictions_for_bb): Update.
23679 (propagate_freq): Update.
23680 (estimate_bb_frequencies): Update.
23681 (force_edge_cold): Update.
23682 * profile-count.c (profile_count::dump): Add missing space into dump.
23683 (profile_count::debug): Add newline.
23684 (profile_count::differs_from_p): Explicitly convert to unsigned.
23685 (profile_count::stream_in): Update.
23686 (profile_probability::dump): New member function.
23687 (profile_probability::debug): New member function.
23688 (profile_probability::differs_from_p): New member function.
23689 (profile_probability::differs_lot_from_p): New member function.
23690 (profile_probability::stream_in): New member function.
23691 (profile_probability::stream_out): New member function.
23692 * profile-count.h (profile_count_quality): Rename to ...
23693 (profile_quality): ... this one.
23694 (profile_probability): New.
23695 (profile_count): Update.
23696 * profile.c (compute_branch_probabilities): Update.
23697 * recog.c (peep2_attempt): Update.
23698 * sched-ebb.c (schedule_ebbs): Update.
23699 * sched-rgn.c (find_single_block_region): Update.
23700 (compute_dom_prob_ps): Update.
23701 (schedule_region): Update.
23702 * sel-sched-ir.c (compute_succs_info): Update.
23703 * stmt.c (struct case_node): Update.
23704 (do_jump_if_equal): Update.
23705 (get_outgoing_edge_probs): Update.
23706 (conditional_probability): Update.
23707 (emit_case_dispatch_table): Update.
23708 (expand_case): Update.
23709 (expand_sjlj_dispatch_table): Update.
23710 (emit_case_nodes): Update.
23711 * targhooks.c: Update.
23712 * tracer.c (better_p): Update.
23713 (find_best_successor): Update.
23714 * trans-mem.c (expand_transaction): Update.
23715 * tree-call-cdce.c: Update.
23716 * tree-cfg.c (gimple_split_edge): Upate.
23717 (move_sese_region_to_fn): Upate.
23718 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
23719 * tree-eh.c (lower_resx): Upate.
23720 (cleanup_empty_eh_move_lp): Upate.
23721 * tree-if-conv.c (version_loop_for_if_conversion): Update.
23722 * tree-inline.c (copy_edges_for_bb): Update.
23723 (copy_cfg_body): Update.
23724 * tree-parloops.c (gen_parallel_loop): Update.
23725 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
23726 (gimple_gen_time_profiler): Update.
23727 * tree-ssa-dce.c (remove_dead_stmt): Update.
23728 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23729 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
23730 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
23731 (unloop_loops): Update.
23732 (try_peel_loop): Update.
23733 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23734 * tree-ssa-loop-split.c (connect_loops): Update.
23735 (split_loop): Update.
23736 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
23737 (hoist_guard): Update.
23738 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
23739 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
23740 (value_replacement): Update.
23741 * tree-ssa-reassoc.c (branch_fixup): Update.
23742 * tree-ssa-tail-merge.c (replace_block_by): Update.
23743 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
23744 (create_edge_and_update_destination_phis): Update.
23745 (compute_path_counts): Update.
23746 (recompute_probabilities): Update.
23747 (update_joiner_offpath_counts): Update.
23748 (freqs_to_counts_path): Update.
23749 (duplicate_thread_path): Update.
23750 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
23751 (struct switch_conv_info): Update.
23752 (gen_inbound_check): Update.
23753 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23754 (vect_do_peeling): Update.
23755 (vect_loop_versioning): Update.
23756 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23757 (optimize_mask_stores): Update.
23758 * ubsan.c (ubsan_expand_null_ifn): Update.
23759 * value-prof.c (gimple_divmod_fixed_value): Update.
23760 (gimple_divmod_fixed_value_transform): Update.
23761 (gimple_mod_pow2): Update.
23762 (gimple_mod_pow2_value_transform): Update.
23763 (gimple_mod_subtract): Update.
23764 (gimple_mod_subtract_transform): Update.
23765 (gimple_ic): Update.
23766 (gimple_stringop_fixed_value): Update.
23767 (gimple_stringops_transform): Update.
23768 * value-prof.h: Update.
23770 2017-06-29 Carl Love <cel@us.ibm.com>
23772 * config/rs6000/rs6000-c.c: Add support for built-in functions
23773 vector signed int vec_signed (vector float);
23774 vector signed long long vec_signed (vector double);
23775 vector signed int vec_signed2 (vector double, vector double);
23776 vector signed int vec_signede (vector double);
23777 vector signed int vec_signedo (vector double);
23778 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
23779 instruction generator.
23780 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23781 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
23782 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
23784 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
23785 vunsignede_v2df): Add define_expands.
23786 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
23787 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
23788 VEC_UNSIGNEDO): Add definitions.
23789 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23790 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
23791 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
23792 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
23793 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
23794 * config/rs6000/altivec.h (vec_signed, vec_signed2,
23795 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
23796 vec_unsignede, vec_unsignedo): Add builtin defines.
23797 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
23799 * doc/extend.texi: Update the built-in documentation file for the
23800 new built-in functions.
23802 2017-06-29 Richard Biener <rguenther@suse.de>
23804 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
23805 reduction chains to LOOP_VINFO_REDUCTIONS.
23806 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
23807 SLP reductions after processing reduction chains.
23809 2017-06-29 Nathan Sidwell <nathan@acm.org>
23811 * builtins.c (fold_builtin_FUNCTION): Use
23812 lang_hooks.decl_printable_name.
23814 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
23816 PR middle-end/81194
23817 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
23818 with only one label.
23819 * stmt.c (expand_case): Assert NCASES is greater than one.
23821 2017-06-29 Richard Biener <rguenther@suse.de>
23823 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
23825 (group_case_labels): Likewise.
23826 (find_taken_edge): Push sanity checking on val to workers...
23827 (find_taken_edge_cond_expr): ... here
23828 (find_taken_edge_switch_expr): ... and here, handle cases
23829 with just a default label.
23830 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
23831 (group_case_labels): Likewise.
23832 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
23833 group_case_labels does anything cleanup the CFG again.
23835 2017-06-29 Bin Cheng <bin.cheng@arm.com>
23837 PR tree-optimization/81196
23838 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
23839 exit condition comparing two IVs.
23841 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
23843 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
23844 profile to the dummy entry at the end of the list of architectures.
23845 * config/arm/arm-cpu-cdata.h: Regenerated.
23847 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23848 Michael Collison <michael.collison@arm.com>
23851 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
23853 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
23854 (*aarch64_reg_<mode>3_minus_mask): New pattern.
23855 (*aarch64_<optab>_reg_di3_mask2): New pattern.
23856 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
23857 of shift when the shift amount is masked with constant equal to
23858 the size of the mode.
23859 * config/aarch64/predicates.md (subreg_lowpart_operator): New
23862 2017-06-29 Martin Liska <mliska@suse.cz>
23864 * config/i386/i386.opt: Change range from [1,5] to [0,5].
23866 2017-06-29 Yury Gribov <tetra2005@gmail.com>
23869 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
23872 (edge_growth_cache_entry::edge_growth_cache_entry): New
23874 (reset_edge_growth_cache): Update to use constructor.
23876 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23878 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
23879 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23880 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23882 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
23884 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
23885 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
23887 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
23889 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
23890 (*-linux-uclibc*): Add t-uclibc tmake_file.
23891 * config/t-musl: New.
23892 * config/t-uclibc: New.
23894 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
23896 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
23898 (gen_comm_data): Emit architectural setting of arch_prof.
23899 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
23901 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
23902 (armv8-m.base, armv8-m.main): Likewise.
23903 * arm-protos.h (arm_build_target): Add profile field.
23904 (arch_option): Likewise.
23905 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
23907 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
23908 arm_active_target.profile.
23910 2017-06-28 Richard Biener <rguenther@suse.de>
23912 PR middle-end/81227
23913 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
23914 TYPE_OVERFLOW_WRAPS.
23915 * match.pd (negate_expr_p): Likewise.
23916 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
23917 fold_build2, not fold_binary.
23919 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23921 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
23922 Convert memory address to Pmode.
23923 (aarch64_print_operand): Assert MEM operands are always Pmode.
23925 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23928 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
23929 Remove redundant if.
23930 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
23931 * config/arm/aarch-common-protos.h
23932 (aarch_forward_to_shift_is_not_shifted_re): Remove.
23933 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
23935 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
23938 * multiple_target.c (create_dispatcher_calls): Set the default
23939 clone to be static, not public.
23941 2017-06-28 Richard Biener <rguenther@suse.de>
23943 * tree-vect-loop.c (vectorizable_reduction): Move special
23944 cond reduction IV var creation ...
23945 (vect_create_epilog_for_reduction): ... here. Remove induction_index
23946 parameter. Use STMT_VINFO_VECTYPE.
23947 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
23950 2017-06-28 Martin Liska <mliska@suse.cz>
23953 * ipa-visibility.c (non_local_p): Handle visibility.
23955 2017-06-28 Martin Liska <mliska@suse.cz>
23958 * common.opt: Add IntegerRange to various options.
23959 * opt-functions.awk (integer_range_info): New function.
23960 * optc-gen.awk: Add integer_range_info to cl_options struct.
23961 * opts-common.c (decode_cmdline_option): Handle
23962 CL_ERR_INT_RANGE_ARG.
23963 (cmdline_handle_error): Likewise.
23964 * opts.c (print_filtered_help): Show valid interval in
23965 when --help is provided.
23966 * opts.h (struct cl_option): Add range_min and range_max fields.
23967 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
23969 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
23971 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
23972 (x * C EQ/NE y * C): New transformation.
23974 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
23976 * genmultilib (combination_space): Accept '+' in option names.
23978 2017-06-28 Martin Liska <mliska@suse.cz>
23981 * asan.c (instrument_derefs): Bail out inner references
23982 that are hard register variables.
23984 2017-06-28 Jakub Jelinek <jakub@redhat.com>
23987 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
23988 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
23990 2017-06-28 Richard Biener <rguenther@suse.de>
23992 * tree-vectorizer.h (vect_get_vec_defs): Remove.
23993 (vect_get_slp_defs): Adjust.
23994 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
23996 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
23998 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
23999 get_initial_defs_for_reduction instead of vect_get_vec_defs.
24000 (vectorizable_reduction): Adjust.
24001 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
24003 (vect_get_slp_defs): Likewise.
24004 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
24005 (vectorizable_bswap): Adjust.
24006 (vectorizable_call): Likewise.
24007 (vectorizable_conversion): Likewise.
24008 (vectorizable_assignment): Likewise.
24009 (vectorizable_shift): Likewise.
24010 (vectorizable_operation): Likewise.
24011 (vectorizable_store): Likewise.
24012 (vectorizable_condition): Likewise.
24013 (vectorizable_comparison): Likewise.
24015 2017-06-28 Michael Collison <michael.collison@arm.com>
24018 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
24020 (arm_gen_movmemqi): Removed unused variable 'i'.
24021 Convert 'for' loop into 'while' loop.
24022 (arm_expand_prologue): Remove last unnecessary set of insn.
24023 (thumb_pop): Remove unused variable 'pushed_words'.
24024 (thumb_exit): Remove last unnecessary set of regs_to_pop.
24026 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24028 * config/s390/predicates.md: Use s390_rel_address_ok_p.
24029 * config/s390/s390-protos.h: Add prototype of
24030 s390_rel_address_ok_p.
24031 * config/s390/s390.c (s390_got_symbol): New function.
24032 (s390_rel_address_ok_p): New function.
24033 (legitimize_pic_address): Use s390_rel_address_ok_p.
24034 (s390_load_got): Use s390_got_symbol.
24035 (s390_option_override): Issue error if
24036 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
24037 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
24039 * config/s390/s390.opt: New option mpic-data-is-text-relative.
24041 2017-06-27 Andrew Pinski <apinski@cavium.com>
24043 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
24044 (X * copysign (1.0, X)): New pattern.
24045 (X * copysign (1.0, -X)): New pattern.
24046 (copysign (-1.0, CST)): New pattern.
24048 2017-06-27 Joseph Myers <joseph@codesourcery.com>
24050 * genmultilib (combination_space): Remove variable.
24051 Validate reuse rules against regular expression for any sequence
24052 of multilib options in any order.
24054 2017-06-27 Michael Collison <michael.collison@arm.com>
24056 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
24057 call aarch64_split_simd_combine.
24058 * (aarch64_combine_internal<mode>): Delete pattern.
24059 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
24060 Allow register and subreg operands.
24062 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24064 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
24065 specific need, just fallback on defaults.
24066 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
24068 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24069 Olivier Hainque <hainque@adacore.com>
24071 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
24073 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
24074 targets. Pick a default if no particular attempt applied.
24075 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
24078 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24080 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
24081 (x86_64-wrs-vxworks7): Likewise.
24083 2017-06-27 Marek Polacek <polacek@redhat.com>
24086 * ubsan.c (instrument_null): Check get_base_address's result for null.
24088 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
24090 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
24092 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
24094 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
24095 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
24096 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
24097 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
24098 New function types.
24099 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
24100 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
24101 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
24102 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
24103 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
24104 BUILT_IN_FEUPDATEENV): New builtins.
24105 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
24106 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
24107 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
24108 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
24110 (builtin_structptr_types): Adjust size.
24111 * tree.c (builtin_structptr_types): Add four entries.
24113 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24114 Olivier Hainque <hainque@adacore.com>
24116 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
24117 (TLS_SYM): New local macro, forcing reference to __tls__ on
24118 link command lines for VxWorks 7 RTPs, triggering initialization
24120 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
24121 OS features TLS support, true for RTPs on VxWorks 7.
24122 * config/vxworks.c (vxworks_override_options): Setup emutls
24125 2017-06-27 Jakub Jelinek <jakub@redhat.com>
24127 * predict.c (test_prediction_value_range): Use -1U instead of -1
24128 to avoid narrowing conversion warning.
24129 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
24130 to avoid narrowing conversion warning.
24131 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
24133 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
24135 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24137 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
24138 64bit configurations.
24139 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
24140 (SIZE_TYPE): Likewise.
24141 * config/vxworks.c (vxworks_emutls_var_fields): Use
24142 long_unsigned_type_node instead of unsigned_type_node as the offset
24143 field type, which is "pointer" mode in emutls.c.
24145 2017-06-27 Jakub Jelinek <jakub@redhat.com>
24148 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
24150 PR middle-end/81207
24151 * gimple-fold.c (replace_call_with_call_and_fold): Handle
24152 gimple_vuse copying separately from gimple_vdef copying.
24154 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24156 * value-prof.c (free_hist): Remove call to memset and the enclosing if
24159 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
24160 Olivier Hainque <hainque@adacore.com>
24162 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
24163 for all vxworks7 targets.
24164 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
24165 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
24166 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
24167 variations for VX6/VX7 and 32/64bits later on in ...
24168 (VXWORKS_LIB_SPEC): Leverage new macros.
24169 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
24170 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
24172 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
24174 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
24175 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
24177 2017-06-26 Carl Love <cel@us.ibm.com>
24179 * config/rs6000/rs6000-c.c: Add support for built-in functions
24180 vector bool char vec_reve (vector bool char);
24181 vector signed char vec_reve (vector signed char);
24182 vector unsigned char vec_reve (vector unsigned char);
24183 vector bool int vec_reve (vector bool int);
24184 vector signed int vec_reve (vector signed int);
24185 vector unsigned int vec_reve (vector unsigned int);
24186 vector bool long long vec_reve (vector bool long long);
24187 vector signed long long vec_reve (vector signed long long);
24188 vector unsigned long long vec_reve (vector unsigned long long);
24189 vector bool short vec_reve (vector bool short);
24190 vector signed short vec_reve (vector signed short);
24191 vector double vec_reve (vector double);
24192 vector float vec_reve (vector float);
24193 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
24194 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
24195 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
24196 (altivec_vreve): New pattern.
24197 * config/rs6000/altivec.h (vec_reve): New define.
24198 * doc/extend.texi (vec_rev): Update the built-in documentation file
24199 for the new built-in functions.
24201 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24203 PR tree-optimization/71815
24204 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
24206 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
24208 (slsr_process_phi): Likewise.
24209 (replace_uncond_cands_and_profitable_phis): Don't replace a
24210 multiply candidate with a stride of 1 (copy or cast).
24211 (phi_incr_cost): Call uses_consumed_by_stmt rather than
24213 (lowest_cost_path): Likewise.
24214 (total_savings): Likewise.
24216 2017-06-26 Richard Biener <rguenther@suse.de>
24219 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
24220 Use def_builtin_pure for all gather builtins.
24222 2017-06-26 Richard Biener <rguenther@suse.de>
24224 PR tree-optimization/81203
24225 * tree-tailcall.c (find_tail_calls): Do not move stmts into
24226 non-dominating BBs.
24228 2017-06-26 Marek Polacek <polacek@redhat.com>
24231 * doc/invoke.texi: Document -Wmultistatement-macros.
24233 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
24235 * doc/sourcebuild.texi (ARM-specific attributes): Document new
24236 arm_neon_ok_no_float_abi effective target.
24238 2017-06-26 Richard Biener <rguenther@suse.de>
24240 PR tree-optimization/80928
24241 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
24242 (copy_bbs): Set BB_DUPLICATED flag early.
24243 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
24245 (execute_on_shrinking_pred): Likewise.
24246 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
24247 BB_DUPLICATED blocks.
24248 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
24249 iterate over all PHIs considering removal of *gsi.
24251 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
24253 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
24256 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24258 * config/rs6000/rs6000-string.c: (expand_block_clear,
24259 do_load_for_compare, select_block_compare_mode,
24260 compute_current_alignment, expand_block_compare,
24261 expand_strncmp_align_check, expand_strn_compare,
24262 expand_block_move, rs6000_output_load_multiple)
24263 Move functions related to string/block move/compare
24264 to a separate file.
24265 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
24266 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
24267 for this function which is now used in two files.
24268 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
24269 * config.gcc: Add rs6000-string.o to extra_objs for
24270 targets powerpc*-*-* and rs6000*-*-*.
24272 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
24275 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
24276 32-bit, since indexed is not valid for DImode.
24277 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
24278 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
24279 (define_peephole2 for Altivec d-form load): Add 32-bit support.
24280 (define_peephole2 for Altivec d-form store): Likewise.
24283 * multiple_target.c (create_dispatcher_calls): Only create the
24284 dispatcher call if the function is the default clone of a
24285 versioned function.
24287 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
24289 PR middle-end/80902
24290 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
24291 a call, force the call to not be a tail call.
24293 2017-06-23 Jeff Law <law@redhat.com>
24295 * doc/contrib.texi: Add entry for Steven Pemberton's work on
24298 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
24300 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
24301 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
24302 handling for early expansion of vector shifts (sl,sr,sra,rl).
24303 (builtin_function_type): Add vector shift right instructions
24304 to the unsigned argument list.
24306 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
24308 rtl-optimizatoin/79286
24309 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
24310 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
24311 trap. PIC register plus a const unspec without offset can never trap.
24313 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
24315 * tree.h (builtin_structptr_type): New type.
24316 (builtin_structptr_types): Declare new array.
24317 * tree.c (builtin_structptr_types): New array.
24318 (free_lang_data, build_common_tree_nodes): Use it.
24320 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
24323 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
24326 2017-06-22 Matt Turner <mattst88@gmail.com>
24328 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
24329 Lake models to skylake case. Assume skylake for unknown
24330 models with clflushopt.
24332 2017-06-22 Jeff Law <law@redhat.com>
24334 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
24335 frame sizes that do not satisfy aarch64_uimm12_shift.
24337 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
24339 * profile-count.h (apply_probability,
24340 apply_scale, probability_in): Fix checks for zero.
24342 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24344 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
24345 * doc/cppdiropts.texi (-I @var{dir}): Document it.
24347 2016-06-22 Richard Biener <rguenther@suse.de>
24349 * tree-vect-loop.c (vect_model_reduction_cost): Handle
24350 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
24351 REDUC_MAX_EXPR support.
24352 (vectorizable_reduction): Likewise.
24353 (vect_create_epilog_for_reduction): Likewise.
24355 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
24357 * match.pd (A / (1 << B) -> A >> B): New.
24358 * generic-match-head.c: Include optabs-tree.h.
24359 * gimple-match-head.c: Likewise.
24360 * optabs-tree.h (target_supports_op_p): New.
24361 * optabs-tree.c (target_supports_op_p): New.
24363 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24365 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
24366 $gcc_cv_ld --help output.
24367 (gcc_cv_ld_demangle): Likewise.
24368 (gcc_cv_ld_eh_frame_hdr): Likewise.
24369 (gcc_cv_ld_pie): Likewise.
24370 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
24371 (gcc_cv_ld_buildid): Likewise.
24372 (gcc_cv_ld_sysroot): Likewise.
24373 (ld_bndplt_support): Likewise.
24374 (ld_pushpopstate_support): Likewise.
24375 * configure: Regenerate.
24376 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
24378 2017-06-21 Jakub Jelinek <jakub@redhat.com>
24381 * config/i386/sse.md (round<mode>2): Renumber match_dup and
24382 operands indexes to avoid gap between operands and match_dups.
24384 2017-06-21 Andrew Pinski <apinski@cavium.com>
24386 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
24387 Increment Arith_shift and Arith_shift_reg by 1.
24388 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
24390 * config/aarch64/aarch64.c (thunderx_tunings): Enable
24391 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
24392 (aarch64_strip_extend): Add new argument and test for it.
24393 (aarch64_cheap_mult_shift_p): New function.
24394 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
24395 add a cost if it is true.
24396 Update calls to aarch64_strip_extend.
24397 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
24399 2017-06-21 Andrew Pinski <apinski@cavium.com>
24401 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
24403 (thunderxt88): Likewise.
24404 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
24405 (thunderx_prefetch_tune): New variable.
24406 (thunderx2t99_prefetch_tune): Update for the correct values.
24407 (thunderxt88_tunings): New variable.
24408 (thunderx_tunings): Use thunderx_prefetch_tune instead of
24409 generic_prefetch_tune.
24410 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
24412 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24414 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
24415 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
24416 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
24417 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
24418 (aarch64_atomic_cas<mode>, GPI): Likewise.
24420 2017-06-21 Martin Liska <mliska@suse.cz>
24422 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
24423 statements on cold and hot labels.
24424 * predict.c (tree_estimate_probability_bb): Remove the
24425 prediction from this place.
24427 2017-06-21 Martin Liska <mliska@suse.cz>
24429 PR tree-optimization/79489
24430 * gimplify.c (maybe_add_early_return_predict_stmt): New
24432 (gimplify_return_expr): Call the function.
24433 * predict.c (tree_estimate_probability_bb): Remove handling
24435 * predict.def: Update comment about early return predictor.
24436 * gimple-predict.h (is_gimple_predict): New function.
24437 * predict.def: Change default value of early return to 66.
24438 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
24440 * passes.def: Put pass_strip_predict_hints to the beginning of
24443 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
24445 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
24446 FUNCTION_DECL declarations.
24447 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
24449 (dwaf2out_decl): Likewise.
24450 * godump.c (go_early_global_decl): Skip call to the real debug hook
24451 for FUNCTION_DECL declarations.
24452 * passes.c (rest_of_decl_compilation): Skip call to the
24453 early_global_decl debug hook for FUNCTION_DECL declarations, unless
24454 -fdump-go-spec is passed.
24456 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
24458 * config/i386/i386.c (struct builtin_isa): New field pure_p.
24459 Reorder for compactness.
24460 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
24461 (def_builtin_pure, def_builtin_pure2): New functions.
24462 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
24464 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
24466 * match.pd (nop_convert): New predicate.
24467 ((A +- CST1) +- CST2): Allow some NOP conversions.
24469 2017-06-21 Jakub Jelinek <jakub@redhat.com>
24472 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
24473 with ctors/dtors if GOVD_SHARED is set.
24475 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24477 * config/aarch64/aarch64.md (movti_aarch64):
24478 Emit mov rather than orr.
24479 (movtf_aarch64): Likewise.
24480 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
24481 Emit mov rather than orr.
24483 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24485 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
24486 Swap alternatives, make integer dup more expensive.
24488 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24490 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
24491 Return true for non-tls symbols.
24493 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
24495 * config/aarch64/aarch64-cores.def (cortex-a55): New.
24496 (cortex-a75): Likewise.
24497 (cortex-a75.cortex-a55): Likewise.
24498 * config/aarch64/aarch64-tune.md: Regenerate.
24499 * doc/invoke.texi (-mtune): Document new values for -mtune.
24501 2017-06-21 Tom de Vries <tom@codesourcery.com>
24503 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
24504 stack_size feature.
24505 (Effective-Target Keywords, Other attributes): Suggest using
24506 dg-add-options stack_size feature to get stack limit in stack_size
24507 effective target documentation.
24509 2017-06-21 Julian Brown <julian@codesourcery.com>
24510 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24512 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
24513 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
24514 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
24516 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
24517 attribute type list for neon_multiply.
24518 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
24519 attribute type list for neon_multiply.
24520 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
24521 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
24522 attribute type list for neon_multiply.
24523 * config/arm/types.md (crypto_pmull): Add.
24524 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
24525 attribute type list.
24527 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
24529 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
24532 2017-06-20 Jakub Jelinek <jakub@redhat.com>
24534 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
24535 to make sure not to dereference a NULL cost_classes_ptr pointer.
24537 2017-06-20 Carl Love <cel@us.ibm.com>
24539 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24540 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
24541 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
24542 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
24543 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
24544 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
24545 VMULOSW): New enum "unspec" values.
24546 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
24547 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
24548 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
24549 altivec_vmulosw): New patterns.
24550 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
24551 VMULOSW): Add definitions.
24553 2017-06-20 Julia Koval <julia.koval@intel.com>
24555 * config/i386/i386.c: Fix rounding expand for new pattern.
24556 * config/i386/subst.md: Fix pattern (parallel -> unspec).
24558 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24560 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
24561 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
24563 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24565 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
24568 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24570 * config/aarch64/aarch64-cores.def: Rearrange to sort by
24571 architecture, then by implementer ID.
24572 * config/aarch64/aarch64-tune.md: Regenerate.
24574 2017-06-20 Richard Biener <rguenther@suse.de>
24576 PR middle-end/81097
24577 * fold-const.c (split_tree): Fold to type before negating.
24579 2017-06-20 David Malcolm <dmalcolm@redhat.com>
24581 * diagnostic-show-locus.c
24582 (selftest::test_fixit_deletion_affecting_newline): New function.
24583 (selftest::diagnostic_show_locus_c_tests): Call it.
24585 2017-06-20 Andreas Schwab <schwab@suse.de>
24588 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
24591 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24593 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
24594 __ARM_FEATURE_COPROC according to support.
24596 2017-06-20 Jakub Jelinek <jakub@redhat.com>
24598 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
24599 Rewritten to avoid overflow for > 32-bit pointers.
24602 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
24603 by removing enum keyword.
24604 (ubsan_type_descriptor): Likewise. Formatting fix.
24607 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
24608 splitter): Require TARGET_SSE2 in the condition.
24610 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
24613 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
24614 for doing vector set of SFmode on ISA 3.0.
24615 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
24616 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
24618 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
24619 SFmode value into a V4SF variable that was extracted from another
24620 V4SF variable without converting the element to double precision
24621 and back to single precision vector format.
24622 (vsx_insert_extract_v4sf_p9_2): Likewise.
24624 2017-06-19 Jakub Jelinek <jakub@redhat.com>
24626 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
24627 in UWHI to avoid undefined overflow.
24630 * ubsan.h (enum ubsan_encode_value_phase): New.
24631 (ubsan_encode_value): Change second argument to
24632 enum ubsan_encode_value_phase with default value of
24633 UBSAN_ENCODE_VALUE_GENERIC.
24634 * ubsan.c (ubsan_encode_value): Change second argument to
24635 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
24636 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
24637 create_tmp_var_raw instead of create_tmp_var and use a
24639 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
24640 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
24641 ubsan_encode_value callers.
24644 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
24645 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
24646 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
24648 2017-06-19 Richard Biener <rguenther@suse.de>
24650 PR middle-end/81118
24651 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
24652 estimates if we changed anything.
24654 2017-06-19 Richard Biener <rguenther@suse.de>
24656 PR tree-optimization/80887
24657 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
24658 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
24659 simplified lookups, then reset mprts_hook.
24660 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
24662 (try_to_simplify): Likewise.
24664 2017-06-19 Martin Liska <mliska@suse.cz>
24667 * gimplify.c (gimplify_switch_expr):
24668 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
24670 2017-06-19 Martin Liska <mliska@suse.cz>
24672 * doc/install.texi: Document that PGO runs in 4 stages.
24674 2017-06-19 Martin Liska <mliska@suse.cz>
24677 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
24678 to dispatcher function name.
24679 * multiple_target.c (replace_function_decl): New function.
24680 (create_dispatcher_calls): Redirect both edges and references.
24682 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
24684 * profile-count.c (profile_count::dump): Dump quality.
24685 (profile_count::differs_from_p): Update for unsigned val.
24686 * profile-count.h (profile_count_quality): New enum.
24687 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
24689 2017-06-19 Richard Biener <rguenther@suse.de>
24691 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
24692 struct function as arg.
24693 (estimate_numbers_of_iterations): Export overload with loop arg.
24694 (free_numbers_of_iterations_estimates_loop): Use an overload of
24695 free_numbers_of_iterations_estimates instead.
24696 * tree-cfg.c (remove_bb): Adjust.
24697 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
24698 * tree-parloops.c (gen_parallel_loop): Likewise.
24699 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
24701 (tree_unroll_loops_completely): Likewise.
24702 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
24703 Use an overload instead and export.
24704 (estimated_loop_iterations): Adjust.
24705 (max_loop_iterations): Likewise.
24706 (likely_max_loop_iterations): Likewise.
24707 (estimate_numbers_of_iterations): Take struct function as arg
24709 (loop_exits_before_overflow): Adjust.
24710 (free_numbers_of_iterations_estimates_loop): Use an overload.
24711 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
24712 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
24714 2017-06-19 Richard Biener <rguenther@suse.de>
24717 * ipa-prop.c (find_constructor_constant_at_offset): Handle
24718 RANGE_EXPR conservatively.
24720 2017-06-16 Carl Love <cel@us.ibm.com>
24722 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24723 definitions for vec_float, vec_float2, vec_floato,
24724 vec_floate built-ins.
24725 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
24726 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
24728 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
24729 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
24730 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
24731 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
24732 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
24733 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
24734 vec_floato): Add builtin defines.
24735 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
24736 Update the built-in documentation file for the new built-in
24739 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24741 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
24742 (mthumb): Mark as the negative of -marm.
24744 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24746 * doc/invoke.texi (ARM Options, -mcpu): Document supported
24748 (ARM Options, -mtune): Document that this accepts the same
24749 extension options as -mcpu.
24750 (ARM Options, -mfpu): Document addition of -mfpu=auto.
24752 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24754 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
24755 permitted extensions.
24757 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24759 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
24760 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
24761 (armv8-m.main): Add option +nodsp.
24762 * config/arm/arm-cpu-cdata.h: Regenerated.
24764 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24766 * config/arm/t-fuchsia: New file.
24767 * config.gcc (arm*-*-fuchsia*): Use it.
24769 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24771 * config/arm/t-symbian: Rewrite for new option infrastructure.
24773 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24775 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
24776 (MULTILIB_REQUIRED): Likewise.
24778 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24780 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
24781 (MULTILIB_RESUE): Likewise.
24782 (MULTILIB_MATCHES): Likewise.
24783 (MULTLIB_REQUIRED): Likewise.
24785 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24787 * config/arm/t-rtems: Rewrite for new option framework.
24789 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24791 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
24792 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
24793 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
24794 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
24795 * config/arm/t-multilib: ... here.
24796 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
24797 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
24798 armv7-a and armv8*-a when A-profile libraries have not been built.
24799 * config/arm/t-rmprofile: Rewrite.
24801 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24803 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
24804 with a backslash. Remove the backslash after substituting unescaped
24806 * doc/fragments.texi (MULTILIB_REUSE): Document it.
24808 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24810 * config.gcc: (arm*-*-*): When building a-profile libraries, force
24811 the driver to pass through the default setting of -mfloat-abi.
24812 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
24814 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
24815 (all_feat_combs): New rule.
24816 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
24818 * config/arm/t-aprofile: Rewrite.
24820 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24822 * config/arm/arm.h (FPUTYPE_AUTO): Define.
24823 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
24824 fpu is not specified by the user/command-line.
24825 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
24826 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
24827 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
24828 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
24829 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
24830 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
24832 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24834 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
24835 * config/arm/t-arm-elf: Rewritten.
24837 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24839 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
24840 have some floating-point instructions.
24841 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
24842 (TARGET_MAYBE_HARD_FLOAT): New macro.
24843 * config/arm/arm-builtins.c (arm_init_builtins): Use
24844 TARGET_MAYBE_HARD_FLOAT.
24845 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
24847 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24849 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
24850 (configargs.h): Include it.
24851 (arm_print_hint_for_fpu_option): New function.
24852 (arm_parse_fpu_option): New function.
24853 (candidate_extension): New class.
24854 (arm_canon_for_multilib): New function.
24855 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
24856 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
24857 (ARCH_CANONICAL_SPECS): New macro.
24858 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
24860 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24862 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
24863 are set after handling multilib fragments. Set target_cpu_default2
24866 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24868 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
24870 (arm*-*-*): Set target_cpu_default2 to a quoted string.
24871 * config/arm/parsecpu.awk (check_cpu): Validate any extension
24873 (check_arch): Likewise.
24874 * config/arm/arm.c (arm_configure_build_target): Handle
24875 TARGET_CPU_DEFAULT being a string constant. Scan any feature
24876 options in the default.
24878 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24880 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
24881 when an option is an alias of another.
24882 * config/arm/parsecpu.awk (optalias): New parser token.
24883 (gen_comm_data): Mark non-alias options as such. Emit entries
24884 for extension aliases.
24885 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
24886 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
24887 (armv6kz, armv6zk, armv6t2): Likewise.
24888 (armv7): Make vfpv3-d16 an alias.
24889 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
24891 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
24892 Sort in canonical order.
24893 (armv8-a): Sort in canonical order.
24894 (armv8.1-a, armv8.2-a): Likewise.
24895 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
24897 (cortex-a9): Sort in canonical order.
24898 * config/arm/arm.c (selftests.h): Include it.
24899 (arm_test_cpu_arch_data): New function.
24900 (arm_run_self_tests): New function.
24901 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
24902 (targetm): Move declaration to the end of the file.
24903 * arm-cpu-cdata.h: Regenerated.
24905 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24907 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
24908 call to target_mode_check describing the type of option passed.
24909 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
24910 (arm_target_thumb_only): Use arm_parse_arch_option_name or
24911 arm_parse_cpu_option_name to match parameters against list of
24913 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
24914 arm_arch_core_flags data structure.
24915 * config/arm/arm-cpu_cdata.h: Regenerated.
24917 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24919 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
24921 (arm_print_hint_for_cpu_option): Likewise.
24922 (arm_print_hint_for_arch_option): Likewise.
24923 (arm_parse_cpu_option_name): Likewise.
24924 (arm_parse_arch_option_name): Likewise.
24925 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
24926 of entries in the all_fpus list.
24927 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
24928 (arm_parse_cpu_option_name): Declare.
24929 (arm_parse_arch_option_name): Declare.
24930 (arm_parse_option_features): Declare.
24931 (arm_intialize_isa): Declare.
24932 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
24934 (gen_comm_data): ... here. Make definitions non-static.
24935 * config/arm/arm-cpu-data.h: Regenerated.
24936 * config/arm/arm-cpu-cdata.h: Regenerated.
24938 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24940 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
24941 (cpu_arch_extension): New structure.
24942 (cpu_arch_option, arch_option, cpu_option): New structures.
24943 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
24944 architecture types.
24945 (gen_data): Generate new format data tables.
24946 * config/arm/arm.c (cpu_tune): New structure.
24947 (cpu_option, processors): Delete.
24948 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
24949 (arm_print_hint_for_cpu_option): ... this and ...
24950 (arm_print_hint_for_arch_option): ... this.
24951 (arm_parse_arch_cpu_name): Delete. Replace with ...
24952 (arm_parse_cpu_option_name): ... this and ...
24953 (arm_parse_arch_option_name): ... this.
24954 (arm_unrecognized_feature): Change type of target parameter to
24956 (arm_parse_arch_cpu_features): Delete. Replace with ...
24957 (arm_parse_option_features): ... this.
24958 (arm_configure_build_target): Rework to use new configuration data
24960 (arm_print_tune_info): Rework for new configuration data tables.
24961 * config/arm/arm-cpu-data.h: Regenerated.
24962 * config/arm/arm-cpu.h: Regenerated.
24964 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24966 * Makefile.in (OBJS): Move sbitmap.o from here ...
24967 (OBJS-libcommon): ... to here.
24969 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24971 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
24972 (ISA_ALL_CRYPTO): New macro.
24973 (ISA_ALL_SIMD): New macro
24974 (ISA_ALL_FP): New macro.
24975 * config/arm/arm.c (fpu_bitlist): Update initializer.
24976 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
24978 (arm9e): Add fpu. Add option for nofp
24979 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
24980 (arm926ej-s, arm1026ej-s): Likewise.
24981 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
24982 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
24983 neon-fp16, neon-vfpv4, nofp and nosimd.
24984 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
24985 (cortex-a8): Add fpu. Add option for nofp.
24986 (cortex-a9): Add fpu. Add options for nosimd and nofp.
24987 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
24988 (cortex-r4f): Add fpu.
24989 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
24990 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
24992 (cortex-r8): Likewise.
24993 (cortex-m4): Add fpu. Add option for nofp.
24994 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
24995 (cortex-a17.cortex-a7): Likewise.
24996 (cortex-a32): Add fpu. Add options for crypto and nofp.
24997 (cortex-a35, cortex-a53): Likewise.
24998 (cortex-a57): Add fpu. Add option for crypto.
24999 (cortex-a72, cortex-a73): Likewise.
25000 (exynos-m1): Likewise.
25001 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
25002 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
25003 (cortex-m33): Add fpu. Add option for nofp.
25004 * config/arm/arm-cpu-cdata.h: Regenerated
25005 * config/arm/arm-cpu-data.h: Regenerated.
25007 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25009 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
25010 (armv5te, armv5tej): Likewise.
25011 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
25012 (armv7): Add options fp and vfpv3-d16.
25013 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
25014 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
25016 (armv7ve): Likewise.
25017 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
25018 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
25019 (armv8-a): Add nocrypto option.
25020 (armv8.1-a, armv8.2-a): Likewise.
25021 (armv8-m.main): add options fp, fp.dp and nofp.
25023 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25025 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
25027 (armv8-a+crc): Delete.
25028 (armv8.1-a): Add options simd, crypto and nofp.
25029 (armv8.2-a): Add options fp16, simd, crypto and nofp.
25030 (armv8.2-a+fp16): Delete.
25031 (armv8-m.main): Add option dsp.
25032 (armv8-m.main+dsp): Delete.
25033 (cortex-a8): Add fpu. Add nofp option.
25034 (cortex-a9): Add fpu. Add nofp and nosimd options.
25035 * config/arm/parsecpu.awk (gen_data): Generate option tables and
25036 link to main cpu and architecture data structures.
25037 (gen_comm_data): Only put isa attributes from the main architecture
25039 (option): New statement for architecture and CPU entries.
25040 * arm.c (struct cpu_option): New structure.
25041 (struct processors): Add entry for options.
25042 (arm_unrecognized_feature): New function.
25043 (arm_parse_arch_cpu_name): Ignore any characters after the first
25045 (arm_parse_arch_cpu_feature): New function.
25046 (arm_configure_build_target): Separate out any CPU and architecture
25047 features and parse separately. Don't error out if -mfpu=auto is
25048 used with only an architecture string.
25049 (arm_print_asm_arch_directives): New function.
25050 (arm_file_start): Call it.
25051 * config/arm/arm-cpu-cdata.h: Regenerated.
25052 * config/arm/arm-cpu-data.h: Likewise.
25053 * config/arm/arm-tables.opt: Likewise.
25055 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25057 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
25058 assembler when it is not -mfpu=auto.
25060 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25062 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
25063 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
25064 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
25065 (ASM_CPU_SPEC): Rewrite.
25066 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
25067 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
25068 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
25069 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
25070 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
25071 copied string is NUL-terminated. Also strip any characters prefixed
25073 (arm_rewrite_selected_arch): New function.
25074 (arm_rewrite_march): New function.
25076 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25078 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
25079 (x_arm_cpu_string, x_arm_tune_string): Likewise.
25080 (march, mcpu, mtune): Convert to string-based options.
25081 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
25082 (arm_parse_arch_cpu_name): New function.
25083 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
25084 identify selected architecture or CPU.
25085 (arm_option_save): New function.
25086 (TARGET_OPTION_SAVE): Redefine.
25087 (arm_option_restore): Restore string options.
25088 (arm_option_print): Print string options.
25090 2017-06-16 Martin Sebor <msebor@redhat.com>
25092 PR tree-optimization/80933
25093 PR tree-optimization/80934
25094 * builtins.c (fold_builtin_3): Do not handle bcmp here.
25095 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
25096 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
25097 (gimple_fold_builtin): Call them.
25099 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25101 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
25102 as unlikely; update profile.
25104 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25106 * predict.c (force_edge_cold): Handle declaring edges impossible
25109 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25111 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
25113 (try_unroll_loop_completely): Fix reporting.
25115 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25117 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
25119 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
25122 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
25123 if given a non-constant argument for an intrinsic which requires a
25126 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25128 * profile.c (compare_freqs): New function.
25129 (branch_prob): Sort edge list.
25130 (find_spanning_tree): Assume that the list is priority sorted.
25132 2017-06-16 Richard Biener <rguenther@suse.de>
25134 PR tree-optimization/81090
25135 * passes.def (pass_record_bounds): Remove.
25136 * tree-pass.h (make_pass_record_bounds): Likewise.
25137 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
25138 make_pass_record_bounds): Likewise.
25139 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
25140 not free niter estimates at the beginning but at the end.
25141 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
25143 2017-06-16 Richard Biener <rguenther@suse.de>
25145 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
25146 initializer to workaround ICE in host GCC 4.8.
25148 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25150 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
25152 (clone_inlined_nodes): Update.
25154 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25156 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
25157 prefetch settings, and enable prefetching by default at -O3.
25159 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25161 * config/aarch64/aarch64.c (aarch64_override_options_internal):
25162 Set flag_prefetch_loop_arrays according to tuning data.
25164 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25166 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
25167 New tune structure.
25168 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
25169 [Unrelated to main purpose of the patch] Place the pointer field last
25170 to enable type checking errors when tune structure are wrongly merged.
25171 * config/aarch64/aarch64.c (generic_prefetch_tune,)
25172 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
25173 (thunderx2t99_prefetch_tune): New tune constants.
25174 (tune_params *_tunings): Update all tunings (no functional change).
25175 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
25176 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
25177 from tunings structures.
25179 2017-06-16 Jakub Jelinek <jakub@redhat.com>
25182 * ubsan.c (instrument_null): Add T argument, use it instead
25183 of computing it based on IS_LHS.
25184 (instrument_object_size): Likewise.
25185 (pass_ubsan::execute): Adjust instrument_null and
25186 instrument_object_size callers to pass gimple_get_lhs or
25187 gimple_assign_rhs1 result to it. Use instrument_null instead of
25188 calling get_base_address and instrument_mem_ref. Handle
25189 aggregate call arguments for object-size sanitization.
25191 2017-06-16 Yury Gribov <tetra2005@gmail.com>
25193 PR tree-optimization/81089
25194 * tree-vrp.c (is_masked_range_test): Validate operands of
25197 2017-06-15 Martin Sebor <msebor@redhat.com>
25200 * dumpfile.c (dump_register): Avoid calling memset to initialize
25201 a class with a default ctor.
25202 * gcc.c (struct compiler): Remove const qualification.
25203 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
25204 * hash-table.h: Ditto.
25205 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
25207 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
25208 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
25210 * params.h (struct param_info): Make struct members non-const.
25211 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
25212 with default initialization.
25213 * vec.h (vec_copy_construct, vec_default_construct): New helper
25215 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
25216 with vec_copy_construct.
25217 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
25218 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
25219 * doc/invoke.texi (-Wclass-memaccess): Document.
25221 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25223 * emit-rtl.h (is_leaf): Update comment about local
25224 register allocator.
25226 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
25229 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
25230 for a variable to have a section before checking if the section has a
25232 Set section to.persistent if persistent attribute is set.
25233 Warn if .persistent attribute is used on an automatic variable.
25235 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
25237 PR rtl-optimization/80474
25238 * reorg.c (update_block): Do not ignore instructions in a delay slot.
25240 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
25242 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
25245 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
25247 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
25248 (casesi): Emit bounds checking as RTL.
25249 (casesi_internal_mips16_<mode>): Remove bounds checking.
25251 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
25253 * config/xtensa/xtensa.c (xtensa_option_override): Append
25254 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
25255 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
25256 xtensa_doloop_hooks): Define unconditionally.
25257 (xtensa_reorg_loops): Only call reorg_loops in the presence of
25259 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
25260 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
25261 for it in xtensa_option_override.
25262 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
25263 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
25265 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
25267 * doc/cppopts.texi: Document '-' special value to -MF.
25269 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
25271 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
25272 (cortex_a53_fconst): Likewise.
25273 (cortex_a53_fpmul): Likewise.
25274 (cortex_a53_f_load_64): Likewise.
25275 (cortex_a53_f_load_many): Likewise.
25276 (cortex_a53_advsimd_alu): Likewise.
25277 (cortex_a53_advsimd_alu_q): Likewise.
25278 (cortex_a53_advsimd_mul): Likewise.
25279 (cortex_a53_advsimd_mul_q): Likewise.
25280 (fpmac bypass): Add new bypass for fpmac-fpmac case.
25281 Add missing fmul, r2f_cvt and fconst cases.
25283 2017-06-14 Richard Biener <rguenther@suse.de>
25285 PR middle-end/81088
25286 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
25288 (fold_binary_loc): When associating do not treat pre-existing
25289 TREE_OVERFLOW on literal constants as a reason to allow
25290 TREE_OVERFLOW on associated literal constants.
25292 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
25294 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
25295 (MASK_FEATURES): New macro.
25296 * config/sparc/sparc.c (sparc_option_override): Remove the special
25297 handling of -mfpu and generalize it to all MASK_FEATURES switches.
25299 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
25301 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
25302 a division of 0 if non-call exceptions are enabled.
25304 2017-06-14 Andrew Pinski <apinski@cavium.com>
25305 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25308 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
25309 Improve vector initialization code gen for only variable case.
25311 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
25313 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
25315 2017-06-14 Richard Biener <rguenther@suse.de>
25317 PR tree-optimization/81083
25318 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
25321 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25323 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
25324 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
25325 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
25326 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
25327 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
25328 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
25330 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25332 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
25333 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
25335 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25337 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
25339 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25341 * config/rs6000/t-rtems: Don't handle SPE.
25343 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25345 * config/rs6000/t-linux: Don't handle SPE.
25347 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25349 * config/rs6000/eabispe.h: Delete file.
25351 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25353 * config/rs6000/t-spe: Delete file.
25355 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25357 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
25358 (rs6000_legitimate_offset_address_p): Return false for anything in
25359 V2SImode or V2SFmode.
25361 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25363 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
25364 except V2SF and V2SI. Rearrange the vector modes, and add comments.
25365 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
25367 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
25368 (rs6000_legitimate_offset_address_p): Ditto.
25369 (rs6000_emit_move): Ditto.
25370 (rs6000_init_builtins): Remove V4HI_type_node.
25372 2017-06-13 Martin Liska <mliska@suse.cz>
25375 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
25376 (gate_asan): Likewise.
25377 * asan.h (asan_no_sanitize_address_p): Remove the function.
25378 (sanitize_flags_p): New function.
25379 * builtins.def: Fix coding style.
25380 * common.opt: Use renamed enum value.
25381 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
25382 * doc/extend.texi: Document no_sanitize attribute.
25383 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
25384 to SANITIZE_UNDEFINED_NONDEFAULT.
25385 * gcc.c (sanitize_spec_function): Use the renamed enum value.
25386 * gimple-fold.c (optimize_atomic_compare_exchange_p):
25387 Use sanitize_flags_p.
25388 * gimplify.c (gimplify_function_tree): Likewise.
25389 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
25390 * opts.c (parse_no_sanitize_attribute): New function.
25391 (common_handle_option): Use renamed enum value.
25392 * opts.h (parse_no_sanitize_attribute): Declare.
25393 * tree.c (sanitize_flags_p): New function.
25394 * tree.h: Declared here.
25395 * tsan.c: Use sanitize_flags_p.
25396 * ubsan.c (ubsan_expand_null_ifn): Likewise.
25397 (instrument_mem_ref): Likewise.
25398 (instrument_bool_enum_load): Likewise.
25399 (do_ubsan_in_current_function): Remove the function.
25400 (pass_ubsan::execute): Use sanitize_flags_p.
25401 * ubsan.h: Remove do_ubsan_in_current_function
25402 * tree-cfg.c (print_no_sanitize_attr_value): New function.
25403 (dump_function_to_file): Use it here.
25405 2017-06-13 Martin Jambor <mjambor@suse.cz>
25407 PR tree-optimization/80803
25408 PR tree-optimization/81063
25409 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
25410 (propagate_subaccesses_across_link): Enqueue subtree whenever
25411 necessary instead of relying on the caller.
25413 2017-06-13 Martin Jambor <mjambor@suse.cz>
25415 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
25416 that have a first_link.
25417 (sort_and_splice_var_accesses): Do not check first_link before
25419 (subtree_mark_written_and_enqueue): Likewise.
25420 (propagate_all_subaccesses): Likewise and do not stop at first
25421 parent with a first_link.
25423 2017-06-13 Martin Jambor <mjambor@suse.cz>
25425 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
25428 2017-06-13 Yury Gribov <tetra2005@gmail.com>
25430 * match.pd: New pattern.
25432 2017-06-13 Yury Gribov <tetra2005@gmail.com>
25434 * tree-vrp.c (is_masked_range_test): New function.
25435 (register_edge_assert_for): Determine ranges for
25438 2017-06-13 Yury Gribov <tetra2005@gmail.com>
25440 PR tree-optimization/67328
25441 * fold-const.c (maskable_range_p): New function.
25442 (build_range_check): Generate bittests if possible.
25444 2017-06-13 Martin Liska <mliska@suse.cz>
25446 * gimple-pretty-print.c (dump_probability): Add new argument.
25447 (dump_edge_probability): Dump both probability and count.
25448 (dump_gimple_label): Likewise.
25449 (dump_gimple_bb_header): Likewise.
25451 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
25454 * config/avr/avr-devices.c: Fix indentation.
25455 * config/avr/gen-avr-mmcu-specs.c: Dito.
25457 2017-06-13 Richard Biener <rguenther@suse.de>
25459 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
25460 instead get vector type from stmt_info.
25461 (vectorizable_reduction): Adjust. Remove dead code.
25463 2017-06-13 Richard Biener <rguenther@suse.de>
25465 PR middle-end/81065
25466 * fold-const.c (extract_muldiv_1): Remove bogus distribution
25467 case of C * (x * C2 + C3).
25468 (fold_addr_of_array_ref_difference): Properly fold index difference.
25470 2017-06-12 David S. Miller <davem@davemloft.net>
25473 * config/sparc/sparc.md (return expander): Emit frame blockage if
25474 function uses alloca.
25476 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
25478 * combine.c (make_field_assignment): Check len rather than the mode
25479 precision when calling force_to_mode.
25481 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
25483 Support multilibs and devices that see flash in RAM address range.
25486 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
25487 (avr_mcu_t) <flash_pm_offset>: New field.
25488 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
25489 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
25490 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
25491 (AVR_TINY_PM_OFFSET): Remove macro.
25492 * config/avr/avr.opt (-mshort-calls): New option.
25493 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
25494 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
25495 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
25496 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
25497 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
25498 instead of avr_arch->have_jmp_call.
25499 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
25500 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
25501 avr_arch->flash_pm_offset to define.
25502 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
25503 new field flash_pm_offset. Add entry for avrxmega3.
25504 (avr_texinfo): Add entry for avrxmega3.
25505 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
25506 attiny212, attiny214,
25507 attiny412, attiny414, attiny416, attiny417,
25508 attiny814, attiny816, attiny817,
25509 attiny1614, attiny1616, attiny1617,
25510 attiny3214, attiny3216, attiny3217.
25511 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
25512 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
25513 (avr_print_operand_address) [AVR_TINY]: Same.
25514 (avr_asm_init_sections) <readonly_data_section>: Only patch
25515 callback if avr_arch->flash_pm_offset = 0.
25516 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
25517 for rodata if avr_arch->flash_pm_offset != 0.
25518 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
25519 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
25520 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
25521 (m_options): Append opt_rcall.
25522 (m_dirnames): Append dir_rcall.
25523 * config/avr/t-multilib: Regenerate.
25525 * configure.ac [target=avr]: Check whether avrxmega3 default
25526 linker description file works as needed.
25527 * configure: Regenerate.
25528 * doc/avr-mmcu.texi: Regenerate.
25529 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
25530 <__AVR_ARCH__>: Document avrxmega3 and 103.
25531 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
25532 <__AVR_SHORT_CALLS__>: Document it.
25533 <__AVR_PM_BASE_ADDRESS__>: Document it.
25534 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
25535 (AVR Variable Attributes) <progmem>: Document this is
25536 not needed for avrxmega3.
25537 (AVR Named Address Spaces) <__flash>: Dito.
25539 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
25541 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
25543 2017-06-12 Doug Rupp <rupp@adacore.com>
25545 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
25546 Append vxworks-stdint.h to the tm_file list.
25547 * config/vxworks-stdint.h: New file.
25549 2017-06-12 Martin Liska <mliska@suse.cz>
25551 PR tree-optimization/81041
25552 * tree-profile.c (gimple_gen_ic_func_profiler):
25553 Create an extra BB in profile-generate
25554 (gimple_gen_time_profiler): Likewise.
25556 2017-06-12 Jakub Jelinek <jakub@redhat.com>
25558 PR tree-optimization/81003
25559 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
25560 (update_range_test): Use it instead of force_gimple_operand_gsi.
25562 2017-06-12 Richard Biener <rguenther@suse.de>
25564 PR tree-optimization/81053
25565 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
25566 with backedge value not defined in loop. Simplify def stmt
25569 2017-06-11 Tom de Vries <tom@codesourcery.com>
25572 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
25574 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
25575 nvptx_cannot_force_const_mem.
25577 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25579 * opts.c (finish_options): Move test for flag_split_stack after
25580 it has been initialized.
25582 2017-06-11 Jason Merrill <jason@redhat.com>
25584 * tree.h (id_equal): New.
25585 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
25586 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
25587 instead of strcmp of IDENTIFIER_POINTER.
25589 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25591 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
25592 (mark_all_inlined_calls_cdtor): Fix formating.
25593 (inline_transform): Rescale profile before inlining.
25595 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25597 * cgraph.h (cgraph_edge::clone): Update prototype.
25598 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
25599 (cgraph_node::create_clone): Update.
25600 (cgraph_node::create_version_clone): Update.
25601 * tree-inline.c (copy_bb): Update.
25602 (expand_call_inline): Update.
25604 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25606 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
25607 factored out from ...
25608 (rs6000_emit_prologue): ... here.
25610 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25612 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
25613 factored out from ...
25614 (rs6000_emit_prologue): ... here.
25616 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25618 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
25620 (handle_missing_profiles): Fix computation of tp_first_run.
25621 (counts_to_freqs): Do not touch freqs when count is 0.
25623 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25625 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
25628 2017-06-10 Tom de Vries <tom@codesourcery.com>
25630 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
25631 attributes): Document signal effective target.
25633 2017-06-10 Tom de Vries <tom@codesourcery.com>
25635 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
25636 Document effective target stack_size.
25638 2017-06-09 David Malcolm <dmalcolm@redhat.com>
25640 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
25641 to the edit_context if they can be auto-applied.
25643 2017-06-9 Ian Lance Taylor <iant@golang.org>
25645 * opts.c (finish_options): If -fsplit-stack, disable implicit
25646 -forder-blocks-and-partition.
25647 * doc/invoke.texi (Optimize Options): Document that when using
25648 -fsplit-stack -forder-blocks-and-partition is not implicitly
25651 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25653 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
25654 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
25655 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
25656 * builtins.def (abort, trap, unreachable): Declare cold.
25657 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
25658 * tree-core.h (ECF_COLD): New.
25659 * tree.c (set_call_expr_flags): Handle ECF_COLD.
25660 (build_common_builtin_nodes): Mark unreachable and abort as cold.
25662 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25664 * predict.c (unlikely_executed_stmt_p): Cleanup.
25666 2017-06-09 Richard Biener <rguenther@suse.de>
25668 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
25669 model if the ref is always written to.
25671 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25673 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
25675 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25677 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
25680 2017-06-09 Tom de Vries <tom@codesourcery.com>
25683 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
25684 "target cannot support label values" when encountering LABEL_REF.
25686 2017-06-09 Martin Liska <mliska@suse.cz>
25688 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
25689 (gimple_gen_ic_func_profiler): Emit direct comparison
25690 of __gcov_indirect_call_callee with NULL.
25691 (gimple_gen_time_profiler): Change probability from
25692 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
25694 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25696 * profile.c (edge_gcov_counts): Turn to pointer.
25697 (compute_branch_probabilities, compute_branch_probabilities): Update.
25698 (branch_prob): Do not clear edge_gcov_count.
25699 * profile.h (edge_gcov_counts): Turn to pointer.
25700 (edge_gcov_count): Update.
25702 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25704 * gimple.h (gimple_check_failed): Mark cold.
25706 2017-06-09 Richard Biener <rguenther@suse.de>
25708 PR tree-optimization/66623
25709 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
25710 refactor check_reduction into two parts, properly computing
25711 whether we have to check reduction validity for outer loop
25714 2017-06-09 Richard Biener <rguenther@suse.de>
25716 PR tree-optimization/79483
25717 * graphite-scop-detection.c (order): New global.
25718 (get_order): Compute bb to order mapping that satisfies code
25719 generation constraints.
25720 (cmp_pbbs): New helper.
25721 (build_scops): Start domwalk at entry block, sort generated
25724 2017-06-09 Richard Biener <rguenther@suse.de>
25726 PR middle-end/81007
25727 * ipa-polymorphic-call.c
25728 (ipa_polymorphic_call_context::restrict_to_inner_class):
25729 Skip FIELD_DECLs with error_mark_node type.
25730 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
25733 2017-06-09 Martin Liska <mliska@suse.cz>
25735 * predict.c (struct branch_predictor): New struct.
25736 (test_prediction_value_range): New test.
25737 (predict_c_tests): New function.
25738 * selftest-run-tests.c (selftest::run_tests): Run the function.
25739 * selftest.h: Declare new tests.
25741 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
25744 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
25745 gen_add3_insn did not fail.
25746 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
25747 r0, construct that number in a temporary reg and add that reg to r0.
25748 If asked to put the result in r0 as well, fail.
25750 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
25752 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25753 for early expansion of vec_eqv.
25755 2017-06-08 Jakub Jelinek <jakub@redhat.com>
25757 PR middle-end/81005
25758 * ubsan.c (instrument_null): Avoid pointless code temporary.
25759 (pass_ubsan::execute): Instrument aggregate arguments of calls.
25761 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
25765 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
25768 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
25771 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25773 * predict.c (unlikely_executed_edge_p): Move ahead.
25774 (probably_never_executed_edge_p): Use it.
25776 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25778 PR middle-end/79988
25779 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
25780 gimple_call_builtin_p call.
25782 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25784 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
25785 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
25786 rtl_check_failed_type2, rtl_check_failed_code1,
25787 rtl_check_failed_code2, rtl_check_failed_code_mode,
25788 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
25789 rtvec_check_failed_bounds, rtl_check_failed_flag,
25790 _fatal_insn_not_found, _fatal_insn): Likewise.
25791 * tree.h (tree_contains_struct_check_failed,
25792 tree_check_failed, tree_not_check_failed,
25793 tree_class_check_failed, tree_range_check_failed,
25794 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
25795 tree_vec_elt_check_failed, phi_node_elt_check_failed,
25796 tree_operand_check_failed, omp_clause_check_failed,
25797 omp_clause_operand_check_failed, omp_clause_range_check_failed):
25800 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25802 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
25803 flag_branch_probabilities.
25804 * ipa-inline.c (edge_badness): Likewise.
25805 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
25806 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
25807 * predict.c (maybe_hot_frequency_p): Likewise.
25808 (probably_never_executed): Likewise.
25809 * sched-ebb.c (schedule_ebbs): Likewise.
25810 * sched-rgn.c (find_single_block_region): Likewise.
25811 * tracer.c (tail_duplicate): Likewise.
25813 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25815 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
25816 longer requires x_flag_profile_use.
25818 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25820 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
25821 instead of flag_reorder_blocks_and_partition.
25822 * dbxout.c (dbxout_function_end): Likewise.
25823 * dwarf2out.c (gen_subprogram_die): Likewise.
25824 * haifa-sched.c (sched_create_recovery_edges): Likewise.
25825 * hw-doloop.c (reorg_loops): Likewise.
25826 * varasm.c (assemble_start_function,
25827 assemble_end_function): Likewise.
25828 (decide_function_section): Do not check for
25829 flag_reorder_blocks_and_partition.
25831 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25833 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
25835 (chkp_get_hard_register_fake_addr_expr): Ditto.
25836 (chkp_build_addr_expr): Add check for hard reg case.
25837 (chkp_parse_array_and_component_ref): Ditto.
25838 (chkp_find_bounds_1): Ditto.
25839 (chkp_process_stmt): Don't generate bounds store for
25842 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25844 * predict.c (maybe_hot_bb_p): Do not check profile status.
25845 (maybe_hot_edge_p): Likewise.
25846 (probably_never_executed): Check for zero counts even if profile
25848 (unlikely_executed_edge_p): New function.
25849 (unlikely_executed_stmt_p): New function.
25850 (unlikely_executed_bb_p): New function.
25851 (set_even_probabilities): Use unlikely predicates.
25852 (combine_predictions_for_bb): Likewise.
25853 (predict_paths_for_bb): Likewise.
25854 (predict_paths_leading_to_edge): Likewise.
25855 (determine_unlikely_bbs): New function.
25856 (estimate_bb_frequencies): Use it.
25857 (compute_function_frequency): Use zero counts even if profile is
25859 * profile-count.h: Fix typo.
25861 2017-08-08 Julia Koval <julia.koval@intel.com>
25863 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
25864 _mm512_mask_cvtsepi16_storeu_epi8,
25865 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
25866 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
25867 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
25868 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
25869 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
25870 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
25871 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
25872 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
25873 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
25874 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
25875 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
25876 __builtin_ia32_pmovuswb256mem_mask,
25877 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
25878 __builtin_ia32_pmovwb512mem_mask): New builtins.
25880 2017-08-08 Julia Koval <julia.koval@intel.com>
25882 PR target/73350,80862
25883 * config/i386/subst.md (round): Fix round pattern.
25884 * config/i386/i386.c (ix86_erase_embedded_rounding):
25885 Fix erasing rounding for the fixed pattern.
25887 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25889 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
25891 2017-06-08 Martin Liska <mliska@suse.cz>
25893 PR gcov-profile/80911
25894 * gcov.c (block_info::block_info): New constructor.
25896 2017-06-07 Carl Love <cel@us.ibm.com>
25898 * config/rs6000/rs6000-c: The return type of the following
25899 built-in functions was implemented as int not long long. Fix sign
25900 of return value for the unsigned version of vec_mulo and vec_mule.
25901 vector unsigned long long vec_bperm (vector unsigned long long,
25902 vector unsigned char)
25903 vector signed long long vec_mule (vector signed int,
25905 vector unsigned long long vec_mule (vector unsigned int,
25906 vector unsigned int)
25907 vector signed long long vec_mulo (vector signed int,
25909 vector unsigned long long vec_mulo (vector unsigned int,
25910 vector unsigned int)
25911 * doc/extend.texi: Fix the documentation for the built-in
25914 2017-06-07 Carl Love <cel@us.ibm.com>
25917 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
25920 2017-06-07 Carl Love <cel@us.ibm.com>
25922 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
25923 support, Generate doublehv for signed int/float for BE case only.
25925 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
25927 * doc/invoke.texi (mcx16): Rewrite.
25929 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25931 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
25932 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
25933 *mov<mode>_softfloat, and an anonymous splitter): Use
25934 nonimmediate_operand instead of rs6000_nonimmediate_operand.
25936 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25938 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
25940 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25941 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
25942 (rs6000_debug_reg_global): Adjust.
25943 (rs6000_init_hard_regno_mode_ok): Adjust.
25944 (rs6000_dbx_register_number): Adjust.
25945 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
25946 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25947 Remove SPE_ACC and SPEFSCR.
25948 (REG_ALLOC_ORDER): Ditto.
25949 (FRAME_POINTER_REGNUM): Change to 111.
25950 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
25951 (REG_CLASS_NAMES): Ditto.
25952 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
25953 (REGISTER_NAMES): Ditto.
25954 (ADDITIONAL_REG_NAMES): Ditto.
25955 (rs6000_reg_names): Ditto.
25956 * config/rs6000/rs6000.md: Renumber some register number
25959 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25961 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
25963 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25964 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
25966 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
25967 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25968 Delete the SPE high registers.
25969 (REG_ALLOC_ORDER): Ditto.
25970 (enum reg_class): Remove SPE_HIGH_REGS.
25971 (REG_CLASS_NAMES): Ditto.
25972 (REG_CLASS_CONTENTS): Delete the SPE high registers.
25973 (REGISTER_NAMES): Ditto.
25974 (rs6000_reg_names): Ditto.
25975 * doc/tm.texi.in: Remove SPE as example.
25976 * doc/tm.texi: Regenerate.
25978 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25980 * config/rs6000/8540.md (ppc8540_brinc): Delete.
25981 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
25982 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
25983 * config/rs6000/rs6000.md (type): Remove "brinc".
25985 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25987 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
25988 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
25989 * config/rs6000/linuxspe.h: Delete file.
25990 * config/rs6000/rs6000.md: Don't include spe.md.
25991 * config/rs6000/spe.h: Delete file.
25992 * config/rs6000/spe.md: Delete file.
25993 * config/rs6000/t-rs6000: Remove spe.md.
25995 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25997 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
25998 (reg_or_none500mem_operand): Delete.
25999 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
26000 instead of reg_or_none500mem_operand.
26002 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26004 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26005 handling of SPE flags.
26006 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
26008 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26010 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
26012 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
26013 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
26014 paired_divv2sf3): Similar.
26015 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
26016 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
26017 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
26019 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
26020 Rename the paired_* instruction patterns.
26021 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
26023 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
26024 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
26025 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
26026 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
26027 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
26028 PAIRED_VECTOR_MODE.
26029 (struct machine_function): Delete field spe_insn_chain_scanned_p.
26030 (spe_func_has_64bit_regs_p): Delete.
26031 (spe_expand_predicate_builtin): Delete.
26032 (spe_expand_evsel_builtin): Delete.
26033 (TARGET_DWARF_REGISTER_SPAN): Do not define.
26034 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
26035 (invalid_e500_subreg): Delete.
26036 (rs6000_legitimize_address): Always force_reg op2 as well, for
26037 paired single memory accesses.
26038 (rs6000_member_type_forces_blk): Delete.
26039 (rs6000_spe_function_arg): Delete.
26040 (rs6000_expand_unop_builtin): Delete SPE handling.
26041 (rs6000_expand_binop_builtin): Ditto.
26042 (spe_expand_stv_builtin): Delete.
26043 (bdesc_2arg_spe): Delete.
26044 (spe_expand_builtin): Delete.
26045 (spe_expand_predicate_builtin): Delete.
26046 (spe_expand_evsel_builtin): Delete.
26047 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
26048 (spe_init_builtins): Delete.
26049 (spe_func_has_64bit_regs_p): Delete.
26050 (savres_routine_name): Delete "info" parameter. Adjust callers.
26051 (rs6000_emit_stack_reset): Ditto.
26052 (rs6000_dwarf_register_span): Delete.
26053 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
26054 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
26055 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
26057 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
26059 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
26060 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
26061 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
26062 mulv2sf3, divv2sf3): Delete expanders.
26064 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26066 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
26068 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26070 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
26071 * config/rs6000/rs6000.c: Ditto.
26073 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26075 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
26076 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
26077 comparison_operator.
26079 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26081 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
26082 * config/rs6000/rs6000.opt: Ditto.
26083 * config/rs6000/t-rtems: Ditto.
26085 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26087 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
26088 TARGET_E500_SINGLE by 0, simplify.
26089 * config/rs6000/rs6000.c: Ditto.
26090 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
26091 (spe_build_register_parallel): Delete.
26092 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
26093 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
26094 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
26095 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
26096 (E500_CONVERT): Delete.
26097 * config/rs6000/spe.md: Remove many patterns and all define_constants.
26099 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26101 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
26102 * config/rs6000/dfp.md: Ditto.
26103 (negdd2, *negdd2_fpr): Merge.
26104 (absdd2, *absdd2_fpr): Merge.
26105 (negtd2, *negtd2_fpr): Merge.
26106 (abstd2, *abstd2_fpr): Merge.
26107 * config/rs6000/e500.h: Delete file.
26108 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
26109 TARGET_FPRS by 1 and simplify.
26110 * config/rs6000/rs6000-c.c: Ditto.
26111 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
26112 TARGET_DF_SPE by 0.
26113 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
26115 * config/rs6000/rs6000.md: Ditto.
26116 (floatdidf2, *floatdidf2_fpr): Merge.
26117 (move_from_CR_gt_bit): Delete.
26118 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
26119 (E500_CR_IOR_COMPARE): Delete.
26120 (All patterns that require !TARGET_FPRS): Delete.
26121 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
26123 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26125 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
26127 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26129 * graphds.c (add_edge): Intitialize edge's attached data.
26130 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
26131 pointer parameter. Call pointed function on each edge during
26132 graph traversing. Skip traversing the edge when the function
26134 (graphds_dfs, graphds_scc): Ditto.
26135 (for_each_edge): New parameter. Pass the new parameter to callback
26137 * graphds.h (skip_edge_callback): New function pointer type.
26138 (graphds_dfs, graphds_scc): New function pointer parameter.
26139 (graphds_edge_callback, for_each_edge): New parameter.
26141 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26143 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
26144 out code checking if runtime alias check is possible to below ...
26145 Call the new function.
26146 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
26147 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
26149 2017-06-07 Marek Polacek <polacek@redhat.com>
26152 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
26153 TYPE_OVERFLOW_WRAPS check.
26155 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26157 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
26158 if versioning is required.
26159 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
26160 peeling with the check for versioning.
26162 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26164 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
26165 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
26166 Set true to new parameter if new ssa variable is defined.
26167 (vect_gen_vector_loop_niters): Refactor. Set range information
26168 for the new vector loop bound variable.
26169 (vect_do_peeling): Ditto.
26171 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26173 * tree-affine.c (ssa.h): Include header file.
26174 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
26175 has wrapping overflow behavior.
26177 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26179 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
26181 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26183 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
26184 (tree_to_aff_combination): ... here.
26186 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26188 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
26189 reg_pressure model function.
26190 (ivopts_global_cost_for_size): Delete.
26191 (determine_set_costs, iv_ca_recount_cost): Call new model function
26192 ivopts_estimate_reg_pressure.
26194 2017-06-07 Tamar Christina <tamar.christina@arm.com>
26196 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
26197 expensive than udiv. Remove floating point cases from mod.
26199 2017-06-07 Tamar Christina <tamar.christina@arm.com>
26201 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
26202 Increase idiv cost.
26204 2017-06-07 Tamar Christina <tamar.christina@arm.com>
26206 * config/aarch64/aarch64.md
26207 (copysignsf3): Fix mask generation.
26209 2017-06-07 Jakub Jelinek <jakub@redhat.com>
26211 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
26213 (class dump_manager): Add register_dumps method.
26214 * dumpfile.c: Include langhooks.h.
26215 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
26216 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
26217 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
26218 (dump_manager::dump_register): Start with 512 entries instead of 32.
26219 (dump_manager::register_dumps): New method.
26220 * toplev.c (general_init): Instead of invoking register_dumps
26221 langhook, invoke register_dumps method on the dump manager.
26222 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
26225 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
26227 * doc/md.texi: Clarify the restrictions on a define_insn condition.
26228 Say that # requires an associated define_split to exist, and that
26229 the define_split must be suitable for use after register allocation.
26231 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26233 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
26234 (compute_outgoing_frequencies): Also initialize zero counts.
26235 (find_many_sub_basic_blocks): Do not produce uninitialized profile
26236 around loops; preserve more of profile when nothing changes.
26238 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
26240 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
26242 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
26243 * config/arm/arm-cpu-cdata.h: Regenerate.
26244 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
26245 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
26246 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
26247 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
26248 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
26250 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
26252 * config/arm/t-rmprofile: Likewise.
26253 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
26255 2017-06-06 David S. Miller <davem@davemloft.net>
26258 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
26259 blockage if function uses alloca.
26261 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26263 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
26264 New "uid" fields to hold pretty-print IDs of group and ref.
26265 Memory references are now identified as <group_id>:<ref_id>
26266 instead of using [random] addresses.
26267 (dump_mem_details): Simplify, no functional change.
26268 (dump_mem_ref): Simplify and make output more concise.
26269 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
26270 (find_or_create_group): Initialize group uid.
26271 (record_ref): Initialize ref uid. Improve debug output.
26272 (prune_group_by_reuse, should_issue_prefetch_p,)
26273 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
26274 (mark_nontemporal_store, determine_loop_nest_reuse):
26275 Improve debug output.
26277 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26279 * dbgcnt.def (prefetch): New debug counter.
26280 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
26281 (schedule_prefetches): Stop issueing prefetches if debug counter
26284 2017-06-06 Tom de Vries <tom@codesourcery.com>
26286 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
26287 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
26289 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26291 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
26292 Use aarch64_reg_or_zero predicate for operand 4.
26293 (aarch64_compare_and_swap<mode> define_insn_and_split):
26294 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
26295 (aarch64_store_exclusive<mode>): Likewise for operand 2.
26297 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
26299 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
26300 (arm_compute_save_core_reg_mask): This.
26301 (thumb1_compute_save_reg_mask): Rename into ...
26302 (thumb1_compute_save_core_reg_mask): This.
26303 (arm_compute_save_reg0_reg12_mask): Adapt comment.
26304 (arm_compute_frame_layout): Likewise.
26306 2017-06-06 Richard Biener <rguenther@suse.de>
26308 PR tree-optimization/80974
26309 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
26310 keep or clear leaders SSA info.
26312 2017-06-06 Tom de Vries <tom@codesourcery.com>
26314 * config/nvptx/nvptx.c (split_mode_p): New function.
26315 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
26317 2017-06-06 Tom de Vries <tom@codesourcery.com>
26319 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
26321 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26324 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
26327 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26329 * shrink-wrap.c (handle_simple_exit): Update profile.
26330 (try_shrink_wrapping): Upate profile.
26332 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26334 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
26335 (tree_guess_outgoing_edge_probabilities): New.
26336 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
26337 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
26339 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26341 * ipa-split.c (split_function): Initialize return bb profile.
26343 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26345 * profile.c (compute_branch_probabilities): Also initialize
26346 EXIT_BLOCK profile.
26348 2017-06-06 Richard Biener <rguenther@suse.de>
26350 PR tree-optimization/80928
26351 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
26352 (vect_analyze_loop_operations): Properly guard analysis for
26354 (vect_transform_loop): Likewise.
26355 (vect_analyze_loop_2): Also reset SLP type on PHIs.
26356 (vect_model_induction_cost): Do not cost for pure SLP.
26357 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
26358 of induction in inner loop vectorization.
26359 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
26360 (vect_get_and_check_slp_defs): Handle vect_induction_def.
26361 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
26363 (vect_analyze_slp_cost_1): Cost induction.
26364 (vect_detect_hybrid_slp_stmts): Handle PHIs.
26365 (vect_get_slp_vect_defs): Likewise.
26366 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
26367 (vect_transform_stmt): Handle SLP reductions.
26368 * tree-vectorizer.h (vectorizable_induction): Adjust.
26370 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
26372 * config/rs6000/rs6000.c (make_resolver_func): Update
26373 init_lowered_empty_function call.
26375 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
26377 * doc/invoke.texi: Document the -fprofile-abs-path option.
26378 * common.opt (fprofile-abs-path): New option.
26379 * gcov-io.h (gcov_write_filename): Declare.
26380 * gcov-io.c (gcov_write_filename): New function.
26381 * coverage.c (coverage_begin_function): Use gcov_write_filename.
26382 * profile.c (output_location): Likewise.
26384 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26386 * shring-wrap.c: Revert accidental commit.
26388 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
26390 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
26392 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26394 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
26396 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
26397 profile in callgraph edge.
26398 * profile-count.h (apply_probability): If THIS is 0, then result is 0
26399 (apply_scale): Likewise.
26400 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
26401 Also scale profile when inlining function with zero profile.
26402 (initialize_cfun): Update exit block profile even when it is zero.
26403 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
26404 when profile is read.
26406 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
26408 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
26409 (CLONE_*): New constants to define the processors we can generate
26410 code for with the target_clone attribute.
26411 (rs6000_clone_map): New array to identify which clone processors
26412 the current program is running on.
26413 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
26414 target_clone attribute.
26415 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
26416 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
26417 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
26418 (cpu_expand_builtin): Add support for target_clone attribute.
26419 (rs6000_valid_attribute_p): Allow "default" attribute.
26420 (get_decl_name): New debug function to simplify printing the
26421 current function name in debugging statements.
26422 (rs6000_clone_priority): New functions to support the target_clone
26423 attribute, and be able to generate code to switch between ISA 2.05
26424 through ISA 3.0 (power6 through power9).
26425 (rs6000_compare_version_priority): Likewise.
26426 (rs6000_get_function_versions_dispatcher): Likewise.
26427 (make_resolver_func): Likewise.
26428 (add_condition_to_bb): Likewise.
26429 (dispatch_function_versions): Likewise.
26430 (rs6000_generate_version_dispatcher_body): Likewise.
26431 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
26432 (fusion_gpr_load_p): Fix a spacing issue.
26433 * doc/extend.texi (Common Function Attributes): Document that the
26434 PowerPC supports the target_clone attribute.
26436 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
26438 * config/arm/arm.h: explain F symbol found in description of ARM
26439 register allocation in its legend.
26441 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26443 * config/mips/frame-header-opt.c: Include profile-count.h.
26444 * config/riscv/riscv.c: Include profile-count.h
26446 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26448 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
26450 (sm_set_flag_if_changed): Add bbs field.
26451 (execute_sm_if_changed_flag_set): Pass BBS.
26452 (execute_sm): Update.
26454 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26456 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
26459 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26461 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
26462 (peephole2): New peephole2 to emit the above.
26463 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
26465 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26467 * config/aarch64/aarch64.c (define_peephole2 above
26468 *sub_<shift>_<mode>): New peephole.
26470 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26472 * config/i386/i386.c (make_resolver_func): Update.
26473 * Makefile.in: Add profile-count.h and profile-count.o
26474 * auto-profile.c (afdo_indirect_call): Update to new API.
26475 (afdo_set_bb_count): Update.
26476 (afdo_propagate_edge): Update.
26477 (afdo_propagate_circuit): Update.
26478 (afdo_calculate_branch_prob): Update.
26479 (afdo_annotate_cfg): Update.
26480 * basic-block.h: Include profile-count.h
26481 (struct edge_def): Turn count to profile_count.
26482 (struct basic_block_def): Likewie.
26483 (REG_BR_PROB_BASE): Move to profile-count.h
26484 (RDIV): Move to profile-count.h
26485 * bb-reorder.c (max_entry_count): Turn to profile_count.
26486 (find_traces): Update.
26487 (rotate_loop):Update.
26488 (connect_traces):Update.
26489 (sanitize_hot_paths):Update.
26490 * bt-load.c (migrate_btr_defs): Update.
26491 * cfg.c (RDIV): Remove.
26492 (init_flow): Use alloc_block.
26493 (alloc_block): Uninitialize count.
26494 (unchecked_make_edge): Uninitialize count.
26495 (check_bb_profile): Update.
26496 (dump_edge_info): Update.
26497 (dump_bb_info): Update.
26498 (update_bb_profile_for_threading): Update.
26499 (scale_bbs_frequencies_int): Update.
26500 (scale_bbs_frequencies_gcov_type): Update.
26501 (scale_bbs_frequencies_profile_count): New.
26502 * cfg.h (update_bb_profile_for_threading): Update.
26503 (scale_bbs_frequencies_profile_count): Declare.
26504 * cfgbuild.c (compute_outgoing_frequencies): Update.
26505 (find_many_sub_basic_blocks): Update.
26506 * cfgcleanup.c (try_forward_edges): Update.
26507 (try_crossjump_to_edge): Update.
26508 * cfgexpand.c (expand_gimple_tailcall): Update.
26509 (construct_exit_block): Update.
26510 * cfghooks.c (verify_flow_info): Update.
26511 (dump_bb_for_graph): Update.
26512 (split_edge): Update.
26513 (make_forwarder_block): Update.
26514 (duplicate_block): Update.
26515 (account_profile_record): Update.
26516 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
26517 (get_estimated_loop_iterations): Update.
26518 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
26519 (single_likely_exit): Update.
26520 * cfgloopmanip.c (scale_loop_profile): Update.
26522 (set_zero_probability): Update.
26523 (lv_adjust_loop_entry_edge): Update.
26524 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
26525 (purge_dead_edges): Update.
26526 (rtl_account_profile_record): Update.
26527 * cgraph.c (cgraph_node::create): Uninitialize count.
26528 (symbol_table::create_edge): Uninitialize count.
26529 (cgraph_update_edges_for_call_stmt_node): Update.
26530 (cgraph_edge::dump_edge_flags): Update.
26531 (cgraph_node::dump): Update.
26532 (cgraph_edge::maybe_hot_p): Update.
26533 * cgraph.h: Include profile-count.h
26534 (create_clone), create_edge, create_indirect_edge): Update.
26535 (cgraph_node): Turn count to profile_count.
26536 (cgraph_edge0: Likewise.
26537 (make_speculative, clone): Update.
26538 (create_edge): Update.
26539 (init_lowered_empty_function): Update.
26540 * cgraphclones.c (cgraph_edge::clone): Update.
26541 (duplicate_thunk_for_node): Update.
26542 (cgraph_node::create_clone): Update.
26543 * cgraphunit.c (cgraph_node::analyze): Update.
26544 (cgraph_node::expand_thunk): Update.
26545 * final.c (dump_basic_block_info): Update.
26546 * gimple-streamer-in.c (input_bb): Update.
26547 * gimple-streamer-out.c (output_bb): Update.
26548 * graphite.c (print_global_statistics): Update.
26549 (print_graphite_scop_statistics): Update.
26550 * hsa-brig.c: Include basic-block.h.
26551 * hsa-dump.c: Include basic-block.h.
26552 * hsa-gen.c (T sum_slice): Update.
26553 (convert_switch_statements):Update.
26554 * hsa-regalloc.c: Include basic-block.h.
26555 * ipa-chkp.c (chkp_produce_thunks): Update.
26556 * ipa-cp.c (struct caller_statistics): Update.
26557 (init_caller_stats): Update.
26558 (gather_caller_stats): Update.
26559 (ipcp_cloning_candidate_p): Update.
26560 (good_cloning_opportunity_p): Update.
26561 (get_info_about_necessary_edges): Update.
26562 (dump_profile_updates): Update.
26563 (update_profiling_info): Update.
26564 (update_specialized_profile): Update.
26565 (perhaps_add_new_callers): Update.
26566 (decide_about_value): Update.
26567 (ipa_cp_c_finalize): Update.
26568 * ipa-devirt.c (struct odr_type_warn_count): Update.
26569 (struct decl_warn_count): Update.
26570 (struct final_warning_record): Update.
26571 (possible_polymorphic_call_targets): Update.
26572 (ipa_devirt): Update.
26573 * ipa-fnsummary.c (redirect_to_unreachable): Update.
26574 * ipa-icf.c (sem_function::merge): Update.
26575 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
26576 * ipa-inline.c (compute_uninlined_call_time): Update.
26577 (compute_inlined_call_time): Update.
26578 (want_inline_small_function_p): Update.
26579 (want_inline_self_recursive_call_p): Update.
26580 (edge_badness): Update.
26581 (lookup_recursive_calls): Update.
26582 (recursive_inlining): Update.
26583 (inline_small_functions): Update.
26584 (dump_overall_stats): Update.
26585 (dump_inline_stats): Update.
26586 * ipa-profile.c (ipa_profile_generate_summary): Update.
26587 (ipa_propagate_frequency): Update.
26588 (ipa_profile): Update.
26589 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26590 * ipa-utils.c (ipa_merge_profiles): Update.
26591 * loop-doloop.c (doloop_modify): Update.
26592 * loop-unroll.c (report_unroll): Update.
26593 (unroll_loop_runtime_iterations): Update.
26594 * lto-cgraph.c (lto_output_edge): Update.
26595 (lto_output_node): Update.
26596 (input_node): Update.
26597 (input_edge): Update.
26598 (merge_profile_summaries): Update.
26599 * lto-streamer-in.c (input_cfg): Update.
26600 * lto-streamer-out.c (output_cfg): Update.
26601 * mcf.c (create_fixup_graph): Update.
26602 (adjust_cfg_counts): Update.
26603 (sum_edge_counts): Update.
26604 * modulo-sched.c (sms_schedule): Update.
26605 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
26606 * predict.c (maybe_hot_count_p): Update.
26607 (probably_never_executed): Update.
26608 (dump_prediction): Update.
26609 (combine_predictions_for_bb): Update.
26610 (propagate_freq): Update.
26611 (handle_missing_profiles): Update.
26612 (counts_to_freqs): Update.
26613 (rebuild_frequencies): Update.
26614 (force_edge_cold): Update.
26615 * predict.h: Include profile-count.h
26616 (maybe_hot_count_p, counts_to_freqs): UPdate.
26617 * print-rtl-function.c: Do not include cfg.h
26618 * print-rtl.c: Include basic-block.h
26619 * profile-count.c: New file.
26620 * profile-count.h: New file.
26621 * profile.c (is_edge_inconsistent): Update.
26622 (correct_negative_edge_counts): Update.
26623 (is_inconsistent): Update.
26624 (set_bb_counts): Update.
26625 (read_profile_edge_counts): Update.
26626 (compute_frequency_overlap): Update.
26627 (compute_branch_probabilities): Update; Initialize and deinitialize
26629 (branch_prob): Update.
26630 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
26631 (edge_gcov_count): New.
26632 (bb_gcov_count): New.
26633 * shrink-wrap.c (try_shrink_wrapping): Update.
26634 * tracer.c (better_p): Update.
26635 * trans-mem.c (expand_transaction): Update.
26636 (ipa_tm_insert_irr_call): Update.
26637 (ipa_tm_insert_gettmclone_call): Update.
26638 * tree-call-cdce.c: Update.
26639 * tree-cfg.c (gimple_duplicate_sese_region): Update.
26640 (gimple_duplicate_sese_tail): Update.
26641 (gimple_account_profile_record): Update.
26642 (execute_fixup_cfg): Update.
26643 * tree-inline.c (copy_bb): Update.
26644 (copy_edges_for_bb): Update.
26645 (initialize_cfun): Update.
26646 (freqs_to_counts): Update.
26647 (copy_cfg_body): Update.
26648 (expand_call_inline): Update.
26649 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26650 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
26651 (try_unroll_loop_completely): Update.
26652 (try_peel_loop): Update.
26653 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26654 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
26655 * tree-ssa-loop-split.c (connect_loops): Update.
26656 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
26657 * tree-ssa-reassoc.c (branch_fixup): Update.
26658 * tree-ssa-tail-merge.c (replace_block_by): Update.
26659 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
26660 (compute_path_counts): Update.
26661 (update_profile): Update.
26662 (recompute_probabilities): Update.
26663 (update_joiner_offpath_counts): Update.
26664 (estimated_freqs_path): Update.
26665 (freqs_to_counts_path): Update.
26666 (clear_counts_path): Update.
26667 (ssa_fix_duplicate_block_edges): Update.
26668 (duplicate_thread_path): Update.
26669 * tree-switch-conversion.c (case_bit_test_cmp): Update.
26670 (struct switch_conv_info): Update.
26671 * tree-tailcall.c (decrease_profile): Update.
26672 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26673 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26674 * value-prof.c (check_counter): Update.
26675 (gimple_divmod_fixed_value): Update.
26676 (gimple_mod_pow2): Update.
26677 (gimple_mod_subtract): Update.
26678 (gimple_ic_transform): Update.
26679 (gimple_stringop_fixed_value): Update.
26680 * value-prof.h (gimple_ic): Update.
26682 2017-06-02 Carl Love <cel@us.ibm.com>
26684 * config/rs6000/rs6000-c: Add support for built-in functions
26685 vector double vec_doublee (vector signed int);
26686 vector double vec_doublee (vector unsigned int);
26687 vector double vec_doublee (vector float);
26688 vector double vec_doubleh (vector signed int);
26689 vector double vec_doubleh (vector unsigned int);
26690 vector double vec_doubleh (vector float);
26691 vector double vec_doublel (vector signed int);
26692 vector double vec_doublel (vector unsigned int);
26693 vector double vec_doublel (vector float);
26694 vector double vec_doubleo (vector signed int);
26695 vector double vec_doubleo (vector unsigned int);
26696 vector double vec_doubleo (vector float);.
26697 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
26698 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
26700 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
26701 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
26702 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
26704 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
26705 vec_doublel, vec_doubleh.
26706 * doc/extend.texi: Update the built-in documentation file for the
26707 new built-in functions.
26709 2017-06-02 David Malcolm <dmalcolm@redhat.com>
26712 * ipa-inline-analysis.c (free_growth_caches): Set
26713 edge_removal_hook_holder to NULL after removing it.
26715 2017-06-02 Sudakshina Das <sudi.das@arm.com>
26717 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
26718 comparision with zero.
26720 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26721 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26722 for early expansion of vec_min and vec_max builtins.
26723 (builtin_function_type): Add min/max unsigned variants to those
26724 identified as having unsigned arguments.
26726 2017-06-02 Olivier Hainque <hainque@adacore.com>
26728 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
26730 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26732 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
26733 Use VALL_F16 iterator rather than VALL.
26735 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26737 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
26738 Emit CBNZ inside loop when doing a strong exchange and comparing
26739 against zero. Generate the CC flags after the loop.
26741 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
26743 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
26744 (dl_section_ref): New.
26745 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
26746 On AIX, append an expression to subtract the size of the
26747 section length to dl_section_ref.
26749 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26751 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26752 for early expansion of vector absolute builtins.
26754 2017-06-02 Richard Biener <rguenther@suse.de>
26756 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
26757 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
26759 2017-06-02 Richard Biener <rguenther@suse.de>
26761 PR tree-optimization/80948
26762 * tree-tailcall.c (find_tail_calls): Track stmts to move in
26763 stmt order as well.
26765 2017-06-02 Richard Biener <rguenther@suse.de>
26767 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
26769 * tree-vect-stmts.c (process_use): Do not mark backedge defs
26770 for inductions as relevant.
26772 2017-06-02 Richard Biener <rguenther@suse.de>
26774 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
26775 (vectorizable_induction): ... this. Remove dead code.
26777 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
26779 * builtins. (expand_builtin_alloca): Remove second parameter and
26780 infer its value from the first parameter instead.
26781 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
26783 2017-06-02 Jakub Jelinek <jakub@redhat.com>
26785 PR rtl-optimization/80903
26786 * loop-doloop.c (add_test): Unshare sequence.
26788 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
26790 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
26792 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
26794 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
26796 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
26797 xlogue_layout::get_instance, logue_layout::xlogue_layout,
26798 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
26799 (xlogue_layout::get_stub_rtx): Make static.
26800 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
26801 (xlogue_layout::compute_stub_managed_regs): Rename to...
26802 (xlogue_layout::count_stub_managed_regs): ...this.
26803 (xlogue_layout::is_stub_managed_reg): New function.
26804 (xlogue_layout::m_stub_names): Rename to...
26805 (xlogue_layout::s_stub_names): ...this, make static.
26806 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
26807 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
26808 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
26809 xlogue_layout::s_stub_names): Instantiate statics.
26810 (stub_managed_regs): Remove.
26811 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
26812 (disable_call_ms2sysv_xlogues): Rename to...
26813 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
26814 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
26816 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
26817 change after reload_completed.
26818 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
26820 (ix86_expand_prologue): Likewise.
26821 (ix86_expand_epilogue): Likewise.
26822 (ix86_expand_split_stack_prologue): Likewise.
26823 (ix86_compute_frame_layout): Remove frame parameter ...
26824 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
26825 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
26827 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
26828 (ix86_frame): Move from here ...
26829 * config/i386/i386.h (ix86_frame): ... to here.
26830 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
26831 complete ix86_frame data structure instead. Remove some_ld_name.
26833 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26835 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
26836 symbols that hold a DECL_VALUE_EXPR.
26838 2017-06-01 Martin Jambor <mjambor@suse.cz>
26840 PR tree-optimization/80898
26841 * tree-sra.c (process_subtree_disqualification): Removed.
26842 (disqualify_candidate): Do not acll
26843 process_subtree_disqualification.
26844 (subtree_mark_written_and_enqueue): New function.
26845 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
26846 RHS has been disqualified and re-queue LHS if necessary. Apart
26847 from that, ignore disqualified RHS.
26849 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26851 * config/s390/s390.c (s390_emit_epilogue): Disable early return
26852 address fetch for z10 or later.
26854 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26856 * config/arc/arc.md (tst_movb): Add guard when splitting.
26858 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26860 * config/arc/arc.c (arc_can_eliminate): Test against
26861 arc_frame_pointer_needed.
26863 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26865 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
26866 to prevent store reordering.
26867 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
26868 (type): Add block type.
26869 (stack_tie): Define special instruction to be used in
26872 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26874 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
26875 constraint. It is not valid for the pattern.
26876 (noncommutative_binary_comparison): Likewise.
26878 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26880 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
26883 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26885 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
26886 be used by the reg-alloc.
26888 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26890 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
26891 reg-alloc when having mul64 or mul32x16 instructions.
26892 (mulsidi3): Likewise.
26893 (umulsidi3): Likewise.
26894 (mulsi32x16): New pattern.
26895 (mulsi64): Likewise.
26896 (mulsidi64): Likewise.
26897 (umulsidi64): Likewise.
26898 (MUL32x16_REG): Define.
26899 (mul64_600): Use MUL32x16_REG.
26900 (mac64_600): Likewise.
26901 (umul64_600): Likewise.
26902 (umac64_600): Likewise.
26904 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26906 * config/arc/arc.md (mulsi3_700): Make it commutative.
26908 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
26910 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
26912 (*sign_extendsidi2_insn): Likewise for movstosw.
26914 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26916 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
26917 the type of the input discriminant value. Convert the
26918 discriminant value of signedness vary.
26920 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
26922 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
26923 Add to -Wall section.
26925 2017-06-01 Richard Biener <rguenther@suse.de>
26927 PR middle-end/66313
26928 * fold-const.c (fold_plusminus_mult_expr): If the factored
26929 factor may be zero use a wrapping type for the inner operation.
26930 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
26931 and handle moved defs.
26932 (process_assignment): Properly guard the unary op case. Return a
26933 tri-state indicating that moving the stmt before the call may allow
26934 to continue. Pass through to_move.
26935 (find_tail_calls): Handle moving unrelated defs before
26938 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
26941 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
26942 splitter result in the canonical way.
26944 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
26946 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
26947 also for 32bit target. Update insn attributes.
26948 (zero-extendsidi2 splitter): Allow all registers for operand 1.
26950 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
26952 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
26953 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
26954 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
26955 (_mm_maskz_min_ss): New intrinsics.
26957 2017-05-31 Martin Liska <mliska@suse.cz>
26959 * tree-vect-loop.c (vect_create_epilog_for_reduction):
26960 Change comment style to one we normally use.
26961 (vectorizable_reduction): Likewise.
26962 (vectorizable_induction): Likewise.
26963 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
26964 (vectorizable_call): Likewise.
26965 (vectorizable_simd_clone_call): Likewise.
26966 (vectorizable_conversion): Likewise.
26967 (vectorizable_assignment): Likewise.
26968 (vectorizable_shift): Likewise.
26969 (vectorizable_operation): Likewise.
26970 (vectorizable_store): Likewise.
26971 (vectorizable_load): Likewise.
26972 * tree-vectorizer.h: Likewise.
26974 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
26976 * passes.c (emergency_dump_function): New.
26977 * tree-pass.h (emergency_dump_function): Declare.
26978 * plugin.c (plugins_internal_error_function): Remove.
26979 * plugin.h (plugins_internal_error_function): Remove declaration.
26980 * toplev.c (internal_error_function): New static function. Use it...
26981 (general_init): ...here.
26983 2017-05-31 Graham Markall <graham.markall@embecosm.com>
26985 * config/arc/arc.c (arc_print_operand): Handle constant operands.
26986 (arc_rtx_costs): Add costs for new patterns.
26987 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
26988 * config/arc/predicates.md: Add _1_2_3_operand predicate.
26990 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26992 * tree-ssa-strlen.c (get_next_strinfo): New function.
26993 (get_stridx_plus_constant): Use it.
26994 (zero_length_string): Likewise.
26995 (adjust_related_strinfos): Likewise.
26996 (adjust_last_stmt): Likewise.
26998 2017-05-31 Richard Biener <rguenther@suse.de>
27001 * config/i386/i386.c (ix86_expand_builtin): Remove assert
27002 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
27004 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
27006 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
27007 loop_vinfo argument and use of dependence distance vectors.
27008 Check instead whether the two references differ only in their
27009 initial value and assume that they have the same alignment if the
27010 difference is a multiple of the vector alignment.
27011 (vect_analyze_data_refs_alignment): Update call accordingly.
27013 2017-05-31 Martin Liska <mliska@suse.cz>
27016 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
27018 2017-05-31 Bin Cheng <bin.cheng@arm.com>
27020 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
27021 (create_intersect_range_checks): Move from ...
27022 * tree-data-ref.c (create_intersect_range_checks_index)
27023 (create_intersect_range_checks): ... to here.
27024 (create_runtime_alias_checks): New function factored from ...
27025 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
27026 here. Call above function.
27027 * tree-data-ref.h (create_runtime_alias_checks): New function.
27029 2017-05-31 Bin Cheng <bin.cheng@arm.com>
27031 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
27032 segment length for dr_b and compute it in wide_int.
27034 2017-05-31 Richard Biener <rguenther@suse.de>
27036 PR tree-optimization/80906
27037 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
27038 and pass through iv_map.
27039 (copy_bb_and_scalar_dependences): Adjust.
27040 (translate_pending_phi_nodes): Likewise.
27041 (copy_loop_close_phi_args): Handle code-generating IVs instead
27044 2017-05-30 David Malcolm <dmalcolm@redhat.com>
27046 * diagnostic-color.c (color_dict): Add "type-diff".
27047 (parse_gcc_colors): Update comment.
27048 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
27049 -fdiagnostics-show-template-tree and -fno-elide-type.
27050 (GCC_COLORS): Add type-diff to example.
27052 (-fdiagnostics-show-template-tree): New.
27053 (-fno-elide-type): New.
27054 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
27055 the pp_format_decoder callback. Call any m_format_postprocessor's
27057 (pretty_printer::pretty_printer): Initialize
27058 m_format_postprocessor.
27059 (pretty_printer::~pretty_printer): Delete any
27060 m_format_postprocessor.
27061 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
27062 (class format_postprocessor): New class.
27063 (struct pretty_printer::format_decoder): Document the new parameters.
27064 (struct pretty_printer::m_format_postprocessor): New field.
27065 * tree-diagnostic.c (default_tree_printer): Update for new
27066 bool and const char ** params.
27067 * tree-diagnostic.h (default_tree_printer): Likewise.
27069 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
27071 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
27072 (lwa_operand): Delete rs6000_gen_cell_microcode test.
27073 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27074 rs6000_gen_cell_microcode code.
27075 (rs6000_final_prescan_insn): Delete.
27076 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
27077 "warn-cell-microcode" entries.
27078 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
27079 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
27080 throughout. Change cc_reg_not_micro_cr0_operand to
27081 cc_reg_not_cr0_operand throughout.
27082 (*extendhi<mode>2_noload): Delete.
27083 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
27084 (mwarn-cell-microcode): Delete.
27085 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
27086 -mgen-cell-microcode and -mwarn-cell-microcode.
27088 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
27091 * config/i386/constraints.md (Yd): New constraint.
27093 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
27094 and (?Yd, r) alternatives. Update insn attributes.
27095 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
27096 and (?*Yd, r) alternatives. Update insn attributes.
27097 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
27099 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
27101 * gimplify.c (gimplify_modify_expr): Don't create a
27102 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
27105 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
27107 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
27109 2017-05-30 Richard Biener <rguenther@suse.de>
27111 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
27112 and reduc_def fields.
27113 (STMT_VINFO_REDUC_TYPE): New define.
27114 (STMT_VINFO_REDUC_DEF): Likewise.
27115 (vect_force_simple_reduction): Adjust prototype.
27116 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
27117 (vect_is_simple_reduction): Remove check_reduction argument.
27118 (vect_force_simple_reduction): Adjust and set
27119 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
27120 (vectorizable_reduction): Do not re-do reduction analysis
27121 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
27122 * tree-parloops.c (gather_scalar_reductions): Adjust.
27124 2017-05-30 Richard Biener <rguenther@suse.de>
27126 PR middle-end/80901
27127 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
27130 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
27132 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
27133 Introduce unknown_misalignment parameter and remove vf.
27134 (vect_peeling_hash_get_lowest_cost):
27135 Pass unknown_misalignment parameter.
27136 (vect_enhance_data_refs_alignment):
27137 Fix unsupportable data ref treatment.
27139 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27141 * tree-vect-data-refs.c (vect_get_data_access_cost):
27142 Workaround for SLP handling.
27143 (vect_enhance_data_refs_alignment):
27144 Compute costs for doing no peeling at all, compare to the best
27145 peeling costs so far and avoid peeling if cheaper.
27147 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27149 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
27150 Return peeling info and set costs to zero for unlimited cost
27152 (vect_enhance_data_refs_alignment): Also inspect all datarefs
27153 with unknown misalignment. Compute and costs for unknown
27154 misalignment, compare them to the costs for known misalignment
27155 and choose the cheapest for peeling.
27157 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27159 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
27160 (vect_get_peeling_costs_all_drs): Create function.
27161 (vect_peeling_hash_get_lowest_cost):
27162 Use vect_get_peeling_costs_all_drs.
27163 (vect_peeling_supportable): Create function.
27164 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
27166 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27168 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
27169 DR_HAS_NEGATIVE_STEP.
27170 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
27171 (vect_enhance_data_refs_alignment): Use.
27172 (vect_duplicate_ssa_name_ptr_info): Use.
27173 * tree-vectorizer.h (dr_misalignment): Use.
27174 (known_alignment_for_access_p): Use.
27176 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27179 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
27181 (has_section_name): New function.
27183 2017-05-30 Martin Liska <mliska@suse.cz>
27186 * auto-profile.c (get_function_decl_from_block): Fix
27189 2017-05-30 Richard Biener <rguenther@suse.de>
27191 PR middle-end/80876
27192 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
27194 2017-05-30 Martin Liska <mliska@suse.cz>
27196 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
27197 * dumpfile.h (struct dump_file_info): Remove ctors.
27199 2017-05-30 Martin Liska <mliska@suse.cz>
27201 * predict.def: Fix GNU coding style.
27203 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
27205 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
27206 Mark 'to' argument with ATTRIBUTE_UNUSED.
27208 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
27210 * config/xtensa/xtensa.c (xtensa_emit_call): Use
27211 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
27212 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
27215 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
27217 * doc/install.texi (Options specification): Restore entry of
27218 --enable-sjlj-exceptions.
27220 2017-05-27 Michael Eager <eager@eagercon.com>
27223 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
27225 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
27227 * config/microblaze/microblaze.h
27228 (FIXED_REGISTERS): Update in macro.
27229 (CALL_USED_REGISTERS): Update in macro.
27231 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
27233 * doc/install.texi: Add links to macOS binary distributions.
27235 2017-05-27 Jakub Jelinek <jakub@redhat.com>
27239 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27241 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
27243 2017-05-26 Martin Liska <mliska@suse.cz>
27245 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
27247 2017-05-26 Martin Liska <mliska@suse.cz>
27249 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
27250 always leading ';; '.
27251 (dump_bb_info): Likewise.
27252 (brief_dump_cfg): Likewise.
27253 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
27254 * dumpfile.c: Remove usage of TDF_VERBOSE.
27255 * dumpfile.h (enum dump_kind): Likewise.
27256 (dump_gimple_bb_header): Do not use TDF_COMMENT.
27257 * print-tree.c (debug_verbose): Remove.
27258 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
27259 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
27260 * tree-diagnostic.c (default_tree_printer): Replace
27261 TDF_DIAGNOSTIC with TDF_SLIM.
27263 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27265 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
27266 in parameter loop, rather than loop_vinfo.
27267 (create_intersect_range_checks): Ditto.
27268 (vect_create_cond_for_alias_checks): Update call to above functions.
27270 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27272 PR tree-optimization/80815
27273 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
27274 for merging runtime alias checks. Handle negative DR_STEPs.
27276 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27278 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
27280 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
27281 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
27282 out code pruning runtime alias checks.
27283 * tree-data-ref.c (prune_runtime_alias_test_list): New function
27284 factored out from above.
27285 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
27287 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
27289 (prune_runtime_alias_test_list): New decalaration.
27291 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27293 * tree-vect-data-refs.c (compare_tree): Rename and move ...
27294 * tree-data-ref.c (data_ref_compare_tree): ... to here.
27295 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
27296 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
27297 (operator==, comp_dr_with_seg_len_pair): Ditto.
27298 (vect_prune_runtime_alias_test_list): Ditto.
27300 2017-05-26 Martin Liska <mliska@suse.cz>
27303 * params.def: Bound partial-inlining-entry-probability param.
27305 2017-05-26 Marek Polacek <polacek@redhat.com>
27308 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
27311 2017-05-26 Richard Biener <rguenther@suse.de>
27313 PR tree-optimization/80842
27314 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
27317 2017-05-26 Richard Biener <rguenther@suse.de>
27319 PR tree-optimization/80844
27320 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
27322 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
27324 * doc/md.texi (Machine Constraints): Update x86 family
27325 machine constraints section to match 'config/i386/constraints.md'.
27327 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
27329 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
27331 2017-05-25 Nathan Sidwell <nathan@acm.org>
27333 * doc/invoke.texi (--enable-languages): Update documentation.
27335 2017-05-25 Martin Liska <mliska@suse.cz>
27337 * dumpfile.c: Add TDF_FOLDING.
27338 * dumpfile.h (enum dump_kind): Likewise.
27339 * genmatch.c (dt_simplify::gen_1): Use it.
27341 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27343 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
27345 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27347 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
27348 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
27350 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27352 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
27353 * match.pd (X == C): Rewrite it here.
27354 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
27355 with_certain_nonzero_bits2): New predicates.
27356 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
27358 2017-05-24 Nathan Sidwell <nathan@acm.org>
27360 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
27363 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
27366 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
27368 * config/powerpcspe: New port. Files are copied from the rs6000
27369 port, with "rs6000" in filenames replaced by "powerpcspe".
27371 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
27373 PR rtl-optimization/80754
27374 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
27376 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
27378 * config/sparc/sparc.md (length): Return the correct value for -mflat
27379 sibcalls to match output_sibcall.
27381 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
27385 * config/rs6000/rs6000.c (struct machine_function): Add new field
27387 (rs6000_get_separate_components): Init that field, use it.
27388 (rs6000_components_for_bb): Use the field.
27390 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
27392 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
27394 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
27396 PR middle-end/80823
27397 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
27399 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27402 * config/s390/s390.c (s390_check_qrst_address): Check incoming
27403 address against address_operand predicate.
27404 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
27406 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
27408 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
27409 parameters passed indirectly.
27411 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
27413 * config/i386/i386.md (*movdi_internal): Remove SSE4
27414 alternative 18 (?r, *v). Update insn attributes.
27415 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
27416 Update insn attributes.
27417 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
27418 Update insn attributes.
27419 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
27420 alternative 1 (r, v). Remove isa attribute.
27421 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
27422 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
27423 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
27425 2017-05-23 Tom de Vries <tom@codesourcery.com>
27427 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
27430 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27432 * cgraphunit.c (symbol_table::process_new_functions): Update.
27433 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
27434 (inline_generate_summary): Rename to ...
27435 (ipa_fn_summary_generate): ... this one.
27436 (inline_read_summary): Rename to ...
27437 (ipa_fn_summary_read): ... this one.
27438 (inline_write_summary): Rename to ...
27439 (ipa_fn_summary_write): ... this one.
27440 (inline_free_summary): Rename to ...
27441 (ipa_free_fn_summary): ... this one.
27442 (pass_data_local_fn_summary, pass_local_fn_summary,
27443 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
27444 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
27445 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
27446 make_pass_ipa_fn_summary): New.
27447 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
27448 inline_write_summary, inline_free_summary): Remove.
27449 (ipa_free_fn_summary) : New.
27450 * ipa-inline.c (ipa_inline): Update.
27451 (pass_ipa_inline): Do not generate summaries.
27452 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
27454 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
27455 and add pass_ipa_fn_summary.
27456 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
27458 (make_pass_inline_parameters): Remove.
27460 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
27462 * omp-low.c (struct omp_context): Remove "default_kind" member.
27465 * omp-offload.c (execute_oacc_device_lower): Remove the
27466 parallelism dimensions function attributes for unparallelized
27467 OpenACC kernels constructs.
27469 2017-05-23 Martin Liska <mliska@suse.cz>
27471 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
27473 (cgraph_edge::make_speculative): Likewise.
27474 (cgraph_edge::resolve_speculation): Likewise.
27475 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
27476 (cgraph_node::dump): Likewise.
27477 * cgraph.h: Likewise.
27478 * cgraphunit.c (analyze_functions): Likewise.
27479 (symbol_table::compile): Likewise.
27480 * ipa-cp.c (print_all_lattices): Likewise.
27481 (determine_versionability): Likewise.
27482 (initialize_node_lattices): Likewise.
27483 (ipcp_verify_propagated_values): Likewise.
27484 (estimate_local_effects): Likewise.
27485 (update_profiling_info): Likewise.
27486 (create_specialized_node): Likewise.
27487 (perhaps_add_new_callers): Likewise.
27488 (decide_about_value): Likewise.
27489 (decide_whether_version_node): Likewise.
27490 (identify_dead_nodes): Likewise.
27491 (ipcp_store_bits_results): Likewise.
27492 * ipa-devirt.c (dump_targets): Likewise.
27493 (ipa_devirt): Likewise.
27494 * ipa-icf.c (sem_item::dump): Likewise.
27495 (sem_function::equals): Likewise.
27496 (sem_variable::equals): Likewise.
27497 (sem_item_optimizer::read_section): Likewise.
27498 (sem_item_optimizer::execute): Likewise.
27499 (congruence_class::dump): Likewise.
27500 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
27501 (dump_inline_summary): Likewise.
27502 (estimate_node_size_and_time): Likewise.
27503 (inline_analyze_function): Likewise.
27504 * ipa-inline-transform.c (inline_call): Likewise.
27505 * ipa-inline.c (report_inline_failed_reason): Likewise.
27506 (want_early_inline_function_p): Likewise.
27507 (edge_badness): Likewise.
27508 (update_edge_key): Likewise.
27509 (inline_small_functions): Likewise.
27510 * ipa-profile.c (ipa_profile): Likewise.
27511 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
27512 (ipa_make_edge_direct_to_target): Likewise.
27513 (remove_described_reference): Likewise.
27514 (ipa_impossible_devirt_target): Likewise.
27515 (propagate_controlled_uses): Likewise.
27516 (ipa_print_node_params): Likewise.
27517 (ipcp_transform_function): Likewise.
27518 * ipa-pure-const.c (pure_const_read_summary): Likewise.
27519 (propagate_pure_const): Likewise.
27520 * ipa-reference.c (generate_summary): Likewise.
27521 (read_write_all_from_decl): Likewise.
27522 (propagate): Likewise.
27523 (ipa_reference_read_optimization_summary): Likewise.
27524 * ipa-utils.c (ipa_merge_profiles): Likewise.
27525 * ipa.c (walk_polymorphic_call_targets): Likewise.
27526 (symbol_table::remove_unreachable_nodes): Likewise.
27527 (ipa_single_use): Likewise.
27528 * passes.c (execute_todo): Likewise.
27529 * predict.c (drop_profile): Likewise.
27530 * symtab.c (symtab_node::get_dump_name): New function.
27531 (symtab_node::dump_name): Likewise.
27532 (symtab_node::dump_asm_name): Likewise.
27533 (symtab_node::dump_references): Likewise.
27534 (symtab_node::dump_referring): Likewise.
27535 (symtab_node::dump_base): Likewise.
27536 (symtab_node::debug_symtab): Likewise.
27537 * tree-sra.c (convert_callers_for_node): Likewise.
27538 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
27539 * value-prof.c (init_node_map): Likewise.
27541 2017-05-23 Martin Liska <mliska@suse.cz>
27543 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
27544 and symtab_node::debug_symtab to symbol_table::debug.
27545 * cgraphunit.c (analyze_functions): Use the renamed function.
27546 (symbol_table::compile): Likewise.
27547 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
27548 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
27549 * passes.c (execute_todo): Likewise.
27550 * symtab.c (symbol_table::dump): New function.
27551 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
27553 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27555 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
27556 that nonconst implies exec.
27558 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27560 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27561 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27562 (inline_edge_summary_vec): Turn into ...
27563 (ipa_call_summaries): ... this one.
27564 (redirect_to_unreachable, edge_set_predicate,
27565 evaluate_properties_for_edge, inline_summary_alloc,
27566 reset_ipa_call_summary, reset_inline_summary,
27567 inline_summary_t::duplicate): Update.
27568 (inline_edge_duplication_hook): Turn to ...
27569 (ipa_call_summary_t::duplicate): ... this one.
27570 (inline_edge_removal_hook): Turn to ...
27571 (ipa_call_summary_t::remove): ... this one.
27572 (dump_inline_edge_summary): Turn to ...
27573 (dump_ipa_call_summary): ... this one.
27574 (estimate_function_body_sizes): Update.
27575 (inline_update_callee_summaries): Update.
27576 (remap_edge_change_prob): Update.
27577 (remap_edge_summaries): Update.
27578 (inline_merge_summary): Update.
27579 (do_estimate_edge_time): Update.
27580 (inline_generate_summary): Update.
27581 (inline_read_section): Update.
27582 (inline_read_summary): Update.
27583 (inline_free_summary): Update.
27584 * ipa-inline.c (can_inline_edge_p): Update.
27585 (compute_inlined_call_time): Update.
27586 (want_inline_small_function_p): Update.
27587 (edge_badness): Update.
27588 (early_inliner): Update.
27589 * ipa-inline.h (inline_edge_summary): Turn to ...
27590 (ipa_call_summary): ... this one.
27591 (ipa_call_summary_t): New class.
27592 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27593 (ipa_call_summaries): New.
27594 (inline_edge_summary): Remove.
27595 (estimate_edge_growth): Update.
27596 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27597 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27598 * ipa-split.c (execute_split_functions): Update.
27599 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27601 2017-05-23 Tom de Vries <tom@codesourcery.com>
27603 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27604 attributes): Document rdrand effective target.
27606 2017-05-23 Tom de Vries <tom@codesourcery.com>
27608 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27609 attributes): Sort alphabetically.
27611 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
27613 * config/avr/genmultilib.awk: Use gsub instead of gensub.
27615 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
27618 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
27619 V2DF/V2DI splat into two separate patterns, one that handles
27620 registers, and the other that only handles memory. Drop support
27621 for splatting from a GPR on ISA 2.07 and then splitting the
27622 splat into direct move and splat.
27623 (vsx_splat_<mode>_reg): Likewise.
27624 (vsx_splat_<mode>_mem): Likewise.
27626 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
27628 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
27630 2017-05-22 Jakub Jelinek <jakub@redhat.com>
27632 PR middle-end/80809
27633 * omp-low.c (finish_taskreg_remap): New function.
27634 (finish_taskreg_scan): If unit size of ctx->record_type
27635 is non-constant, unshare the size expression and replace
27636 decls in it with possible outer var refs.
27638 PR middle-end/80809
27639 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
27640 GOVD_SHARED rather than GOVD_PRIVATE with it.
27641 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
27642 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
27644 PR middle-end/80853
27645 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
27646 as last argument to build_outer_var_ref for pointer bases of array
27647 section reductions.
27649 2017-05-19 Martin Sebor <msebor@redhat.com>
27651 * print-tree.c (print_node): Print DECL_READ_P flag.
27653 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27655 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
27656 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
27657 * cgraph.c: Likewise.
27658 * cgraphunit.c: Likewise.
27659 * gengtype.c: Likewise.
27660 * ipa-cp.c: Likewise.
27661 * ipa-devirt.c: Likewise.
27662 * ipa-icf.c: Likewise.
27663 * ipa-predicate.c: Likewise.
27664 * ipa-profile.c: Likewise.
27665 * ipa-prop.c: Likewise.
27666 * ipa-split.c: Likewise.
27668 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
27669 edge_predicate_pool, dump_inline_hints,
27670 inline_summary::account_size_time, redirect_to_unreachable,
27671 edge_set_predicate, set_hint_predicate,
27672 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
27673 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
27674 inline_summary_t::remove, remap_hint_predicate_after_duplication,
27675 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
27676 ipa_call_summary_t::remove, initialize_growth_caches,
27677 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
27678 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
27679 mark_modified, unmodified_parm_1, unmodified_parm,
27680 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
27681 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
27682 compute_bb_predicates, will_be_nonconstant_expr_predicate,
27683 will_be_nonconstant_predicate, record_modified_bb_info,
27684 get_minimal_bb, record_modified, param_change_prob,
27685 phi_result_unknown_predicate, predicate_for_phi_result,
27686 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
27687 estimate_function_body_sizes, compute_inline_parameters,
27688 compute_inline_parameters_for_curren, pass_data_inline_parameters,
27689 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
27690 inline_update_callee_summaries, remap_edge_change_prob,
27691 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
27692 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
27693 inline_analyze_function, inline_summary_t::insert,
27694 inline_generate_summary, read_ipa_call_summary, inline_read_section,
27695 inline_read_summary, write_ipa_call_summary, inline_write_summary,
27696 inline_free_summary): Move to ipa-fnsummary.h
27697 (predicate_t): Remove.
27698 * ipa-fnsummary.c: New file.
27699 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
27700 (enum inline_hints_vals, inline_hints, agg_position_info,
27701 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
27702 inline_summaries, ipa_call_summary, ipa_call_summary_t,
27703 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
27704 dump_inline_summary, dump_inline_hints, inline_generate_summary,
27705 inline_read_summary, inline_write_summary, inline_free_summary,
27706 inline_analyze_function, initialize_inline_failed,
27707 inline_merge_summary, inline_update_overall_summary,
27708 compute_inline_parameters): Move to ipa-fnsummary.h
27709 * ipa-fnsummary.h: New file.
27710 * ipa-inline-transform.h: Include ipa-inline.h.
27711 * ipa-inline.c: LIkewise.
27713 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27715 * ipa-inline.c (edge_badness): Use inlined_time instead of
27716 inline_summaries->get.
27718 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27720 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
27722 2017-05-22 Nathan Sidwell <nathan@acm.org>
27724 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
27725 (fdump-lang): Document 'raw' option.
27726 * dumpfile.h (TDI_tu): Delete.
27727 * dumpfile.c (dump_files): Remove translation-unit.
27728 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
27730 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
27732 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
27733 command option from $(AWK) call.
27734 * config/avr/genmultilib.awk: Simplify and rewrite so that it
27735 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
27736 [FORMAT]: Remove handling of variable.
27737 * config/avr/t-multilib: Regenerate.
27739 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27741 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
27743 (dump_inline_summary): Do not print self_time.
27744 (estimate_function_body_sizes): Do not set self_time.
27745 (compute_inline_parameters): Likewise.
27746 (inline_read_section, inline_write_summary): Do not stream self_time.
27747 * ipa-inline.h (inline_summary): Drop self_time.
27749 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27751 * ipa-inline-analysis.c (account_size_time): Rename to ...
27752 (inline_summary::account_size_time): ... this one.
27753 (reset_ipa_call_summary): Turn to ...
27754 (ipa_call_summary::reset): ... this one.
27755 (reset_inline_summary): Turn to ...
27756 (inline_summary::reset): ... this one.
27757 (inline_summary_t::remove): Update.
27758 (inline_summary_t::duplicate): Update.
27759 (ipa_call_summary_t::remove): Update.
27760 (dump_inline_summary): Update.
27761 (estimate_function_body_sizes): Update.
27762 (compute_inline_parameters): Update.
27763 (estimate_node_size_and_time): Update.
27764 (inline_merge_summary): Update.
27765 (inline_update_overall_summary): Update.
27766 (inline_read_section): Update.
27767 (inline_write_summary): Update.
27768 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
27769 add account_size_time and reset member functions.
27770 (ipa_call_summary): Add reset function.
27771 * ipa-predicate.h (predicate::operator &): Constify.
27773 2017-05-22 Richard Biener <rguenther@suse.de>
27775 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
27777 2017-05-19 Jason Merrill <jason@redhat.com>
27779 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
27781 2017-05-19 Marek Polacek <polacek@redhat.com>
27784 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
27785 TYPE_OVERFLOW_WRAPS checks.
27787 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
27789 * tree-core.h (enum omp_clause_default_kind): Add
27790 "OMP_CLAUSE_DEFAULT_PRESENT".
27791 * tree-pretty-print.c (dump_omp_clause): Handle it.
27792 * gimplify.c (enum gimplify_omp_var_data): Add
27793 "GOVD_MAP_FORCE_PRESENT".
27794 (gimplify_adjust_omp_clauses_1): Map it to
27795 "GOMP_MAP_FORCE_PRESENT".
27796 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
27798 * gimplify.c (oacc_default_clause): Clarify.
27800 2017-05-19 Nathan Sidwell <nathan@acm.org>
27802 LANG_HOOK_REGISTER_DUMPS
27803 * toplev.c (general_init): Call register dump lang hook.
27804 * doc/invoke.texi: Document -fdump-lang option family.
27805 * dumpfile.c (dump_files): Remove class dump here.
27806 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
27807 * dumpfile.h (tree_dump_index): Remove TDI_class.
27808 * langhooks-def.h (lhd_register_dumps): Declare.
27809 (LANG_HOOKS_REGISTER_DUMPS): Define.
27810 (LANG_HOOKS_INITIALIZER): Add it.
27811 * langhooks.c (lhd_register_dumps): Define.
27812 * langhooks.h (struct lang_hooks): Add register_dumps.
27814 2017-05-19 Nathan Sidwell <nathan@acm.org>
27816 * context.h (context::set_passes): New.
27817 * context.c (context::context): Do not create pass manager.
27818 * toplev.c (general_init): Create pass manager here.
27820 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
27822 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
27823 use this splitter if two add or or instructions would also work for
27824 the constant we want to generate.
27826 2017-05-19 Richard Biener <rguenther@suse.de>
27829 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
27830 predicate evaluation.
27832 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27834 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
27836 * ipa-inline.c (want_inline_small_function_p): Do not cast to
27839 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27841 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27842 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27843 (inline_edge_summary_vec): Turn into ...
27844 (ipa_call_summaries): ... this one.
27845 (redirect_to_unreachable, edge_set_predicate,
27846 evaluate_properties_for_edge, inline_summary_alloc,
27847 reset_ipa_call_summary, reset_inline_summary,
27848 inline_summary_t::duplicate): Update.
27849 (inline_edge_duplication_hook): Turn to ...
27850 (ipa_call_summary_t::duplicate): ... this one.
27851 (inline_edge_removal_hook): Turn to ...
27852 (ipa_call_summary_t::remove): ... this one.
27853 (dump_inline_edge_summary): Turn to ...
27854 (dump_ipa_call_summary): ... this one.
27855 (estimate_function_body_sizes): Update.
27856 (inline_update_callee_summaries): Update.
27857 (remap_edge_change_prob): Update.
27858 (remap_edge_summaries): Update.
27859 (inline_merge_summary): Update.
27860 (do_estimate_edge_time): Update.
27861 (inline_generate_summary): Update.
27862 (inline_read_section): Update.
27863 (inline_read_summary): Update.
27864 (inline_free_summary): Update.
27865 * ipa-inline.c (can_inline_edge_p): Update.
27866 (compute_inlined_call_time): Update.
27867 (want_inline_small_function_p): Update.
27868 (edge_badness): Update.
27869 (early_inliner): Update.
27870 * ipa-inline.h (inline_edge_summary): Turn to ...
27871 (ipa_call_summary): ... this one.
27872 (ipa_call_summary_t): New class.
27873 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27874 (ipa_call_summaries): New.
27875 (inline_edge_summary): Remove.
27876 (estimate_edge_growth): Update.
27877 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27878 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27879 * ipa-split.c (execute_split_functions): Update.
27880 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27882 2017-05-19 Richard Biener <rguenther@suse.de>
27884 PR middle-end/80764
27885 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
27887 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
27889 * config/rs6000/rs6000.c (struct machine_function): Add field
27890 fpr_is_wrapped_separately.
27891 (rs6000_get_separate_components): Use 64 components. Handle the
27892 new FPR components.
27893 (rs6000_components_for_bb): Handle the FPR components.
27894 (rs6000_emit_prologue_components): Handle the FPR components.
27895 (rs6000_emit_epilogue_components): Handle the FPR components.
27896 (rs6000_set_handled_components): Handle the FPR components.
27897 (rs6000_emit_prologue): Don't output prologue code for those FPRs
27898 that are already separately shrink-wrapped.
27899 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
27900 that are already separately shrink-wrapped.
27902 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27905 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
27908 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
27909 (define_peephole2 for Altivec d-form load): Add peepholes to catch
27910 cases where the register allocator uses a move and an offsettable
27911 memory operation to/from a FPR register on ISA 2.06/2.07.
27912 (define_peephole2 for Altivec d-form store): Likewise.
27914 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
27917 * config/i386/mmx.md (*mov<mode>_internal): Enable
27918 alternatives 11, 12, 13 and 14 also for 32bit targets.
27919 Remove alternatives 15, 16, 17 and 18.
27920 * config/i386/sse.md (vec_concatv2di): Change
27921 alternative (!x, *y) to (x, ?!*Yn).
27923 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
27925 * dumpfile.h (enum dump_kind): Remove stray comma.
27927 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27929 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
27930 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
27931 predicate::num_conditions
27932 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
27933 (CHANGED): turn into predicate::changed.
27934 (agg_position_info): Move to ipa-predicate.h
27935 (add_condition, predicate::add_clause, predicate::operator &=,
27936 predicate::or_with, predicate::evaluate, predicate::probability,
27937 dump_condition, dump_clause, predicate::dump,
27938 predicate::remap_after_duplication, predicate::remap_after_inlining,
27939 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
27940 (evaluate_conditions_for_known_args): Update.
27941 (set_cond_stmt_execution_predicate): Update.
27942 * ipa-inline.h: Include ipa-predicate.h
27943 (condition, inline_param_summary, conditions, agg_position_info,
27944 predicate): Move to ipa-predicate.h
27945 * ipa-predicate.c: New file.
27946 * ipa-predicate.h: New file.
27948 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
27950 * final.c (leaf_function_p): Check we are not in a sequence.
27952 2017-05-18 Martin Liska <mliska@suse.cz>
27954 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
27955 * dumpfile.c (dump_register): Use new enum dump_kind.
27956 (get_dump_file_name): Likewise.
27957 (dump_enable_all): Likewise.
27958 (dump_switch_p_1): Likewise.
27959 (enable_rtl_dump_file): Remove usage of TDF_RTL.
27960 * dumpfile.h (enum dump_kind): New enum type.
27961 (struct dump_file_info): Create constructor and
27962 format fields and comments.
27963 * passes.c (pass_manager::register_one_dump_file):
27965 * statistics.c (statistics_early_init): Likewise.
27966 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
27967 TDF_TREE with TDF_SLIM.
27968 (gather_memory_references_ref): Likewise.
27970 2017-05-18 Martin Liska <mliska@suse.cz>
27972 * vec.h (struct vnull): Use it.
27974 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27976 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
27977 (true_predicate, false_predicate, true_predicate_p,
27978 false_predicate_p): Remove.
27979 (single_cond_predicate, not_inlined_predicate): Turn to member function
27981 (add_condition): Update.
27982 (add_clause): Turn to...
27983 (predicate::add_clause): ... this one; update; allow passing NULL
27985 (and_predicates): Turn to ...
27986 (predicate::operator &=): ... this one.
27987 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
27988 (or_predicates): Turn to ...
27989 (predicate::or_with): ... this one.
27990 (evaluate_predicate): Turn to ...
27991 (predicate::evaluate): ... this one.
27992 (predicate_probability): Turn to ...
27993 (predicate::probability): ... this one.
27994 (dump_condition): Update.
27995 (dump_predicate): Turn to ...
27996 (predicate::dump): ... this one.
27997 (account_size_time): Update.
27998 (edge_set_predicate): Update.
27999 (set_hint_predicate): UPdate.
28000 (evaluate_conditions_for_known_args): Update.
28001 (evaluate_properties_for_edge): Update.
28002 (remap_predicate_after_duplication): Turn to...
28003 (predicate::remap_after_duplication): ... this one.
28004 (remap_hint_predicate_after_duplication): Update.
28005 (inline_summary_t::duplicate): UPdate.
28006 (dump_inline_edge_summary): Update.
28007 (dump_inline_summary): Update.
28008 (set_cond_stmt_execution_predicate): Update.
28009 (set_switch_stmt_execution_predicate): Update.
28010 (compute_bb_predicates): Update.
28011 (will_be_nonconstant_expr_predicate): Update.
28012 (will_be_nonconstant_predicate): Update.
28013 (phi_result_unknown_predicate): Update.
28014 (predicate_for_phi_result): Update.
28015 (array_index_predicate): Update.
28016 (estimate_function_body_sizes): Update.
28017 (estimate_node_size_and_time): Update.
28018 (estimate_ipcp_clone_size_and_time): Update.
28019 (remap_predicate): Rename to ...
28020 (predicate::remap_after_inlining): ... this one.
28021 (remap_hint_predicate): Update.
28022 (inline_merge_summary): Update.
28023 (inline_update_overall_summary): Update.
28024 (estimate_size_after_inlining): Update.
28025 (read_predicate): Rename to ...
28026 (predicate::stream_in): ... this one.
28027 (read_inline_edge_summary): Update.
28028 (write_predicate): Rename to ...
28029 (predicate::stream_out): ... this one.
28030 (write_inline_edge_summary): Update.
28031 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
28032 (clause_t): Turn to uint32_t
28033 (predicate): Turn to class; implement constructor and operators
28035 (size_time_entry): Update.
28036 (inline_summary): Update.
28037 (inline_edge_summary): Update.
28039 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
28041 * fold-const.c (fold_binary_loc): Move transformation...
28042 * match.pd (C - X CMP X): ... here.
28044 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
28046 * config/sparc/sparc.c (sparc_option_override): Set function
28047 alignment for -mcpu=niagara7 to 64 to match the I$ line.
28048 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
28050 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
28052 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
28054 2017-05-18 Marek Polacek <polacek@redhat.com>
28057 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
28058 (pass_ubsan::execute): Call gimple_assign_single_p instead of
28059 gimple_assign_load_p.
28061 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
28063 PR middle-end/80692
28064 * real.c (do_compare): Give decimal_do_compare preference over
28065 comparing just the signs.
28067 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
28069 * doc/md.texi (Canonicalization of Instructions): Describe the
28070 canonical form of instructions that inherently set a condition
28073 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
28075 PR middle-end/80775
28076 * tree-cfg.c: Move deletion of unreachable case statements to after
28077 the merging of consecutive case labels.
28079 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
28081 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
28082 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
28083 restoring of callee-saved registers.
28085 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
28087 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
28088 * config/visium/visium.c (single_set_and_flags): Likewise.
28089 * config/visium/visium.md (Substitutions): Likewise.
28091 2017-05-17 Martin Liska <mliska@suse.cz>
28093 * cfg.c: Introduce dump_flags_t type and
28094 use it instead of int type.
28096 * cfghooks.c: Likewise.
28097 * cfghooks.h (struct cfg_hooks): Likewise.
28098 * cfgrtl.c: Likewise.
28099 * cfgrtl.h: Likewise.
28100 * cgraph.c (cgraph_node::get_body): Likewise.
28101 * coretypes.h: Likewise.
28102 * domwalk.c: Likewise.
28103 * domwalk.h: Likewise.
28104 * dumpfile.c (struct dump_option_value_info): Likewise.
28105 (dump_enable_all): Likewise.
28106 (dump_switch_p_1): Likewise.
28107 (opt_info_switch_p): Likewise.
28108 * dumpfile.h (enum tree_dump_index): Likewise.
28109 (struct dump_file_info): Likewise.
28110 * genemit.c: Likewise.
28111 * generic-match-head.c: Likewise.
28112 * gengtype.c (open_base_files): Likewise.
28113 * gimple-pretty-print.c: Likewise.
28114 * gimple-pretty-print.h: Likewise.
28115 * graph.c (print_graph_cfg): Likewise.
28116 * graphite-scop-detection.c (dot_all_sese): Likewise.
28117 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
28118 * loop-unroll.c (report_unroll): Likewise.
28119 * passes.c (pass_manager::register_one_dump_file): Likewise.
28120 * print-tree.c: Likewise.
28121 * statistics.c: Likewise.
28122 * tree-cfg.c: Likewise.
28123 * tree-cfg.h: Likewise.
28124 * tree-dfa.c: Likewise.
28125 * tree-dfa.h: Likewise.
28126 * tree-dump.c (dump_function): Likewise.
28127 * tree-dump.h (struct dump_info): Likewise.
28128 * tree-pretty-print.c: Likewise.
28129 * tree-pretty-print.h: Likewise.
28130 * tree-ssa-live.c: Likewise.
28131 * tree-ssa-live.h: Likewise.
28132 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
28133 * tree-vect-loop.c: Likewise.
28134 * tree-vect-slp.c: Likewise.
28136 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
28137 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28139 PR tree-optimization/80457
28140 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
28141 of all arguments to a statement as scalar_to_vec operations.
28142 (vectorizable_call): Adjust call to vect_model_simple_cost for
28144 (vectorizable_conversion): Likewise.
28145 (vectorizable_assignment): Likewise.
28146 (vectorizable_shift): Likewise.
28147 (vectorizable_operation): Likewise.
28148 (vectorizable_comparison): Likewise.
28149 (vect_is_simple_cond): Record the def types for operands.
28150 (vectorizable_condition): Likewise, call vect_model_simple_cost.
28151 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
28152 for statement argument count.
28154 2017-05-16 Carl Love <cel@us.ibm.com>
28156 * config/rs6000/rs6000-c: Add support for built-in functions
28157 vector unsigned long long vec_bperm (vector unsigned long long,
28158 vector unsigned char)
28159 vector signed long long vec_mule (vector signed int,
28161 vector unsigned long long vec_mule (vector unsigned int,
28162 vector unsigned int)
28163 vector signed long long vec_mulo (vector signed int,
28165 vector unsigned long long vec_mulo (vector unsigned int,
28166 vector unsigned int)
28167 vector signed char vec_sldw (vector signed char,
28168 vector signed char,
28170 vector unsigned char vec_sldw (vector unsigned char,
28171 vector unsigned char,
28173 vector signed short vec_sldw (vector signed short,
28174 vector signed short,
28176 vector unsigned short vec_sldw (vector unsigned short,
28177 vector unsigned short,
28179 vector signed int vec_sldw (vector signed int,
28182 vector unsigned int vec_sldw (vector unsigned int,
28183 vector unsigned int,
28185 vector signed long long vec_sldw (vector signed long long,
28186 vector signed long long,
28188 vector unsigned long long vec_sldw (vector unsigned long long,
28189 vector unsigned long long,
28191 * config/rs6000/rs6000-c: Add support for built-in functions
28192 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
28193 * config/rs6000/altivec.h: Add defintion for vec_sldw.
28194 * doc/extend.texi: Update the built-in documentation for the
28195 new built-in functions.
28197 2017-05-16 Marek Polacek <polacek@redhat.com>
28201 * tree.c (save_expr): Don't fold the expression.
28203 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
28205 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
28206 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
28207 and (?*y,m). Update insn attributes.
28209 2017-05-16 Martin Liska <mliska@suse.cz>
28211 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
28212 flags argument of print_gimple_stmt, print_gimple_expr,
28213 print_generic_stmt and print_generic_expr.
28214 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
28215 * coretypes.h: Likewise.
28216 * except.c (dump_eh_tree): Likewise.
28217 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
28218 * gimple-pretty-print.h: Likewise.
28219 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
28220 (backprop::push_to_worklist): Likewise.
28221 (backprop::pop_from_worklist): Likewise.
28222 (backprop::process_use): Likewise.
28223 (backprop::intersect_uses): Likewise.
28224 (note_replacement): Likewise.
28225 * gimple-ssa-store-merging.c
28226 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
28227 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
28228 (pass_store_merging::execute): Likewise.
28229 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
28230 (ssa_base_cand_dump_callback): Likewise.
28231 (dump_incr_vec): Likewise.
28232 (replace_refs): Likewise.
28233 (replace_mult_candidate): Likewise.
28234 (create_add_on_incoming_edge): Likewise.
28235 (create_phi_basis): Likewise.
28236 (insert_initializers): Likewise.
28237 (all_phi_incrs_profitable): Likewise.
28238 (introduce_cast_before_cand): Likewise.
28239 (replace_one_candidate): Likewise.
28240 * gimplify.c (gimplify_expr): Likewise.
28241 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
28242 (set_rename): Likewise.
28243 (rename_uses): Likewise.
28244 (copy_loop_phi_nodes): Likewise.
28245 (add_close_phis_to_merge_points): Likewise.
28246 (copy_loop_close_phi_args): Likewise.
28247 (copy_cond_phi_args): Likewise.
28248 (graphite_copy_stmts_from_block): Likewise.
28249 (translate_pending_phi_nodes): Likewise.
28250 * graphite-poly.c (print_pdr): Likewise.
28251 (dump_gbb_cases): Likewise.
28252 (dump_gbb_conditions): Likewise.
28253 (print_scop_params): Likewise.
28254 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
28255 (build_cross_bb_scalars_use): Likewise.
28256 (gather_bbs::before_dom_children): Likewise.
28257 * hsa-dump.c (dump_hsa_immed): Likewise.
28258 * ipa-cp.c (print_ipcp_constant_value): Likewise.
28259 (get_replacement_map): Likewise.
28260 * ipa-inline-analysis.c (dump_condition): Likewise.
28261 (estimate_function_body_sizes): Likewise.
28262 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
28263 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
28264 * ipa-prop.c (ipa_dump_param): Likewise.
28265 (ipa_print_node_jump_functions_for_edge): Likewise.
28266 (ipa_modify_call_arguments): Likewise.
28267 (ipa_modify_expr): Likewise.
28268 (ipa_dump_param_adjustments): Likewise.
28269 (ipa_dump_agg_replacement_values): Likewise.
28270 (ipcp_modif_dom_walker::before_dom_children): Likewise.
28271 * ipa-pure-const.c (check_stmt): Likewise.
28272 (pass_nothrow::execute): Likewise.
28273 * ipa-split.c (execute_split_functions): Likewise.
28274 * omp-offload.c (dump_oacc_loop_part): Likewise.
28275 (dump_oacc_loop): Likewise.
28276 * trans-mem.c (tm_log_emit): Likewise.
28277 (tm_memopt_accumulate_memops): Likewise.
28278 (dump_tm_memopt_set): Likewise.
28279 (dump_tm_memopt_transform): Likewise.
28280 * tree-cfg.c (gimple_verify_flow_info): Likewise.
28281 (print_loop): Likewise.
28282 * tree-chkp-opt.c (chkp_print_addr): Likewise.
28283 (chkp_gather_checks_info): Likewise.
28284 (chkp_get_check_result): Likewise.
28285 (chkp_remove_check_if_pass): Likewise.
28286 (chkp_use_outer_bounds_if_possible): Likewise.
28287 (chkp_reduce_bounds_lifetime): Likewise.
28288 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
28289 (chkp_mark_completed_bounds): Likewise.
28290 (chkp_register_incomplete_bounds): Likewise.
28291 (chkp_mark_invalid_bounds): Likewise.
28292 (chkp_maybe_copy_and_register_bounds): Likewise.
28293 (chkp_build_returned_bound): Likewise.
28294 (chkp_get_bound_for_parm): Likewise.
28295 (chkp_build_bndldx): Likewise.
28296 (chkp_get_bounds_by_definition): Likewise.
28297 (chkp_generate_extern_var_bounds): Likewise.
28298 (chkp_get_bounds_for_decl_addr): Likewise.
28299 * tree-chrec.c (chrec_apply): Likewise.
28300 * tree-data-ref.c (dump_data_reference): Likewise.
28301 (dump_subscript): Likewise.
28302 (dump_data_dependence_relation): Likewise.
28303 (analyze_overlapping_iterations): Likewise.
28304 * tree-inline.c (expand_call_inline): Likewise.
28305 (tree_function_versioning): Likewise.
28306 * tree-into-ssa.c (dump_defs_stack): Likewise.
28307 (dump_currdefs): Likewise.
28308 (dump_names_replaced_by): Likewise.
28309 (dump_update_ssa): Likewise.
28310 (update_ssa): Likewise.
28311 * tree-object-size.c (pass_object_sizes::execute): Likewise.
28312 * tree-parloops.c (build_new_reduction): Likewise.
28313 (try_create_reduction_list): Likewise.
28314 (ref_conflicts_with_region): Likewise.
28315 (oacc_entry_exit_ok_1): Likewise.
28316 (oacc_entry_exit_single_gang): Likewise.
28317 * tree-pretty-print.h: Likewise.
28318 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
28319 (get_scalar_evolution): Likewise.
28320 (add_to_evolution): Likewise.
28321 (get_loop_exit_condition): Likewise.
28322 (analyze_evolution_in_loop): Likewise.
28323 (analyze_initial_condition): Likewise.
28324 (analyze_scalar_evolution): Likewise.
28325 (instantiate_scev): Likewise.
28326 (number_of_latch_executions): Likewise.
28327 (gather_chrec_stats): Likewise.
28328 (final_value_replacement_loop): Likewise.
28329 (scev_const_prop): Likewise.
28330 * tree-sra.c (dump_access): Likewise.
28331 (disqualify_candidate): Likewise.
28332 (create_access): Likewise.
28333 (reject): Likewise.
28334 (maybe_add_sra_candidate): Likewise.
28335 (create_access_replacement): Likewise.
28336 (analyze_access_subtree): Likewise.
28337 (analyze_all_variable_accesses): Likewise.
28338 (sra_modify_assign): Likewise.
28339 (initialize_constant_pool_replacements): Likewise.
28340 (find_param_candidates): Likewise.
28341 (decide_one_param_reduction): Likewise.
28342 (replace_removed_params_ssa_names): Likewise.
28343 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
28344 * tree-ssa-copy.c (dump_copy_of): Likewise.
28345 (copy_prop_visit_cond_stmt): Likewise.
28346 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
28347 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
28348 (record_equivalences_from_stmt): Likewise.
28349 * tree-ssa-dse.c (compute_trims): Likewise.
28350 (delete_dead_call): Likewise.
28351 (delete_dead_assignment): Likewise.
28352 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
28353 (forward_propagate_into_cond): Likewise.
28354 (pass_forwprop::execute): Likewise.
28355 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
28356 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
28358 (move_computations_worker): Likewise.
28359 (execute_sm): Likewise.
28360 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
28361 (remove_exits_and_undefined_stmts): Likewise.
28362 (remove_redundant_iv_tests): Likewise.
28363 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
28364 (adjust_iv_update_pos): Likewise.
28365 * tree-ssa-math-opts.c (bswap_replace): Likewise.
28366 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
28367 (value_replacement): Likewise.
28368 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
28369 * tree-ssa-pre.c (print_pre_expr): Likewise.
28370 (get_representative_for): Likewise.
28371 (create_expression_by_pieces): Likewise.
28372 (insert_into_preds_of_block): Likewise.
28373 (eliminate_insert): Likewise.
28374 (eliminate_dom_walker::before_dom_children): Likewise.
28375 (eliminate): Likewise.
28376 (remove_dead_inserted_code): Likewise.
28377 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
28378 * tree-ssa-reassoc.c (get_rank): Likewise.
28379 (eliminate_duplicate_pair): Likewise.
28380 (eliminate_plus_minus_pair): Likewise.
28381 (eliminate_not_pairs): Likewise.
28382 (undistribute_ops_list): Likewise.
28383 (eliminate_redundant_comparison): Likewise.
28384 (update_range_test): Likewise.
28385 (optimize_range_tests_var_bound): Likewise.
28386 (optimize_vec_cond_expr): Likewise.
28387 (rewrite_expr_tree): Likewise.
28388 (rewrite_expr_tree_parallel): Likewise.
28389 (linearize_expr): Likewise.
28390 (break_up_subtract): Likewise.
28391 (linearize_expr_tree): Likewise.
28392 (attempt_builtin_powi): Likewise.
28393 (attempt_builtin_copysign): Likewise.
28394 (transform_stmt_to_copy): Likewise.
28395 (transform_stmt_to_multiply): Likewise.
28396 (dump_ops_vector): Likewise.
28397 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
28398 (print_scc): Likewise.
28399 (set_ssa_val_to): Likewise.
28400 (visit_reference_op_store): Likewise.
28401 (visit_use): Likewise.
28402 (sccvn_dom_walker::before_dom_children): Likewise.
28403 (run_scc_vn): Likewise.
28404 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
28406 (expr_hash_elt::print): Likewise.
28407 (const_and_copies::pop_to_marker): Likewise.
28408 (const_and_copies::record_const_or_copy_raw): Likewise.
28409 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
28410 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
28411 (dump_predicates): Likewise.
28412 (find_uninit_use): Likewise.
28413 (warn_uninitialized_phi): Likewise.
28414 (pass_late_warn_uninitialized::execute): Likewise.
28415 * tree-ssa.c (verify_vssa): Likewise.
28416 (verify_ssa): Likewise.
28417 (maybe_optimize_var): Likewise.
28418 * tree-vrp.c (dump_value_range): Likewise.
28419 (dump_all_value_ranges): Likewise.
28420 (dump_asserts_for): Likewise.
28421 (register_edge_assert_for_2): Likewise.
28422 (vrp_visit_cond_stmt): Likewise.
28423 (vrp_visit_switch_stmt): Likewise.
28424 (vrp_visit_stmt): Likewise.
28425 (vrp_visit_phi_node): Likewise.
28426 (simplify_cond_using_ranges_1): Likewise.
28427 (fold_predicate_in): Likewise.
28428 (evrp_dom_walker::before_dom_children): Likewise.
28429 (evrp_dom_walker::push_value_range): Likewise.
28430 (evrp_dom_walker::pop_value_range): Likewise.
28431 (execute_early_vrp): Likewise.
28433 2017-05-16 Richard Biener <rguenther@suse.de>
28435 * dwarf2out.c (loc_list_from_tree_1): Do not create
28436 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
28438 2017-05-16 Richard Biener <rguenther@suse.de>
28440 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
28442 (note_variable_value_in_expr): If we resolved the decl ref
28443 do not push to the stack.
28445 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
28447 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
28448 operations in fast-math mode.
28449 (vaddq_f16): Likewise.
28450 (vmul_f16): Likewise.
28451 (vmulq_f16): Likewise.
28452 (vsub_f16): Likewise.
28453 (vsubq_f16): Likewise.
28454 * config/arm/neon.md (add<mode>3): New.
28456 (fma:<VH:mode>3): New. Also remove outdated comment.
28459 2017-05-16 Martin Liska <mliska@suse.cz>
28463 * ipa-devirt.c (warn_types_mismatch): Fix typo.
28464 (odr_types_equivalent_p): Likewise.
28466 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
28468 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
28470 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
28473 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
28474 non-interunit SSE move alternatives with '?'.
28475 (zero-extendsidi peephole2): New peephole to skip intermediate
28476 general register in SSE zero-extend sequence.
28478 2017-05-15 Jeff Law <law@redhat.com>
28480 * reorg.c (relax_delay_slots): Create a new variable to hold
28481 the temporary target rather than clobbering TARGET_LABEL.
28483 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
28484 missing argument to extract_bit_field call.
28485 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
28487 2017-05-15 Martin Liska <mliska@suse.cz>
28490 * gcc.c (process_command): Do not allow empty argument of -o option.
28492 2017-05-15 Renlin Li <renlin.li@arm.com>
28494 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
28495 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
28496 * config/aarch64/constraints.md (Usf): Add long call check.
28497 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
28498 (call_value): Likewise.
28499 (sibcall): Likewise.
28500 (sibcall_value): Likewise.
28502 (call_value_insn): New.
28503 (sibcall_insn): Update rtx pattern.
28504 (sibcall_value_insn): Likewise.
28505 (call_internal): Remove.
28506 (call_value_internal): Likewise.
28507 (sibcall_internal): Likewise.
28508 (sibcall_value_internal): Likewise.
28509 (call_reg): Likewise.
28510 (call_symbol): Likewise.
28511 (call_value_reg): Likewise.
28512 (call_value_symbol): Likewise.
28514 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
28517 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
28519 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
28521 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
28522 compatible with CCGOCmode and with CCZmode.
28524 2017-05-14 Martin Sebor <msebor@redhat.com>
28526 PR middle-end/77671
28527 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
28528 (gimple_fold_builtin_snprintf): Same.
28529 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
28530 (gimple_fold_builtin_snprintf): Same.
28531 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
28532 of character types.
28533 (is_call_safe): New function.
28534 (try_substitute_return_value): Call it.
28535 (try_simplify_call): New function.
28536 (pass_sprintf_length::handle_gimple_call): Call it.
28538 2017-05-14 Martin Sebor <msebor@redhat.com>
28540 PR middle-end/80669
28541 * builtins.c (expand_builtin_stpncpy): Simplify.
28543 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
28545 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
28546 * config/i386/i386.h
28547 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
28548 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
28549 (struct machine_function): Add new members call_ms2sysv,
28550 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
28551 (struct machine_frame_state): New fields sp_realigned and
28552 sp_realigned_offset.
28553 * config/i386/i386.c
28554 (enum xlogue_stub): New enum.
28555 (enum xlogue_stub_sets): New enum.
28556 (class xlogue_layout): New class.
28557 (struct ix86_frame): New fields stack_realign_allocate_offset,
28558 stack_realign_offset and outlined_save_offset. Modify comments to
28559 detail stack layout when using out-of-line stubs.
28560 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
28561 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
28562 -mcall-ms2sysv-xlogues.
28563 (stub_managed_regs): New static variable.
28564 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
28565 registers managed by out-of-line stub.
28566 (disable_call_ms2sysv_xlogues): New function.
28567 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
28568 m->call_ms2sysv when appropriate and compute frame layout for
28570 (sp_valid_at, fp_valid_at): New inline functions.
28571 (choose_basereg): New function.
28572 (choose_baseaddr): Add align parameter, use choose_basereg and modify
28574 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
28575 Use align parameter of choose_baseaddr to generated aligned SSE movs
28577 (pro_epilogue_adjust_stack): Modify to track
28578 machine_frame_state::sp_realigned.
28579 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
28580 (ix86_nsaved_sseregs): Likewise.
28581 (ix86_emit_save_regs): Likewise.
28582 (ix86_emit_save_regs_using_mov): Likewise.
28583 (ix86_emit_save_sse_regs_using_mov): Likewise.
28584 (get_scratch_register_on_entry): Likewise.
28585 (gen_frame_set): New function.
28586 (gen_frame_load): Likewise.
28587 (gen_frame_store): Likewise.
28588 (emit_outlined_ms2sysv_save): Likewise.
28589 (emit_outlined_ms2sysv_restore): Likewise.
28590 (ix86_expand_prologue): Modify stack re-alignment code and call
28591 emit_outlined_ms2sysv_save when appropriate.
28592 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
28593 parameter rtx_insn *insn, which allows the function to be used to only
28594 generate the notes.
28595 (ix86_expand_epilogue): Modify validity checks of frame and stack
28596 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
28597 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
28598 * config/i386/predicates.md
28599 (save_multiple): New predicate.
28600 (restore_multiple): Likewise.
28601 * config/i386/sse.md
28602 (save_multiple<mode>): New pattern.
28603 (save_multiple_realign<mode>): Likewise.
28604 (restore_multiple<mode>): Likewise.
28605 (restore_multiple_and_return<mode>): Likewise.
28606 (restore_multiple_leave_return<mode>): Likewise.
28607 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
28609 2017-05-14 Julia Koval <julia.koval@intel.com>
28611 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
28612 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
28613 (__builtin_ia32_xsetbv): New builtins.
28614 * config/i386/i386.c (ix86_expand_special_args_builtin):
28616 (ix86_expand_builtin): Special expand for new intrinsics.
28617 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
28618 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
28619 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
28621 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28623 * cfganal.c (inverted_post_order_compute): Change argument type
28625 * cfganal.h (inverted_post_order_compute): Adjust prototype.
28626 * df-core.c (rest_of_handle_df_initialize): Adjust.
28627 (rest_of_handle_df_finish): Likewise.
28628 (df_analyze_1): Likewise.
28629 (df_analyze): Likewise.
28630 (loop_inverted_post_order_compute): Change argument to be a vec *.
28631 (df_analyze_loop): Adjust.
28632 (df_get_n_blocks): Likewise.
28633 (df_get_postorder): Likewise.
28634 * df.h (struct df_d): Change field to be a vec.
28635 * lcm.c (compute_laterin): Adjust.
28636 (compute_available): Likewise.
28637 * lra-lives.c (lra_create_live_ranges_1): Likewise.
28638 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
28639 * tree-ssa-pre.c (compute_antic): Likewise.
28641 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28643 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
28644 (depth_first_search::depth_first_search): Change structure init
28645 function to this constructor.
28646 (depth_first_search::add_bb): Rename function to this member.
28647 (depth_first_search::execute): Likewise.
28648 (flow_dfs_compute_reverse_finish): Adjust.
28650 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28652 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
28653 (longest_simple_path): Likewise.
28654 * shrink-wrap.c (spread_components): Likewise.
28655 (disqualify_problematic_components): Likewise.
28656 (emit_common_heads_for_components): Likewise.
28657 (emit_common_tails_for_components): Likewise.
28658 (insert_prologue_epilogue_for_components): Likewise.
28660 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28662 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
28665 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28667 * df-core.c (df_set_blocks): Start using auto_bitmap.
28668 (df_compact_blocks): Likewise.
28669 * df-problems.c (df_rd_confluence_n): Likewise.
28670 * df-scan.c (df_insn_rescan_all): Likewise.
28671 (df_process_deferred_rescans): Likewise.
28672 (df_update_entry_block_defs): Likewise.
28673 (df_update_exit_block_uses): Likewise.
28674 (df_entry_block_bitmap_verify): Likewise.
28675 (df_exit_block_bitmap_verify): Likewise.
28676 (df_scan_verify): Likewise.
28677 * lra-constraints.c (lra_constraints): Likewise.
28678 (undo_optional_reloads): Likewise.
28679 (lra_undo_inheritance): Likewise.
28680 * lra-remat.c (calculate_gen_cands): Likewise.
28681 (do_remat): Likewise.
28682 * lra-spills.c (assign_spill_hard_regs): Likewise.
28683 (spill_pseudos): Likewise.
28684 * tree-ssa-pre.c (bitmap_set_and): Likewise.
28685 (bitmap_set_subtract_values): Likewise.
28687 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28689 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
28690 management with auto_bitmap.
28691 (fix_inter_tick): Likewise.
28692 (fix_recovery_deps): Likewise.
28693 * ira.c (add_store_equivs): Likewise.
28694 (find_moveable_pseudos): Likewise.
28695 (split_live_ranges_for_shrink_wrap): Likewise.
28696 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
28697 (rtx_reuse_manager::seen_def_p): Likewise.
28698 (rtx_reuse_manager::set_seen_def): Likewise.
28699 * print-rtl.h (class rtx_reuse_manager): Likewise.
28701 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28703 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
28705 (migrate_btr_def): Likewise.
28706 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
28707 * df-core.c (loop_post_order_compute): Likewise.
28708 (loop_inverted_post_order_compute): Likewise.
28709 * hsa-common.h: Likewise.
28710 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
28711 * init-regs.c (initialize_uninitialized_regs): Likewise.
28712 * ipa-inline.c (resolve_noninline_speculation): Likewise.
28713 (inline_small_functions): Likewise.
28714 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
28715 * ira.c (combine_and_move_insns): Likewise.
28716 (build_insn_chain): Likewise.
28717 * loop-invariant.c (find_invariants): Likewise.
28718 * lower-subreg.c (propagate_pseudo_copies): Likewise.
28719 * predict.c (tree_predict_by_opcode): Likewise.
28720 (predict_paths_leading_to): Likewise.
28721 (predict_paths_leading_to_edge): Likewise.
28722 (estimate_loops_at_level): Likewise.
28723 (estimate_loops): Likewise.
28724 * shrink-wrap.c (try_shrink_wrapping): Likewise.
28725 (spread_components): Likewise.
28726 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
28727 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
28728 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
28729 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
28730 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
28731 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
28732 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
28733 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
28734 (mark_threaded_blocks): Likewise.
28735 (thread_through_all_blocks): Likewise.
28736 * tree-ssa.c (verify_ssa): Likewise.
28737 (execute_update_addresses_taken): Likewise.
28738 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
28740 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28742 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
28744 (post_order_compute): Likewise.
28745 (inverted_post_order_compute): Likewise.
28746 (pre_and_rev_post_order_compute_fn): Likewise.
28748 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28750 * genrecog.c (int_set::int_set): Explicitly construct our
28751 auto_vec base class.
28752 * vec.h (auto_vec::auto_vec): New constructor.
28754 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28756 * bitmap.h (class auto_bitmap): New constructor taking
28757 bitmap_obstack * argument.
28759 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28761 * bitmap.h (class auto_bitmap): Change type of m_bits to
28762 bitmap_head, and adjust ctor / dtor and member operators.
28764 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
28766 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
28767 when returned register mode doesn't match original mode.
28769 2017-05-12 Jeff Law <law@redhat.com>
28770 Jakub Jelinek <jakub@redhat.com>
28772 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
28773 we look for cc setter after the compare-elim changes.
28774 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
28775 within the vector to match what compare-elim now expects.
28776 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
28777 (xorsi3_flags, one_cmplsi2_flags): Likewise.
28779 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
28780 after the compare-elim changes.
28781 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
28782 the vector to match what compare-elim now expects.
28783 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
28784 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
28785 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
28786 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
28787 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
28789 * config/visium/visium.c (single_set_and_flags): Fix where
28790 we look for cc setter after the compare-elim changes.
28791 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
28792 with the vector to match what compare-elim now expects.
28793 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
28794 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
28795 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
28796 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
28797 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
28798 (neg<mode>2_insn_set_overflow): Likewise.
28800 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
28802 PR middle-end/79794
28803 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
28804 maybe_expand_insn call, set ops[0].target. If still set after call,
28805 set alt_rtl. Add extra arg to recursive calls.
28806 (extract_bit_field): Add alt_rtl argument. Pass to
28808 * expmed.h (extract_bit_field): Fix prototype.
28809 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
28810 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
28811 to extract_bit_field_calls.
28812 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
28813 Pass alt_rtl to extract_bit_field calls.
28814 * calls.c (store_unaligned_arguments_into_psuedos)
28815 load_register_parameters): Pass extra NULL to extract_bit_field calls.
28816 * optabs.c (maybe_legitimize_operand): Clear op->target when call
28818 * optabs.h (struct expand_operand): Add target bitfield.
28820 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28822 * compare-elim.c (try_eliminate_compare): Canonicalize
28823 operation with embedded compare to
28824 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
28825 (set (reg) (operation)].
28827 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
28829 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28832 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
28833 cost of adding a carry flag for ADC instruction.
28834 [case MINUS]: Ignore the cost of subtracting a carry flag
28835 for SBB instruction.
28837 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
28839 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
28841 * config/rs6000/bmiintrin.h: New file.
28842 * config/rs6000/bmi2intrin.h: New file.
28843 * config/rs6000/x86intrin.h: New file.
28845 2017-05-12 Jeff Law <law@redhat.com>
28847 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
28850 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
28852 PR middle-end/80707
28853 * tree-cfg.c: Remove cfg edges of unreachable case statements.
28855 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28857 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28858 early expansion of vector divide builtins.
28859 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
28860 builtins identified as having unsigned arguments.
28862 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28864 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
28865 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
28866 expansion of vector logical operations (and, andc, or, xor,
28869 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28871 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
28872 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
28874 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28876 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28877 early GIMPLE expansion of vector multiplies.
28879 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28881 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
28882 TARGET_HAVE_MOVT conditional.
28883 (movt splitter): Likewise.
28885 2017-05-12 Richard Biener <rguenther@suse.de>
28887 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
28888 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28889 Fold all stmts not inplace.
28891 2017-05-12 Richard Biener <rguenther@suse.de>
28893 PR tree-optimization/80713
28894 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
28895 inserted_exprs bit for not removed stmts.
28897 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
28899 PR middle-end/69921
28900 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
28901 parallelized" attribute for parallelized OpenACC kernels.
28902 * omp-offload.c (execute_oacc_device_lower): Use it.
28904 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
28905 Set "oacc kernels" attribute.
28906 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
28907 parameter. Adjust all users.
28908 (oacc_fn_attrib_kernels_p): Remove function.
28909 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
28910 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
28911 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
28912 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
28913 assert "oacc kernels" attribute is set.
28915 2017-05-11 Carl Love <cel@us.ibm.com>
28917 * config/rs6000/rs6000-c: Add support for built-in functions
28918 vector unsigned char vec_popcnt (vector signed char)
28919 vector unsigned char vec_popcnt (vector unsigned char)
28920 vector unsigned short vec_popcnt (vector signed short)
28921 vector unsigned short vec_popcnt (vector unsigned short)
28922 vector unsigned int vec_popcnt (vector signed int)
28923 vector unsigned int vec_popcnt (vector unsigned int)
28924 vector unsigned long long vec_popcnt (vector signed long long)
28925 vector unsigned long long vec_popcnt (vector unsigned long long)
28926 vector signed long long vec_slo (vector signed long long,
28927 vector signed char)
28928 vector signed long long vec_slo (vector signed long long,
28929 vector unsigned char)
28930 vector unsigned long long vec_slo (vector unsigned long long,
28931 vector signed char)
28932 vector unsigned long long vec_slo (vector unsigned long long,
28933 vector unsigned char)
28934 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
28935 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
28936 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
28937 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
28938 * doc/extend.texi: Update the built-in documentation file for the
28939 new built-in functions.
28941 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
28943 * attribs.h (sorted_attr_string): Move machine independent
28944 functions for target clone support from the i386 port to common
28945 code. Rename ix86_function_versions to common_function_versions.
28946 Rename make_name to make_unique_name.
28947 (common_function_versions): Likewise.
28948 (make_unique_name): Likewise.
28949 (make_dispatcher_decl): Likewise.
28950 (is_function_default_version): Likewise.
28951 * attribs.c (attr_strcmp): Likewise.
28952 (sorted_attr_string): Likewise.
28953 (common_function_versions): Likewise.
28954 (make_unique_name): Likewise.
28955 (make_dispatcher_decl): Likewise.
28956 (is_function_default_version): Likewise.
28957 * config/i386/i386.c (attr_strcmp): Likewise.
28958 (sorted_attr_string): Likewise.
28959 (ix86_function_versions): Likewise.
28960 (make_name): Likewise.
28961 (make_dispatcher_decl): Likewise.
28962 (is_function_default_version): Likewise.
28963 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
28965 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28968 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
28969 Account for direct move costs for vec_construct of integer
28972 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
28975 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
28976 (UNSPEC_STX_ATOMIC): Ditto.
28977 (loaddi_via_sse): New insn.
28978 (storedi_via_sse): Ditto.
28979 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
28980 Update corresponding peephole2 patterns.
28981 (atomic_storedi_fpu): Ditto.
28983 2017-05-11 Julia Koval <julia.koval@intel.com>
28985 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
28986 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
28988 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
28989 (__builtin_ia32_rsqrt14ss_mask): New builtins.
28990 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
28992 2017-05-11 Nathan Sidwell <nathan@acm.org>
28994 * graphite-poly.c: Include dumpfile.h.
28996 * dumpfle.h (dump_function): Declare here ...
28997 * tree-dump.h (dump_function): ... not here.
28998 * dumpfile.c: #include tree-cfg.h.
28999 (dump_function): Move here from ...
29000 * tree-dump.c (dump_function): ... here.
29001 * gimplify.c: #include splay-tree.h, not tree-dump.h.
29002 * graphite-poly.c: Don't include tree-dump.h.
29003 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
29004 * print-tree.c: Likewise.
29005 * stor-layout.c: Likewise.
29006 * tree-nested.c: Likewise.
29008 * dumpfile.c (dump_start): Use TDF_FLAGS.
29009 (dump_enable_all): Fix TDF_KIND check thinko.
29011 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
29013 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
29014 array entries to represent two legal parameterizations of the
29015 overloaded __builtin_cmpb function, as represented by the
29016 P6_OV_BUILTIN_CMPB constant.
29017 (altivec_resolve_overloaded_builtin): Add special case handling
29018 for the __builtin_cmpb function, as represented by the
29019 P6_OV_BUILTIN_CMPB constant.
29020 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
29021 (BU_P6_64BIT_2): New macro.
29022 (BU_P6_OVERLOAD_2): New macro
29023 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
29024 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
29025 (CMPB): Add overload support to represent both 32-bit and 64-bit
29026 compare-bytes function.
29027 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
29028 support for TARGET_CMPB.
29029 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
29030 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
29031 documentation of the __builtin_cmpb overloaded built-in function.
29033 2017-05-11 Richard Biener <rguenther@suse.de>
29035 PR tree-optimization/80705
29036 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
29037 bases are not vectorizable.
29039 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29041 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
29042 when counting register pressure.
29044 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29046 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
29047 (iv_ca_more_deps): Renamed to ...
29048 (iv_ca_compare_deps): ... this.
29049 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
29051 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29053 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
29055 (determine_group_iv_costs): ... here.
29056 (find_inv_vars_cb): Record inv var if it's not recorded before.
29058 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29060 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
29061 (get_shiftadd_cost): Ditto.
29063 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29065 * tree-ssa-address.c: Include header file.
29066 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
29068 (add_to_parts): Refactor.
29069 (addr_to_parts): New parameter. Update use of move_hint_to_base.
29070 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
29073 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29075 PR tree-optimization/53090
29076 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
29078 (extract_cond_operands): Detect condition with IV on both sides
29079 and return COMP_IV_EXPR_2.
29080 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
29081 (rewrite_use_compare): Simplify by removing call to function
29082 extract_cond_operands.
29084 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29086 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
29087 (extract_cond_operands): Detect condition comparing against non-
29088 invariant bound and return appropriate enum value.
29089 (find_interesting_uses_cond): Update use of extract_cond_operands.
29090 Handle its return value accordingly.
29091 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
29093 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29095 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
29096 nonlinear iv_use computation in loop invariant sensitive way.
29098 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29100 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
29101 (find_iv_candidates): Call relate_compare_use_with_all_cands.
29103 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29105 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
29106 (dump_cand): Support iv_cand.inv_exprs.
29107 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
29109 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
29112 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29114 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
29116 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
29117 as local function. Include necessary header files.
29118 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
29120 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29122 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
29124 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29126 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
29127 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
29128 RSHIFT_EXPR and BIT_NOT_EXPR.
29130 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29132 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
29133 (adjust_setup_cost): New parameter supporting round up adjustment.
29134 (struct address_cost_data): Delete.
29135 (force_expr_to_var_cost): Don't bound cost with spill_cost.
29136 (split_address_cost, ptr_difference_cost): Delete.
29137 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
29138 (struct ainc_cost_data): New struct.
29139 (get_address_cost_ainc): New function.
29140 (get_address_cost, get_computation_cost): Reimplement.
29141 (determine_group_iv_cost_address): Record inv_expr for all uses of
29143 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
29144 (iv_ca_has_deps): Reimplemented to ...
29145 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
29147 (iv_ca_extend): Call iv_ca_more_deps.
29149 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29151 * tree-ssa-address.c (struct mem_address): Move to header file.
29152 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
29153 * tree-ssa-address.h (struct mem_address): Move from C file.
29154 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
29156 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29158 * tree-affine.h (aff_combination_type): New interface.
29159 (aff_combination_zero_p): Remove static.
29160 (aff_combination_const_p): New interface.
29161 (aff_combination_singleton_var_p): New interfaces.
29163 2017-05-11 Richard Biener <rguenther@suse.de>
29165 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
29166 Skip unreachable blocks and destinations.
29167 (eliminate): Move stmt removal and fixup ...
29168 (fini_eliminate): ... here. Skip inserted exprs.
29169 (pass_pre::execute): Move fini_pre after fini_eliminate.
29170 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
29171 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
29172 PRE to get rid of dead code that has invalid SSA form and
29173 split critical edges again.
29175 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29177 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29179 2017-05-11 Richard Biener <rguenther@suse.de>
29181 * passes.c (execute_function_todo): Verify loops if they are
29182 said to be up-to-date.
29183 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
29184 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
29186 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
29189 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
29190 handle calling assemble_external ourself.
29193 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
29194 modes with zero size. Enhance comment.
29196 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29198 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
29199 built-ins for vec_xl and vec_xst with short and char pointer
29202 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
29204 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
29205 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
29206 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
29207 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
29208 (_mm_maskz_min_round_ss): New intrinsics.
29209 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
29210 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
29211 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
29212 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
29213 (__builtin_ia32_minss_mask_round): New builtins.
29214 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29215 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29216 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
29218 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
29219 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
29221 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
29224 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
29226 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
29227 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
29228 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
29229 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
29230 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
29231 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
29232 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
29233 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29234 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
29235 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
29236 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
29237 (__builtin_ia32_mulss_mask_round): New builtins.
29238 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29239 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29240 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
29242 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
29243 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
29245 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
29248 2017-05-10 Julia Koval <julia.koval@intel.com>
29250 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
29251 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
29252 (_mm256_setr_m128i): New intrinsics.
29254 2017-05-10 Julia Koval <julia.koval@intel.com>
29256 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
29257 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
29258 (_mm_maskz_rcp14_ss): New intrinsics.
29259 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
29260 (__builtin_ia32_rcp14ss_mask): New builtins.
29261 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
29263 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
29265 PR tree-optimization/51513
29266 * tree-cfg.c (gimple_seq_unreachable_p): New function.
29267 (assert_unreachable_fallthru_edge_p): Use it.
29268 (group_case_labels_stmt): Likewise.
29269 * tree-cfg.h: Prototype it.
29270 * stmt.c: Include cfghooks.h and tree-cfg.h.
29271 (emit_case_dispatch_table) <gap_label>: New local variable.
29272 Use it to fill dispatch table gaps.
29273 Test for default_label before updating probabilities.
29274 (expand_case) <default_label>: Remove unneeded initialization.
29275 Test for unreachable default case statement and remove its edge.
29276 Set default_label accordingly.
29277 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
29279 2017-05-10 Carl Love <cel@us.ibm.com>
29281 * config/rs6000/rs6000-c: Add support for built-in functions
29282 vector signed char vec_neg (vector signed char)
29283 vector signed short int vec_neg (vector short int)
29284 vector signed int vec_neg (vector signed int)
29285 vector signed long long vec_neg (vector signed long long)
29286 vector float vec_neg (vector float)
29287 vector double vec_neg (vector double)
29288 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
29290 * config/rs6000/altivec.h: Add define for vec_neg
29291 * doc/extend.texi: Update the built-in documentation for the
29292 new built-in functions.
29294 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29296 PR tree-optimization/77644
29297 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
29299 2017-05-10 Nathan Sidwell <nathan@acm.org>
29301 * dumpfile.h (TDI_lang_all): New.
29302 (TDF_KIND): New. Renumber others
29303 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
29305 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
29307 (get_dump_file_name): Adjust suffix generation.
29308 (dump_enable_all): Use TDF_KIND.
29309 * doc/invoke.texi (-fdump-lang-all): Document.
29311 * dumpfile.h: Tabify.
29313 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
29316 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
29317 Move member access before delete.
29319 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
29321 * tree-inline.c (expand_call_inline): Split block at stmt
29324 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
29327 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
29328 are now unused after splitting mov{sf,sd}_hardfloat.
29329 (f32_lr2): Likewise.
29330 (f32_lm): Likewise.
29331 (f32_lm2): Likewise.
29332 (f32_li): Likewise.
29333 (f32_li2): Likewise.
29334 (f32_lv): Likewise.
29335 (f32_sr): Likewise.
29336 (f32_sr2): Likewise.
29337 (f32_sm): Likewise.
29338 (f32_sm2): Likewise.
29339 (f32_si): Likewise.
29340 (f32_si2): Likewise.
29341 (f32_sv): Likewise.
29342 (f32_dm): Likewise.
29343 (f32_vsx): Likewise.
29344 (f32_av): Likewise.
29345 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
29346 For movsf, order stores so the VSX stores occur before the GPR
29347 store which encourages the register allocator to use a traditional
29348 FPR instead of a GPR. For movsd, order the stores so that the GPR
29349 store comes before the VSX stores to allow the power6 to work.
29350 This is due to the power6 not having a 32-bit integer store
29351 instruction from a FPR.
29352 (movsf_hardfloat): Likewise.
29353 (movsd_hardfloat): Likewise.
29355 2017-05-09 Martin Sebor <msebor@redhat.com>
29357 PR translation/80280
29358 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
29361 PR translation/80280
29362 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
29363 data member added in r247778.
29364 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
29366 2017-05-09 Nathan Sidwell <nathan@acm.org>
29368 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
29370 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
29373 2017-05-09 Marek Polacek <polacek@redhat.com>
29375 * doc/invoke.texi: Fix typo.
29377 2017-05-09 Richard Biener <rguenther@suse.de>
29379 * tree-vrp.c (vrp_val_is_max): Adjust comment.
29380 (vrp_val_is_min): Likewise.
29381 (set_value_range_to_value): Likewise.
29382 (set_value_range_to_nonnegative): Likewise.
29383 (gimple_assign_nonzero_p): Likewise.
29384 (gimple_stmt_nonzero_p): Likewise.
29385 (vrp_int_const_binop): Likewise. Remove unreachable case.
29386 (adjust_range_with_scev): Adjust comments.
29387 (compare_range_with_value): Likewise.
29388 (extract_range_from_phi_node): Likewise.
29389 (test_for_singularity): Likewise.
29391 2017-05-09 Richard Biener <rguenther@suse.de>
29393 * tree-vrp.c (get_single_symbol): Add assert that we don't
29394 get overflowed constants as invariant part.
29395 (compare_values_warnv): Add comment before the TREE_NO_WARNING
29396 checks. Use wi::cmp instead of recursing for integer constants.
29397 (compare_values): Just ignore whether we assumed undefined
29398 overflow instead of failing the compare.
29399 (extract_range_for_var_from_comparison_expr): Add comment before the
29400 TREE_NO_WARNING sets.
29401 (test_for_singularity): Likewise.
29402 (extract_range_from_comparison): Do not disable optimization
29403 when we assumed undefined overflow.
29404 (extract_range_basic): Remove init of unused var.
29406 2017-05-09 Richard Biener <rguenther@suse.de>
29408 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
29409 (extract_range_from_multiplicative_op_1): Adjust.
29410 (extract_range_from_binary_expr_1): Use int_const_binop.
29412 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
29415 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
29416 rs6000_store_data_bypass_p in seven define_bypass directives and
29417 in several comments.
29418 * config/rs6000/rs6000-protos.h: Add prototype for
29419 rs6000_store_data_bypass_p function.
29420 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
29421 function implements slightly different (rs6000-specific) semantics
29422 than store_data_bypass_p, returning false rather than aborting
29423 with assertion error when arguments do not satisfy the
29424 requirements of store data bypass.
29425 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
29426 rs6000_store_data_bypass_p.
29428 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
29430 * config/xtensa/xtensa-protos.h
29431 (xtensa_initial_elimination_offset): New declaration.
29432 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29433 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
29434 macro definition, add case for FRAME_POINTER_REGNUM when
29435 FRAME_GROWS_DOWNWARD.
29436 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
29437 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
29438 xtensa_initial_elimination_offset.
29440 2017-05-08 Nathan Sidwell <nathan@acm.org>
29442 * doc/invoke.texi: Alphabetize -fdump options.
29444 2017-05-08 Martin Sebor <msebor@redhat.com>
29446 PR translation/80280
29447 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
29449 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
29451 * target.def (compute_frame_layout): New optional target hook.
29452 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
29453 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
29454 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
29456 * reload1.c (verify_initial_elim_offsets): Likewise.
29457 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
29458 (use_simple_return_p): Call arm_compute_frame_layout if needed.
29459 (arm_get_frame_offsets): Split up into this ...
29460 (arm_compute_frame_layout): ... and this function.
29462 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
29464 * config/aarch64/constraints.md (Usa): New constraint.
29465 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
29467 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
29469 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
29470 with_multilib_list after it has been checked.
29472 2017-05-08 Richard Biener <rguenther@suse.de>
29474 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
29475 (bitmap_set_subtract_values): Likewise.
29477 2017-05-08 Richard Biener <rguenther@suse.de>
29479 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
29480 (gimple_assign_nonzero): ... this and remove strict_overflow_p
29482 (gimple_stmt_nonzero_warnv_p): Rename to ...
29483 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
29485 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
29486 (extract_range_basic): Adjust, do not disable propagation on
29487 strict overflow sensitive simplification.
29488 (vrp_visit_cond_stmt): Likewise.
29490 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
29492 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
29493 body size unconditionally.
29495 2017-05-07 Jeff Law <law@redhat.com>
29498 2017-05-06 Jeff Law <law@redhat.com>
29499 PR tree-optimization/78496
29500 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
29503 PR tree-optimization/78496
29504 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
29505 (simplify_stmt_using_ranges): Call it.
29506 (vrp_dom_walker::before_dom_children): Extract equivalences
29507 from an ASSERT_EXPR with an equality comparison against a
29510 2017-05-06 Jeff Law <law@redhat.com>
29512 PR tree-optimization/78496
29513 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
29516 PR tree-optimization/78496
29517 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
29518 (simplify_stmt_using_ranges): Call it.
29519 (vrp_dom_walker::before_dom_children): Extract equivalences
29520 from an ASSERT_EXPR with an equality comparison against a
29523 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
29525 * lra-constraints.c (lra_copy_reg_equiv): New function.
29526 (split_reg): Use it to copy equivalence information from the
29527 original register to the spill register.
29529 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
29531 PR rtl-optimization/75964
29532 * simplify-rtx.c (simplify_const_relational_operation): Remove
29533 invalid handling of comparisons of integer ABS.
29535 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
29537 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
29538 initialize to zero.
29539 (init_regs): Remove declaration.
29540 (function_arg_advance_32): Initialize error_p as boolean variable.
29542 2017-05-05 Nathan Sidwell <nathan@acm.org>
29544 * store-motion.c (remove_reachable_equiv_notes): Reformat long
29545 lines. Use for (;;).
29547 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29549 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
29550 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
29551 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
29552 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
29553 VF=2 that require versioning.
29555 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29557 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
29560 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29562 * diagnostic.h (diagnostic_override_option_index): Convert from
29563 macro to inline function.
29565 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29567 * diagnostic.c (last_module_changed_p): New function.
29568 (set_last_module): New function.
29569 (diagnostic_report_current_module): Convert macro usage to
29570 the above functions.
29571 * diagnostic.h (diagnostic_context::last_module): Strengthen
29572 from const line_map * to const line_map_ordinary *.
29573 (diagnostic_last_module_changed): Delete macro.
29574 (diagnostic_set_last_module): Delete macro.
29576 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29578 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
29579 with diagnostic_report_diagnostic.
29580 (diagnostic_n_impl_richloc): Likewise.
29581 * diagnostic.h (report_diagnostic): Delete macro.
29582 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
29583 with diagnostic_report_diagnostic.
29584 * substring-locations.c (format_warning_va): Likewise.
29586 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29588 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
29589 save/restor of format_spec. Move option-printing code to...
29590 (print_option_information): ...this new function, and
29591 reimplement by simply printing to the pretty_printer,
29592 rather than appending to the format string.
29594 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29596 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
29597 handling logic into...
29598 (update_effective_level_from_pragmas): ...this new function.
29600 2017-05-04 Andrew Waterman <andrew@sifive.com>
29602 * config/riscv/riscv.opt (mstrict-align): New option.
29603 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
29604 (SLOW_UNALIGNED_ACCESS): Define.
29605 (riscv_slow_unaligned_access): Declare.
29606 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
29608 (riscv_slow_unaligned_access): New variable.
29609 (rocket_tune_info): Set slow_unaligned_access to true.
29610 (optimize_size_tune_info): Set slow_unaligned_access to false.
29611 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
29612 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
29613 (riscv_option_override): Set riscv_slow_unaligned_access.
29614 * doc/invoke.texi: Add -mstrict-align to RISC-V.
29616 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
29618 * config/riscv/riscv.md: Unify indentation.
29620 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29625 * config/rs6000/rs6000.md (u code attribute): Add FIX and
29627 (extendsi<mode>2): Add support for doing sign extension via
29628 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
29629 don't have ISA 3.0 instructions.
29630 (extendsi<mode>2 splitter): Likewise.
29631 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
29632 generate the normal insns since SImode can now go in vector
29633 registers. Disallow the special UNSPECs needed for previous
29634 machines to hide SImode being used. Add new insns
29635 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
29636 (fix_trunc<mode>si2_stfiwx): Likewise.
29637 (fix_trunc<mode>si2_internal): Likewise.
29638 (fixuns_trunc<mode>si2): Likewise.
29639 (fixuns_trunc<mode>si2_stfiwx): Likewise.
29640 (fctiw<u>z_<mode>_smallint): Likewise.
29641 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
29642 of floating point to 32-bit integer from doing a direct move to
29643 the GPR registers to do a store.
29644 (fctiwz_<mode>): Break long line.
29646 2017-05-05 Bin Cheng <bin.cheng@arm.com>
29648 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
29649 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
29650 (addr_list, addr_offset_valid_p): New.
29651 (split_address_groups): Check offset validity with above function.
29652 (gt-tree-ssa-loop-ivopts.h): Include header file.
29654 2017-05-05 Nathan Sidwell <nathan@acm.org>
29656 * config.gcc (arm*-*-*): Add missing 'fi'.
29658 2017-05-05 Steve Ellcey <sellcey@cavium.com>
29660 * doc/invoke.texi (-fopt-info): Explicitly say order of options
29661 included in -fopt-info does not matter.
29662 * doc/optinfo.texi (-fopt-info): Fix description of default
29663 behavour. Explicitly say order of options included in -fopt-info
29666 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29668 * config.gcc: Allow combinations of aprofile and rmprofile values for
29669 --with-multilib-list.
29670 * config/arm/t-multilib: New file.
29671 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
29672 variables. Remove setting of ISA and floating-point ABI in
29673 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
29674 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
29675 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
29676 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
29678 * config/arm/t-rmprofile: Likewise except for the matches changes.
29679 * doc/install.texi (--with-multilib-list): Document the combination of
29680 aprofile and rmprofile values and warn about pitfalls in doing that.
29682 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29684 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
29685 (movdi_aarch64): Likewise.
29687 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29689 PR tree-optimization/80632
29690 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
29692 (build_arrays): Initialize it for virtual phis.
29693 (fix_phi_nodes): Use it for virtual phis.
29695 PR tree-optimization/80558
29696 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
29697 [x, y] op z into [x op, y op z] for op & or | if conditions
29700 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
29701 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29704 * config/arm/arm.md (use_literal_pool): Remove.
29705 (64-bit immediate split): No longer takes cost into consideration
29706 if arm_disable_literal_pool is enabled.
29707 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
29708 used when arm_disable_literal_pool is enabled.
29709 (arm_max_const_double_inline_cost): Remove use of
29710 arm_disable_literal_pool.
29711 (push_minipool_fix): Add assert.
29712 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
29713 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
29714 (no_literal_pool_sf_immediate): New.
29716 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29718 PR tree-optimization/80613
29719 * tree-ssa-dce.c (propagate_necessity): Remove cases for
29720 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
29722 2017-05-05 Richard Biener <rguenther@suse.de>
29724 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
29726 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
29728 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
29729 of this flag from insn conditions due to removal from r247495.
29731 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29733 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
29735 (arm_early_store_addr_dep_ptr): Likewise.
29736 * config/arm/aarch-common-protos.h
29737 (arm_early_load_addr_dep_ptr): Add prototype.
29738 (arm_early_store_addr_dep_ptr): Likewise.
29739 * config/arm/cortex-a53.md: Add new bypasses.
29741 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29743 * tree.c (next_type_uid): Change type to unsigned.
29744 (type_hash_canon): Decrement back next_type_uid if
29745 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
29746 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
29749 2017-05-04 Martin Sebor <msebor@redhat.com>
29751 * builtins.c: Fix a trivial typo in a comment.
29753 PR middle-end/79234
29754 * builtins.c (check_sizes): Adjust to handle reading past the end.
29755 Avoid printing excessive upper bound of ranges. Use %E to print
29756 tree nodes instead of converting them to %wu.
29757 (expand_builtin_memchr): New function.
29758 (compute_dest_size): Rename...
29759 (compute_objsize): ...to this.
29760 (expand_builtin_memcpy): Adjust.
29761 (expand_builtin_mempcpy): Adjust.
29762 (expand_builtin_strcat): Adjust.
29763 (expand_builtin_strcpy): Adjust.
29764 (check_strncat_sizes): Adjust.
29765 (expand_builtin_strncat): Adjust.
29766 (expand_builtin_strncpy): Adjust and simplify.
29767 (expand_builtin_memset): Adjust.
29768 (expand_builtin_bzero): Adjust.
29769 (expand_builtin_memcmp): Adjust.
29770 (expand_builtin): Handle memcmp.
29771 (maybe_emit_chk_warning): Check strncat just once.
29773 2017-05-04 Martin Sebor <msebor@redhat.com>
29775 PR preprocessor/79214
29776 PR middle-end/79222
29777 PR middle-end/79223
29778 * builtins.c (check_sizes): Add inlining context and issue
29779 warnings even when -Wno-system-headers is set.
29780 (check_strncat_sizes): Same.
29781 (expand_builtin_strncat): Same.
29782 (expand_builtin_memmove): New function.
29783 (expand_builtin_stpncpy): Same.
29784 (expand_builtin): Handle memmove and stpncpy.
29786 2017-05-04 Bin Cheng <bin.cheng@arm.com>
29788 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
29789 which is not used any more.
29791 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29793 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
29795 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29797 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
29798 (cortexa53_tunings): Likewise.
29799 (cortexa57_tunings): Likewise.
29800 (cortexa72_tunings): Likewise.
29801 (cortexa73_tunings): Likewise.
29803 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29805 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
29806 Set loop alignment to 8.
29808 2017-05-04 Martin Sebor <msebor@redhat.com>
29810 PR translation/80280
29811 * builtins.c (expand_builtin_object_size): Add missing quoting to
29812 %D and like directives.
29813 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
29814 (hsa_type_for_tree_type): Same.
29815 (verify_function_arguments): Same.
29816 * symtab.c (symbol_table::change_decl_assembler_name): Same.
29817 * varasm.c (get_section): Same.
29820 2017-05-04 Martin Sebor <msebor@redhat.com>
29822 PR translation/80280
29823 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
29825 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29827 * config/aarch64/aarch64.c (generic_addrcost_table):
29828 Change HI/TI mode setting.
29830 2017-05-04 Martin Jambor <mjambor@suse.cz>
29832 PR tree-optimization/80622
29833 * tree-sra.c (comes_initialized_p): New function.
29834 (build_accesses_from_assign): Only set write lazily when
29835 comes_initialized_p is false.
29836 (analyze_access_subtree): Use comes_initialized_p.
29837 (propagate_subaccesses_across_link): Assert !comes_initialized_p
29838 instead of testing for PARM_DECL.
29840 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29842 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
29843 constraint on operand 0 to allow more general addressing modes.
29844 Adjust output template.
29845 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
29847 * config/aarch64/aarch64-protos.h
29848 (aarch64_address_valid_for_prefetch_p): Declare prototype.
29849 * config/aarch64/constraints.md (Dp): New address constraint.
29850 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
29853 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
29855 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
29856 update use of estimate_ipcp_clone_size_and_time.
29857 (estimate_local_effects): Update use of
29858 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
29859 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
29860 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
29861 Return nonspecialized time.
29863 2017-05-04 Richard Biener <rguenther@suse.de>
29865 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
29866 for the last VUSE which def dominates the PHI. Directly call
29868 (get_continuation_for_phi_1): Remove.
29870 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
29872 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
29873 to explain the use of truncating division. Cap the number of
29874 iterations to the maximum given by nb_iterations_upper_bound,
29877 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
29879 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
29880 * configure: Regenerate.
29881 * config.in: Regenerate.
29882 * config/i386/driver-mingw32.c: new file.
29883 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
29884 * config.host: Link driver-mingw32.o on MinGW host.
29885 * doc/install.texi: Document new --enable-mingw-wildcard configure
29888 2017-05-04 Marek Polacek <polacek@redhat.com>
29890 PR tree-optimization/80612
29891 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
29893 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29894 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
29896 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
29897 (movt splitter): Likewise.
29898 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
29899 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
29900 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
29901 block for Thumb-1 with MOVT.
29902 (thumb2_legitimate_address_p): Move code block ...
29903 (can_avoid_literal_pool_for_label_p): ... into this new function.
29904 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
29906 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
29907 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
29908 "M-profile targets with the MOVT instruction".
29910 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29912 * config/arm/arm-builtins.c (arm_init_builtins): Rename
29913 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
29914 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
29916 2017-05-04 Martin Liska <mliska@suse.cz>
29918 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
29919 variable cond_code.
29921 2017-05-04 Richard Biener <rguenther@suse.de>
29923 * tree.c (array_at_struct_end_p): Handle arrays at struct
29924 end with flexarrays more conservatively. Refactor and treat
29925 arrays of arrays or aggregates more strict. Fix
29926 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
29927 * tree.c (array_at_struct_end_p): Adjust prototype.
29928 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
29929 * gimple-fold.c (get_range_strlen): Likewise.
29930 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
29932 2017-05-04 Richard Biener <rguenther@suse.de>
29934 PR tree-optimization/31130
29935 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
29937 (supports_overflow_infinity): Likewise.
29938 (is_negative_overflow_infinity): Likewise.
29939 (is_positive_overflow_infinity): Likewise.
29940 (is_overflow_infinity): Likewise.
29941 (stmt_overflow_infinity): Likewise.
29942 (overflow_infinity_range_p): Likewise.
29943 (usable_range_p): Remove as always returning true.
29944 (make_overflow_infinity): Remove.
29945 (negative_overflow_infinity): Likewise.
29946 (positive_overflow_infinity): Likewise.
29947 (avoid_overflow_infinity): Likewise.
29948 (set_value_range): Adjust accordingly.
29949 (set_value_range_to_nonnegative): Likewise, remove now unused
29950 overflow_infinity arg.
29951 (vrp_operand_equal_p): Adjust.
29952 (update_value_range): Likewise.
29953 (range_int_cst_singleton_p): Likewise.
29954 (operand_less_p): Likewise.
29955 (compare_values_warnv): Likewise.
29956 (extract_range_for_var_from_comparison_expr): Likewise.
29957 (vrp_int_const_binop): Likewise.
29958 (zero_nonzero_bits_from_vr): Likewise.
29959 (extract_range_from_multiplicative_op_1): Likewise.
29960 (extract_range_from_binary_expr_1): Likewise.
29961 (extract_range_from_unary_expr): Likewise.
29962 (extract_range_from_comparison): Likewise.
29963 (extract_range_basic): Likewise.
29964 (adjust_range_with_scev): Likewise.
29965 (compare_ranges): Likewise.
29966 (compare_range_with_value): Likewise.
29967 (dump_value_range): Likewise.
29968 (test_for_singularity): Likewise, remove strict_overflow_p parameter
29970 (simplify_cond_using_ranges): Adjust.
29972 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
29974 * brig-builtins.def: Added a builtin for class_f64.
29975 * builtin-types.def: Added a builtin type needed by class_f64.
29977 2017-05-03 Jason Merrill <jason@redhat.com>
29979 * timevar.def: Add TV_CONSTEXPR.
29981 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29983 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
29985 2017-05-03 Martin Jambor <mjambor@suse.cz>
29987 * ipa-prop.c (ipa_update_after_lto_read): Removed.
29988 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
29989 * ipa-cp.c (ipcp_propagate_stage): Do not call
29990 ipa_update_after_lto_read.
29991 * ipa-inline.c (ipa_inline): Likewise.
29993 2017-05-03 Martin Jambor <mjambor@suse.cz>
29995 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
29996 tag. Added a default constructor and a destructor.
29997 (ipa_edge_args_sum_t): New class;
29998 (ipa_edge_args_sum): Declare.
29999 (ipa_edge_args_vector): Remove declaration.
30000 (IPA_EDGE_REF): Use ipa_edge_args_sum.
30001 (ipa_free_edge_args_substructures): Remove declaration.
30002 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
30003 (ipa_edge_args_info_available_for_edge_p): Likewise.
30004 * ipa-prop.c (ipa_edge_args_vector): Removed.
30005 (edge_removal_hook_holder): Likewise.
30006 (edge_duplication_hook_holder): Likewise.
30007 (ipa_edge_args_sum): New variable.
30008 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
30009 ipa_edge_args_vector.
30010 (ipa_free_edge_args_substructures): Likewise.
30011 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
30012 ipa_edge_args_vector.
30013 (ipa_edge_removal_hook): Turned into method
30014 ipa_edge_args_sum_t::remove.
30015 (ipa_edge_duplication_hook): Turned into method
30016 ipa_edge_args_sum_t::duplicate.
30017 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
30018 registering edge hooks.
30019 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
30020 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
30021 ipa_edge_args_sum instead of ipa_edge_args_vector.
30022 * ipa-profile.c (ipa_profile): Likewise.
30024 2017-05-03 Martin Jambor <mjambor@suse.cz>
30026 * symbol-summary.h (function_summary): New method exists.
30027 (function_summary::symtab_removal): Deallocate through release.
30028 (call_summary): New class.
30029 (gt_ggc_mx): New overload.
30030 (gt_pch_nx): Likewise.
30031 (gt_pch_nx): Likewise.
30033 2017-05-03 Jeff Law <law@redhat.com>
30035 PR tree-optimization/78496
30036 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
30037 from simplify_cond_using_ranges. Split off code to walk
30038 backwards through casts into ...
30039 (simplify_cond_using_ranges_2): New function.
30040 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
30041 (execute_vrp): After identifying jump threads, call
30042 simplify_cond_using_ranges_2.
30044 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
30047 * ipa-inline.h (inline_summary): Add ctor.
30048 (create_ggc): Do not use ggc_cleared_alloc.
30050 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
30051 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
30053 * gcc.c (handle_braces): Support escaping in switch matching
30055 * doc/invoke.texi (Spec Files): Document it.
30056 Remove superfluous @code markup in items.
30058 2017-05-03 David Malcolm <dmalcolm@redhat.com>
30060 * diagnostic-show-locus.c (struct column_range): New struct.
30061 (get_affected_columns): New function.
30062 (get_printed_columns): New function.
30063 (struct correction): New struct.
30064 (correction::ensure_capacity): New function.
30065 (correction::ensure_terminated): New function.
30066 (struct line_corrections): New struct.
30067 (line_corrections::~line_corrections): New dtor.
30068 (line_corrections::add_hint): New function.
30069 (layout::print_trailing_fixits): Reimplement in terms of the new
30071 (selftest::test_overlapped_fixit_printing): New function.
30072 (selftest::diagnostic_show_locus_c_tests): Call it.
30074 2017-05-03 Nathan Sidwell <nathan@acm.org>
30076 Canonicalize canonical type hashing
30077 * tree.h (type_hash_canon_hash): Declare.
30078 * tree.c (type_hash_list, attribute_hash_list): Move into
30079 type_hash_canon_hash.
30080 (build_type_attribute_qual_variant): Break out hash code calc into
30081 type_hash_canon_hash.
30082 (type_hash_canon_hash): New. Generic type hash computation.
30083 (build_range_type_1, build_array_type_1, build_function_type,
30084 build_method_type_directly, build_offset_type, build_complex_type,
30085 make_vector_type): Call it.
30087 2017-05-03 Richard Biener <rguenther@suse.de>
30089 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
30090 When all DRs have unknown misaligned do not always peel
30091 when there is a store but apply the same costing model as if
30092 there were only loads.
30094 2017-05-03 Richard Biener <rguenther@suse.de>
30097 PR tree-optimization/80492
30098 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30099 compare_base_decls returning dont-know properly.
30101 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
30103 * config/arm/iterators.md (CCSI): New mode iterator.
30104 (arch): New mode attribute.
30105 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
30106 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
30107 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
30108 code iterator for success result mode.
30109 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
30110 the corresponding new insn generators.
30112 2017-05-03 Bin Cheng <bin.cheng@arm.com>
30115 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30116 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30118 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
30120 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
30121 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
30122 (DDR_A): Wrap DDR argument in brackets.
30123 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
30124 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
30125 (DDR_REVERSED_P): Likewise.
30127 2017-05-03 Jakub Jelinek <jakub@redhat.com>
30129 PR tree-optimization/79472
30130 * tree-switch-conversion.c (struct switch_conv_info): Add
30131 contiguous_range and default_case_nonstandard fields.
30132 (collect_switch_conv_info): Compute contiguous_range and
30133 default_case_nonstandard fields, don't clear final_bb if
30134 contiguous_range and only the default case doesn't have the required
30136 (check_all_empty_except_final): Set default_case_nonstandard instead
30137 of failing if contiguous_range and the default case doesn't have empty
30139 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
30140 and only the default case doesn't have the required constants. Skip
30142 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
30143 if default_case_nonstandard.
30144 (build_constructors): Build constant 1 just once. Assert that default
30145 values aren't inserted in between cases if contiguous_range. Skip
30147 (build_arrays): Skip virtual phis.
30148 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
30149 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
30150 Handle virtual phis.
30151 (gen_inbound_check): Handle default_case_nonstandard case.
30152 (process_switch): Adjust check_final_bb caller. Call
30153 gather_default_values with the first non-default case instead of
30154 default case if default_case_nonstandard.
30156 2017-05-02 Nathan Sidwell <nathan@acm.org>
30158 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
30159 check. Fix formatting.
30161 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
30163 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
30164 errors when comparing specialized and unspecialized times.
30166 2017-05-02 David Malcolm <dmalcolm@redhat.com>
30168 * diagnostic-show-locus.c
30169 (layout::should_print_annotation_line_p): Make private.
30170 (layout::print_annotation_line): Make private.
30171 (layout::annotation_line_showed_range_p): Make private.
30172 (layout::show_ruler): Make private.
30173 (layout::print_source_line): Make private. Pass in line and
30174 line_width, rather than calling location_get_source_line. Drop
30176 (layout::print_leading_fixits): New method.
30177 (layout::print_any_fixits): Rename to...
30178 (layout::print_trailing_fixits): ...this, and make private.
30179 Don't print newline fixits.
30180 (diagnostic_show_locus): Move logic for printing one row into...
30181 (layout::print_line): ...this new function. Move the
30182 location_get_source_line call and error-handling from
30183 print_source_line to here. Call print_leading_fixits, and rename
30184 print_any_fixits to print_trailing_fixits.
30185 (selftest::test_fixit_insert_containing_newline): Update now that
30186 newlines are partially supported.
30187 (selftest::test_fixit_insert_containing_newline_2): New test.
30188 (selftest::test_fixit_replace_containing_newline): Update comments.
30189 (selftest::diagnostic_show_locus_c_tests): Call the new test.
30190 * edit-context.c (class added_line): New class.
30191 (class edited_line): Describe newline handling in comment.
30192 (edited_line::actually_edited_p): New method.
30193 (edited_line::print_content): Delete redundant decl.
30194 (edited_line::m_predecessors): New field.
30195 (edited_file::print_content): Call edited_line::print_content.
30196 (edited_file::print_diff): Update to support newlines.
30197 (edited_file::print_diff_hunk): Likewise.
30198 (edited_file::print_run_of_changed_lines): New function.
30199 (edited_file::print_diff_line): Convert to...
30200 (print_diff_line): ...this.
30201 (edited_file::get_effective_line_count): New function.
30202 (edited_line::edited_line): Initialize new field m_predecessors.
30203 (edited_line::~edited_line): Clean up m_predecessors.
30204 (edited_line::apply_fixit): Handle newlines.
30205 (edited_line::get_effective_line_count): New function.
30206 (edited_line::print_content): New function.
30207 (edited_line::print_diff_lines): New function.
30208 (selftest::test_applying_fixits_insert_containing_newline): New
30210 (selftest::test_applying_fixits_replace_containing_newline): New
30212 (selftest::insert_line): New function.
30213 (selftest::test_applying_fixits_multiple_lines): Add example of
30215 (selftest::edit_context_c_tests): Call the new tests.
30217 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30219 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
30220 parameter cand. Update dump information.
30221 (get_computation_cost): Update uses.
30223 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30225 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
30226 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
30227 (get_computation_at, rewrite_use_address): Update use of
30228 get_computation_aff.
30230 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30232 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
30233 (get_computation): Delete.
30234 (get_computation_cost): Implement like get_computation_cost_at.
30235 Use get_computation_at.
30236 (get_computation_cost_at): Delete.
30237 (rewrite_use_nonlinear_expr): Use get_computation_at.
30238 (rewrite_use_compare, remove_unused_ivs): Ditto.
30240 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30242 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
30244 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30246 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
30247 (ivopts_global_cost_for_size): Rename parameter and update uses.
30248 (iv_ca_recount_cost): Update uses.
30249 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
30250 candidates seperately in n_invs and n_cands.
30251 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
30253 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30255 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
30256 (find_inv_vars_cb): New.
30257 (find_depends): Renamed to ...
30258 (find_inv_vars): ... this.
30259 (add_candidate_1, force_var_cost): Call find_inv_vars.
30260 (split_address_cost, determine_group_iv_cost_cond): Ditto.
30262 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30264 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
30265 inv_vars. Add inv_exprs.
30266 (struct iv_cand): Rename depends_on to inv_vars.
30267 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
30268 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
30269 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
30270 (dump_cand): Dump inv_vars.
30271 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
30272 (record_invariant, find_depends, add_candidate_1): Ditto.
30273 (set_group_iv_cost, force_var_cost): Ditto.
30274 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
30275 (get_computation_cost_at, get_computation_cost): Ditto.
30276 (determine_group_iv_cost_generic): Ditto.
30277 (determine_group_iv_cost_address): Ditto.
30278 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
30279 (determine_group_iv_costs): Ditto.
30280 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
30281 (iv_ca_set_remove_invariants): Renamed to ...
30282 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
30283 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
30284 (iv_ca_set_add_invariants): Renamed to ...
30285 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
30286 (iv_ca_set_cp): Use iv_ca_set_add_invs.
30287 (iv_ca_has_deps): Support inv_vars and inv_exprs.
30288 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
30289 (create_new_ivs): Remove useless dump.
30291 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30293 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
30295 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
30296 (iv_ca_set_no_cp, create_new_iv): Ditto.
30298 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30300 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30302 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
30304 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
30305 function tree_check2.
30307 2017-05-02 Martin Liska <mliska@suse.cz>
30309 * doc/gcov.texi: Add missing preposition.
30310 * gcov.c (function_info::function_info): Properly fill up
30311 all member variables.
30313 2017-05-02 Tamar Christina <tamar.christina@arm.com>
30315 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
30317 2017-05-02 Tamar Christina <tamar.christina@arm.com>
30319 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
30321 2017-05-02 Martin Liska <mliska@suse.cz>
30324 * lto-streamer-in.c (lto_read_tree_1): Remove
30325 LTO_STREAMER_DEBUG.
30326 * lto-streamer.c (struct tree_hash_entry): Likewise.
30327 (struct tree_entry_hasher): Likewise.
30328 (tree_entry_hasher::hash): Likewise.
30329 (tree_entry_hasher::equal): Likewise.
30330 (lto_streamer_init): Likewise.
30331 (lto_orig_address_map): Likewise.
30332 (lto_orig_address_get): Likewise.
30333 (lto_orig_address_remove): Likewise.
30334 * lto-streamer.h: Likewise.
30335 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
30336 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
30338 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
30340 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
30341 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
30342 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
30343 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
30344 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
30345 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
30346 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
30347 (mm_maskz_sub_ss): New intrinsics.
30348 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30349 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
30350 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
30351 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
30352 (__builtin_ia32_subss_mask_round): New builtins.
30353 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30354 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30355 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
30357 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
30358 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
30360 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
30363 2017-05-02 Martin Jambor <mjambor@suse.cz>
30365 PR tree-optimization/78687
30366 * tree-sra.c (access): New field parent.
30367 (process_subtree_disqualification): New function.
30368 (disqualify_candidate): Call it.
30369 (build_accesses_from_assign): Reset write flag if creating an
30371 (build_access_subtree): Fill in parent field and also prpagate
30372 down grp_write flag.
30373 (create_artificial_child_access): New parameter set_grp_write, set
30374 grp_write to its value.
30375 (propagate_subaccesses_across_link): Also propagate grp_write flag
30377 (propagate_all_subaccesses): Push the closest parent back to work
30378 queue if add_access_to_work_queue returned true.
30380 2017-05-02 Richard Biener <rguenther@suse.de>
30382 * common.opt (fstrict-overflow): Alias negative to fwrapv.
30383 * doc/invoke.texi (fstrict-overflow): Remove all traces of
30384 -fstrict-overflow documentation.
30385 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
30386 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
30387 flag_strict_overflow.
30388 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
30389 * lto-opts.c (lto_write_options): Do not stream it.
30390 * lto-wrapper.c (merge_and_complain): Do not handle it.
30391 * opts.c (default_options_table): Do not set -fstrict-overflow.
30392 (finish_options): Likewise do not clear it when sanitizing.
30393 * simplify-rtx.c (simplify_const_relational_operation): Do not
30394 test flag_strict_overflow.
30396 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
30398 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
30399 using enabled attribute.
30400 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
30401 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
30402 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
30403 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
30404 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
30405 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
30406 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
30407 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
30408 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
30409 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
30411 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
30413 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
30415 2017-05-02 Richard Biener <rguenther@suse.de>
30417 PR tree-optimization/80591
30419 2017-04-10 Richard Biener <rguenther@suse.de>
30421 * tree-ssa-structalias.c (find_func_aliases): Properly handle
30424 2017-05-02 Richard Biener <rguenther@suse.de>
30426 PR tree-optimization/80549
30427 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
30428 (cleanup_tree_cfg_noloop): Create forwarders to known loop
30429 headers if they do not have a preheader.
30431 2017-05-02 Martin Liska <mliska@suse.cz>
30434 * common.opt: Fix typo.
30435 * doc/invoke.texi: Likewise.
30437 2017-05-01 Jan Beulich <jbeulich@suse.com>
30439 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
30440 swapping, add (x,x,m,x,n) alternative.
30442 2017-05-01 Nathan Sidwell <nathan@acm.org>
30444 * calls.c (combine_pending_stack_adjustment_and_call): Remove
30445 unnecessary unadjusted_alignment check.
30447 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
30450 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
30452 * gimplify.c (gimplify_cilk_detach): New function.
30453 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
30454 * tree-core.h: Document EXPR_CILK_SPAWN.
30455 * tree.h (EXPR_CILK_SPAWN): Define.
30457 2017-05-01 David Malcolm <dmalcolm@redhat.com>
30459 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
30460 to use new fixit_hint representation, using the "replace" logic.
30461 (get_line_span_for_fixit_hint): Likewise.
30462 (layout::print_any_fixits): Likewise.
30463 (selftest::test_one_liner_many_fixits): Rename to...
30464 (selftest::test_one_liner_many_fixits_1): ...this, and update
30465 comment and expected output to reflect that the multiple fix-it
30466 hints are now consolidated into one insertion.
30467 (selftest::test_one_liner_many_fixits_2): New test.
30468 (selftest::test_diagnostic_show_locus_one_liner): Update for
30470 (selftest::test_fixit_consolidation): Update for fix-it API
30472 * diagnostic.c (print_parseable_fixits): Likewise.
30473 * edit-context.c (edited_line::m_line_events): Convert from
30474 auto_vec <line_event *> to auto_vec <line_event>.
30475 (class line_event): Convert from abstract base class to a concrete
30476 class, taking over the role of replace_event.
30477 (class insert_event): Delete.
30478 (class replace_event): Rename to class line_event. Convert to
30480 (edit_context::add_fixits): Reimplement.
30481 (edit_context::apply_insert): Delete.
30482 (edit_context::apply_replace): Rename to...
30483 (edit_context::apply_fixit): ...this. Convert to half-open range.
30484 (edited_file::apply_insert): Delete.
30485 (edited_file::apply_replace): Rename to...
30486 (edited_file::apply_fixit): ...this.
30487 (edited_line::~edited_line): Drop deletion of events.
30488 (edited_line::apply_insert): Delete.
30489 (edited_line::apply_replace): Rename to...
30490 (edited_line::apply_fixit): ...this. Convert to half-open range.
30491 Update for change to type of m_line_events.
30492 * edit-context.h (edit_context::apply_insert): Delete.
30493 (edit_context::apply_replace): Rename to...
30494 (edit_context::apply_fixit): ...this.
30496 2017-05-01 Martin Sebor <msebor@redhat.com>
30498 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
30501 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
30504 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
30505 __get_cpuid_max returns 0.
30506 (__get_cpuid_count): Ditto.
30508 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
30510 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
30511 replacement expression is another instance of one of its arguments.
30513 2017-05-01 Jakub Jelinek <jakub@redhat.com>
30516 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
30517 check for stack push/pop autoinc.
30518 * config/i386/i386.c (ix86_agi_dependent): Return false
30519 if the only reason why modified_in_p returned true is that
30520 addr is SP based and set_insn is a push or pop.
30522 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
30524 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
30527 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
30530 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
30531 (account_size_time): Use two predicates - exec_pred and
30533 (evaluate_conditions_for_known_args): Compute both clause and
30535 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
30536 (inline_summary_t::duplicate): Update.
30537 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
30539 (compute_inline_parameters): Likewise.
30540 (estimate_edge_size_and_time): Update caluclation of time.
30541 (estimate_node_size_and_time): Compute both time and nonspecialized
30543 (estimate_ipcp_clone_size_and_time): Update.
30544 (inline_merge_summary): Update.
30545 (do_estimate_edge_time): Update.
30546 (do_estimate_edge_size): Update.
30547 (do_estimate_edge_hints): Update.
30548 (inline_read_section, inline_write_summary): Stream both new predicates.
30549 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
30551 (compute_inlined_call_time): Cleanup.
30552 (big_speedup_p): Update.
30553 (edge_badness): Update.
30554 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
30555 (size_time_entry): Replace predicate by exec_predicate and
30556 nonconst_predicate.
30557 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
30558 (estimate_edge_time): Return also nonspec_time.
30559 (reset_edge_growth_cache): Update.
30561 2017-04-29 Jakub Jelinek <jakub@redhat.com>
30563 PR rtl-optimization/80491
30564 * ifcvt.c (noce_process_if_block): When looking for x setter
30565 with missing else_bb, don't check only the insn right before
30566 cond_earliest, but look for the last insn that x is modified in
30567 within the same bb.
30569 PR rtl-optimization/80491
30570 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
30572 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
30574 PR tree-optimization/80487
30575 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
30577 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30579 PR tree-optimization/79697
30580 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
30581 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
30582 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
30584 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
30585 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
30587 2017-04-28 Martin Sebor <msebor@redhat.com>
30589 PR tree-optimization/80523
30590 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
30591 (init_target_to_host_charmap, target_to_host, target_strtol10): New
30593 (maybe_warn, format_directive, parse_directive): Use new functions.
30594 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
30596 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
30598 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
30600 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
30602 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
30603 target_header_dir): Set correctly.
30604 * configure: Regenerated.
30605 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30606 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30607 instead of SYSTEM_HEADER_DIR.
30609 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
30611 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
30612 (estimate_local_effects): Likewise.
30613 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
30614 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
30615 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
30616 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
30617 do_estimate_edge_time, estimate_edge_time): Likewise.
30618 * ipa-inline-analysis.c (estimate_node_size_and_time,
30619 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
30620 (estimate_time_after_inlining): Remove.
30622 2017-04-28 Martin Liska <mliska@suse.cz>
30624 * doc/gcov.texi: Enhance documentation of gcov.
30626 2017-04-28 Martin Liska <mliska@suse.cz>
30628 * doc/gcov.texi: Sort options in alphabetic order.
30629 * doc/gcov-dump.texi: Likewise.
30630 * doc/gcov-tool.texi: Likewise.
30631 * gcov.c (print_usage): Likewise.
30632 * gcov-dump.c (print_usage): Likewise.
30633 * gcov-tool.c (print_merge_usage_message): Likewise.
30634 (print_rewrite_usage_message): Likewise.
30635 (print_overlap_usage_message): Likewise.
30637 2017-04-28 Martin Liska <mliska@suse.cz>
30639 PR gcov-profile/53915
30640 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
30642 2017-04-28 Martin Liska <mliska@suse.cz>
30644 PR gcov-profile/79891
30645 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
30646 is marked by compiler as living on a line.
30647 (get_cycles_count): Remove usage of the union.
30648 (output_intermediate_file): Likewise.
30649 (find_source): Fix GNU coding style.
30650 (accumulate_line_counts): Remove old non-all block mode.
30651 (output_lines): Remove usage of the union.
30652 * profile.c (output_location): Include all BBs, even if
30653 belonging to a same line (and file) as a previous BB.
30655 2017-04-28 Martin Liska <mliska@suse.cz>
30657 * gcov.c (process_args): Handle new argument 'w'.
30658 (read_graph_file): Assign ID to BBs.
30659 (output_branch_count): Display BB # if verbose flag is set.
30660 (output_lines): Likewise for arcs.
30661 (print_usage): Add '--verbose' option help.
30662 * doc/gcov.texi: Document --verbose (-w) option.
30664 2017-04-28 Martin Liska <mliska@suse.cz>
30666 * gcov.c (struct block_location_info): New struct.
30667 (process_file): Fill up the new structure.
30668 (read_graph_file): Replace usage of encoding by the newly added
30670 (add_line_counts): Likewise.
30671 (accumulate_line_counts): Remove usage of the union.
30672 (function_info::function_info): New function.
30673 (function_info::~function_info): Likewise.
30674 (process_file): Call delete instead of release_function.
30675 (release_function): Release the function.
30676 (release_structures): Call delete instead of release_function.
30677 (solve_flow_graph): Replace usage of num_blocks.
30678 (find_exception_blocks): Likewise.
30679 (output_lines): Fix GNU coding style.
30681 2017-04-28 Martin Liska <mliska@suse.cz>
30684 * coverage.c (coverage_remove_note_file): New function.
30685 * coverage.h: Declare the function.
30686 * toplev.c (finalize): Clean if an error has been seen.
30688 2017-04-28 Martin Liska <mliska@suse.cz>
30690 PR gcov-profile/80031
30691 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
30692 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
30693 * gcov.c (read_graph_file): Read just number of blocks.
30694 * profile.c (branch_prob): Do not stream 0 flags per a basic
30697 2017-04-28 Martin Liska <mliska@suse.cz>
30699 * gcov-dump.c (tag_*): Add new argument to declarations.
30700 (dump_gcov_file): Likewise.
30701 (tag_blocks): Add and use new argument depth.
30702 (tag_arcs): Likewise.
30703 (tag_lines): Likewise.
30704 (tag_counters): Likewise.
30705 (tag_summary): Likewise.
30706 (dump_working_sets): Use depth to do a proper indentation.
30708 2017-04-28 Jakub Jelinek <jakub@redhat.com>
30711 * cgraph.h (symtab_node::debug_symtab): No longer inline.
30712 * symtab.c (symtab_node::debug_symtab): Move definition here.
30714 2017-04-28 Richard Biener <rguenther@suse.de>
30716 * lto-streamer.h (LTO_major_version): Bump to 7.
30718 2017-04-28 Richard Biener <rguenther@suse.de>
30720 * tree-vrp.c (assert_info): New struct.
30721 (add_assert_info): New helper.
30722 (register_edge_assert_for_2): Refactor to add asserts to a vector
30724 (register_edge_assert_for_1): Likewise.
30725 (register_edge_assert_for): Likewise.
30726 (finish_register_edge_assert_for): New helper actually registering
30727 asserts where live on edge.
30728 (find_conditional_asserts): Adjust.
30729 (find_switch_asserts): Likewise.
30730 (evrp_dom_walker::try_find_new_range): Generalize.
30731 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
30733 2017-04-27 Marek Polacek <polacek@redhat.com>
30736 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
30737 arg10 and arg11 to itype.
30739 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
30741 * doc/extend.texi (Object Size Checking): Improve grammar.
30743 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
30746 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
30747 that the logic for permitting reciprocal estimates matches that
30750 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30753 * tree.c (type_cache_hasher::equal): Only compare
30754 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
30755 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
30756 non-aggregate element types.
30757 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
30758 about the flag on ARRAY_TYPEs in the comment, formatting fix.
30760 2017-04-27 Richard Biener <rguenther@suse.de>
30762 PR middle-end/80533
30763 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
30764 stripping ARRAY_REFs from MEM_EXPR make sure we're not
30765 keeping a reference to a trailing array.
30767 2017-04-27 Richard Biener <rguenther@suse.de>
30769 PR middle-end/80539
30770 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
30771 being in loop-closed SSA form conservatively.
30772 (chrec_fold_multiply_poly_poly): Likewise.
30774 2017-04-27 Tamar Christina <tamar.christina@arm.com>
30776 PR middle-end/79665
30777 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
30778 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
30780 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30783 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
30784 (aarch64_function_arg_alignment): Return unsigned int again, but still
30785 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
30786 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30787 Don't emit -Wpsabi note.
30788 (aarch64_function_arg_boundary): Likewise.
30789 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30792 2017-04-26 Nathan Sidwell <nathan@acm.org>
30794 * tree.h (crc32_unsigned_n): Declare.
30795 (crc32_unsigned, crc32_unsigned): Make inline.
30796 * tree.c (crc32_unsigned_bits): Replace with ...
30797 (crc32_unsigned_n): ... this.
30798 (crc32_unsigned, crc32_byte): Remove.
30799 (crc32_string): Remove unnecessary braces.
30801 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30803 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
30804 * ipa-inline-analysis.c (MAX_TIME): Remove.
30805 (account_size_time): Use sreal for time.
30806 (dump_inline_summary): Update.
30807 (estimate_function_body_sizes): Update.
30808 (estimate_edge_size_and_time): Update.
30809 (estimate_calls_size_and_time): Update.
30810 (estimate_node_size_and_time): Update.
30811 (inline_merge_summary): Update.
30812 (inline_update_overall_summary): Update.
30813 (estimate_time_after_inlining): Update.
30814 (inline_read_section): Update.
30815 (inline_write_summary): Update.
30816 * ipa-inline.c (compute_uninlined_call_time): Update.
30817 (compute_inlined_call_time): Update.
30818 (recursive_inlining): Update.
30819 (inline_small_functions): Update.
30820 (dump_overall_stats): Update.
30821 * ipa-inline.h: Include sreal.h.
30822 (size_time_entry): Turn time to sreal.
30823 (inline_summary): Turn self_time nad time to sreal.
30825 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30827 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
30829 (sreal::stream_out, sreal::stream_in): New.
30830 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
30832 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30834 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
30837 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
30840 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
30842 (dimode_scalar_chain::compute_convert_gain): Ditto.
30843 (dimode_scalar_chain::make_vector_copies): Ditto.
30844 (dimode_scalar_chain::convert_reg): Ditto.
30845 (dimode_scalar_chain::convert_insn): Ditto.
30846 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
30847 (VI248_AVX512BW_1): New mode iterator.
30848 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
30849 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
30852 2017-04-25 Martin Sebor <msebor@redhat.com>
30854 PR tree-optimization/80497
30855 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
30856 constants are representable in HOST_WIDE_INT.
30857 (parse_directive): Ditto.
30859 2017-04-25 Martin Sebor <msebor@redhat.com>
30862 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
30863 (new_zero_array): Adjust signature.
30864 (dom_info::dom_init): Used unsigned rather that size_t.
30865 (dom_info::dom_info): Same.
30867 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30868 Jakub Jelinek <jakub@redhat.com>
30871 * config/arm/arm.c: Include gimple.h.
30872 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30873 returns negative, increment ncrn only if it returned positive.
30874 (arm_needs_doubleword_align): Return int instead of bool,
30875 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
30876 members, but if there is any such non-FIELD_DECL
30877 > PARM_BOUNDARY aligned decl, return -1 instead of false.
30878 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30879 returns negative, increment nregs only if it returned positive.
30880 (arm_setup_incoming_varargs): Likewise.
30881 (arm_function_arg_boundary): Emit -Wpsabi note if
30882 arm_needs_doubleword_align returns negative, return
30883 DOUBLEWORD_ALIGNMENT only if it returned positive.
30885 2017-04-25 Marek Polacek <polacek@redhat.com>
30888 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30889 first argument to type.
30891 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
30894 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
30895 type checks to test for compatibility instead of equality.
30897 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30898 Jakub Jelinek <jakub@redhat.com>
30901 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
30903 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
30904 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
30905 the alignment computation, but return their maximum in warn_alignment.
30906 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30907 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
30909 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
30910 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30913 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30915 * config/arc/simdext.md (dmpyh): Fix typo.
30917 2017-04-25 Richard Biener <rguenther@suse.de>
30919 PR tree-optimization/80492
30920 * alias.c (compare_base_decls): Handle registers with asm
30921 specification conservatively.
30922 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30923 compare_base_decls returning dont-know properly.
30925 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30927 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
30928 (legitimate_offset_address_p): New function.
30929 (arc_legitimate_address_p): Use above function.
30931 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30933 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
30935 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30937 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
30938 ACCH registers whenever they are available.
30940 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30942 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
30943 double regs fix when not used.
30945 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30947 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
30949 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
30950 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
30952 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30954 * config/arc/arc.c (arc_output_addsi): Check for h-register class
30955 when emitting short ADD instructions.
30957 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30959 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
30961 (cmpsi_cc_c_insn): Likewise.
30962 (cbranchsi4_scratch): Compute proper instruction length using
30963 compact_hreg_operand.
30964 * config/arc/predicates.md (compact_hreg_operand): New predicate.
30966 2017-04-25 Richard Biener <rguenther@suse.de>
30968 PR middle-end/80509
30969 * passes.c (pass_manager::pass_manager): Initialize
30970 m_name_to_pass_map.
30972 2017-04-25 Richard Biener <rguenther@suse.de>
30974 PR tree-optimization/79201
30975 * tree-ssa-sink.c (statement_sink_location): Handle calls.
30977 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30980 * config/s390/vector.md: Split MEM->GPR vector moves for
30981 non-s_operand addresses.
30983 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30986 * config/s390/predicates.md (reload_const_wide_int_operand): New
30988 * config/s390/s390.md ("movti"): Remove d/P alternative.
30989 ("movti_bigconst"): New pattern definition.
30991 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30994 * s390-protos.h (s390_expand_cs_hqi): Removed.
30995 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
30996 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
30997 modes as well as CCZ1mode and CCZmode.
30998 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
30999 signature of s390_emit_compare_and_swap.
31000 (s390_expand_cs_hqi): Likewise, make static.
31001 (s390_expand_cs_tdsi): Generate an explicit compare before trying
31002 compare-and-swap, in some cases.
31003 (s390_expand_cs): Wrapper function.
31004 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
31006 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
31007 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
31008 patterns for small and large integers. Forbid symref memory operands.
31009 Move expander to s390.c. Require cc register.
31010 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
31011 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
31012 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
31013 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
31014 symref memory operands. Remove CC mode and call s390_match_ccmode
31016 ("atomic_exchange<mode>"): Allow and implement all integer modes.
31018 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31020 * config/s390/s390.md (define_peephole2): New peephole to help
31021 combining the load-and-test pattern with volatile memory.
31023 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31025 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
31026 with CCZmode for TARGET_Z196.
31028 2017-04-25 Jakub Jelinek <jakub@redhat.com>
31030 PR rtl-optimization/80501
31031 * combine.c (make_compound_operation_int): Set subreg_code to SET
31032 even for AND with mask of the sign bit of mode.
31034 PR rtl-optimization/80500
31035 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
31036 sum's initial value.
31038 2017-04-25 Julian Brown <julian@codesourcery.com>
31039 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31041 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
31043 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
31045 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
31047 2017-04-25 Julian Brown <julian@codesourcery.com>
31048 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31050 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
31051 (thunderx2t99_sha): New Reservation.
31053 2017-04-25 Julian Brown <julian@codesourcery.com>
31054 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31056 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
31057 type for 1-element load.
31059 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
31061 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
31063 2017-04-24 Martin Jambor <mjambor@suse.cz>
31065 PR tree-optimization/80293
31066 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
31067 char arrays not totally scalarizable if it is false.
31068 (analyze_all_variable_accesses): Pass correct value in the new
31069 parameter. Add a statistics counter.
31071 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
31073 PR middle-end/79931
31074 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
31076 2017-04-24 Richard Biener <rguenther@suse.de>
31078 PR tree-optimization/80494
31079 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
31080 out for complex types.
31082 2017-04-24 Richard Biener <rguenther@suse.de>
31084 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
31085 * tree-ssa-sccvn.c (print_scc): Print SCC size.
31086 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
31087 (DFS): Adjust and never fail.
31088 (sccvn_dom_walker::fail): Remove.
31089 (sccvn_dom_walker::before_dom_children): Adjust.
31090 (run_scc_vn): Likewise and never fail.
31091 * tree-ssa-pre.c (pass_pre::execute): Adjust.
31092 (pass_fre::execute): Likewise.
31094 2017-04-24 Richard Biener <rguenther@suse.de>
31096 PR tree-optimization/79725
31097 * tree-ssa-sink.c (statement_sink_location): Return whether
31098 failure reason was zero uses. Move that check later.
31099 (sink_code_in_bb): Deal with zero uses by removing the stmt
31102 2017-04-24 Richard Biener <rguenther@suse.de>
31105 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
31106 pointer-based references.
31108 2017-04-24 Richard Biener <rguenther@suse.de>
31111 * pass_manager.h (pass_manager::operator new): Remove.
31112 (pass_manager::operator delete): Likewise.
31113 * passes.c (pass_manager::operator new): Remove.
31114 (pass_manager::operator delete): Likewise.
31115 (pass_manager::pass_manager): Zero individual pass members.
31117 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
31120 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
31121 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
31122 Check "XEXP (src, 1)" operand here.
31123 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
31124 Check "XEXP (src, 1)" operand here.
31125 (dimode_scalar_chain::make_vector_copies): Detect count register
31126 of a shift instruction. Zero extend count register from QImode
31127 to DImode to satisfy vector shift pattern count operand predicate.
31128 Substitute vector shift count operand with a DImode copy.
31129 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
31132 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
31134 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
31135 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
31136 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
31137 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
31138 (UNSPEC_NOREX_MEM): Remove definition.
31140 2017-04-21 Richard Biener <rguenther@suse.de>
31142 PR tree-optimization/79547
31143 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
31144 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
31145 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
31146 without any constraints.
31148 2017-04-21 Richard Biener <rguenther@suse.de>
31150 PR tree-optimization/78847
31151 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
31153 2017-04-21 Richard Biener <rguenther@suse.de>
31155 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
31156 (build_distinct_type_copy): Likewise.
31157 (build_variant_type_copy): Likewise.
31158 * tree.c (build_qualified_type): Pass down mem-stat info.
31159 (build_distinct_type_copy): Likewise.
31160 (build_variant_type_copy): Likewise.
31162 2017-04-21 Richard Biener <rguenther@suse.de>
31164 PR tree-optimization/80237
31165 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
31167 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
31168 for a simplified result.
31170 2016-04-21 Richard Biener <rguenther@suse.de>
31172 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
31173 sth as strict as a simple_iv but a chrec without symbols and an
31174 operand defined in the loop we are peeling (and not some subloop).
31175 (propagate_constants_for_unrolling): Propagate all constants.
31177 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
31180 * config/i386/i386.c (print_reg): Remove assert for disalowed
31181 regno values, call output_operand_lossage instead.
31183 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
31186 * config/i386/constraints.md (Yc): New register constraint.
31187 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
31188 Use Yc constraint for alternative 2 of operand 0. Remove
31189 preferred_for_speed attribute.
31191 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
31193 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
31194 lastprivate clauses in SIMT case.
31196 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
31198 * doc/invoke.texi (-Wextra-semi): Document new warning option.
31200 2017-04-20 Richard Biener <rguenther@suse.de>
31202 PR tree-optimization/57796
31203 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
31204 as N scalar stores.
31205 (vect_model_load_cost): Cost gathers as N scalar loads.
31207 2017-04-20 Richard Biener <rguenther@suse.de>
31209 * ggc-page.c (ggc_allocated_p): Rename to ...
31210 (safe_lookup_page_table_entry): ... this and return the lookup
31212 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
31214 2017-04-20 Richard Biener <rguenther@suse.de>
31216 PR tree-optimization/80453
31217 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
31218 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
31219 from the conditions.
31220 (vn_phi_eq): Pass them down.
31221 (vn_phi_lookup): Record them.
31222 (vn_phi_insert): Likewise.
31224 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
31226 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
31227 uninitialized variable warning to avoid buffer overrun.
31229 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
31232 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
31233 is suppressed for '{ 0 }' in C.
31235 2017-04-20 Jakub Jelinek <jakub@redhat.com>
31237 * BASE-VER: Set to 8.0.0.
31239 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
31241 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
31242 priority .init_array and .fini_array section with SECTION_NOTYPE
31245 2017-04-20 Jakub Jelinek <jakub@redhat.com>
31247 PR middle-end/80423
31248 * tree.h (build_array_type): Add typeless_storage default argument.
31249 * tree.c (type_cache_hasher::equal): Also compare
31250 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
31251 (build_array_type): Add typeless_storage argument, set
31252 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
31254 (build_nonshared_array_type): Adjust build_array_type_1 caller.
31255 (build_array_type): Likewise. Add typeless_storage argument.
31257 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
31258 Jakub Jelinek <jakub@redhat.com>
31260 PR tree-optimization/80426
31261 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
31262 operation on symbolic operands, also compute the overflow for the
31263 invariant part when the operation degenerates into a negation.
31265 2017-04-19 Jakub Jelinek <jakub@redhat.com>
31268 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
31269 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
31272 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
31274 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
31277 * config/avr/avr.c (tree.h): Include it.
31278 (cgraph.h): Include it.
31279 (avr_encode_section_info): Don't warn for uninitialized progmem
31280 variable if it's just an alias.
31282 2017-04-19 Richard Biener <rguenther@suse.de>
31285 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
31286 when needed by AutoPGO.
31288 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
31291 * doc/lto.texi: Remove an extra 'that'.
31293 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
31295 PR rtl-optimization/80429
31296 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
31297 are only used in debug insns.
31299 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
31300 Vladimir Makarov <vmakarov@redhat.com>
31302 * config/sparc/predicates.md (input_operand): Add comment. Return
31303 true for any memory operand when LRA is in progress.
31304 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
31306 2017-04-18 Jeff Law <law@redhat.com>
31309 * mips.md ({return,simple_return}_internal): Do not overwrite
31312 2017-04-18 Jakub Jelinek <jakub@redhat.com>
31314 PR tree-optimization/80443
31315 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
31316 instead of adding 1, subtract -1 and similarly instead of subtracting
31319 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
31321 PR rtl-optimization/80357
31322 * haifa-sched.c (tmp_bitmap): New variable.
31323 (model_recompute): Handle duplicate use records.
31324 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
31325 (free_global_sched_pressure_data): Free it.
31327 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
31330 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31331 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31332 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31333 instead of SYSTEM_HEADER_DIR.
31335 2017-04-18 Jeff Law <law@redhat.com>
31337 PR middle-end/80422
31338 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
31339 predecessors after walking up the insn chain.
31341 2017-04-18 Jakub Jelinek <jakub@redhat.com>
31344 * dwarf2out.c (modified_type_die): Try harder not to emit internal
31345 sizetype type into debug info.
31347 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
31350 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
31351 unneeded test for TARGET_UPPER_REGS_SF.
31352 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
31354 2017-04-18 Jakub Jelinek <jakub@redhat.com>
31357 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
31358 instead of gsi_after_labels.
31360 2017-04-18 Jeff Law <law@redhat.com>
31362 * regcprop.c (maybe_mode_change): Avoid creating copies of the
31366 2017-04-13 Jeff Law <law@redhat.com>
31367 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
31368 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
31370 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
31373 * config/avr/avr.c (intl.h): Include it.
31374 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
31376 2017-04-18 Martin Liska <mliska@suse.cz>
31378 PR gcov-profile/78783
31379 * gcov-tool.c (gcov_output_files): Validate that destination
31380 file is either removed by the tool or by a user.
31382 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
31383 Guy Benyei <guybe@mellanox.com>
31385 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
31386 block, and do not negate it, the stored id is already negative.
31388 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
31390 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
31392 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
31395 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
31396 masks of options that should be turned off if the VSX vector
31397 options are turned off.
31398 (OTHER_P8_VECTOR_MASKS): Likewise.
31399 (OTHER_VSX_VECTOR_MASKS): Likewise.
31400 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
31401 rs6000_disable_incompatible_switches to validate no type switches
31403 (rs6000_incompatible_switch): New function to disallow turning on
31404 other vector options if -mno-vsx, -mno-power8-vector, or
31405 -mno-power9-vector are specified.
31407 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
31409 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
31411 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
31413 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
31414 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
31415 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
31416 (ARG_POINTER_CFA_OFFSET): Likewise.
31418 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
31420 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
31421 conditions to take advantage of various optimizations.
31423 2017-04-13 Jeff Law <law@redhat.com>
31425 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
31426 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
31427 (zero_extendsidi2_dext): Likewise.
31429 2017-04-13 Jakub Jelinek <jakub@redhat.com>
31432 * fold-const.c (fold_ternary_loc): Revert
31433 use op0 instead of fold_convert_loc (loc, type, arg0) part of
31436 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
31438 PR rtl-optimization/80343
31439 * lra-remat.c (update_scratch_ops): Assign original hard reg to
31440 new scratch pseudo.
31442 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
31445 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
31446 to ubsan_encode_value.
31448 2017-04-13 Jeff Law <law@redhat.com>
31450 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
31451 appearing in DEBUG_INSNs.
31453 2017-04-13 Martin Liska <mliska@suse.cz>
31455 PR gcov-profile/80413
31456 * gcov-io.c (gcov_write_string): Copy to buffer just when
31457 allocated size is greater than zero.
31459 2017-04-13 Jakub Jelinek <jakub@redhat.com>
31462 * dwarf2out.c (decls_for_scope): Ignore declarations of
31463 current_function_decl in BLOCK_NONLOCALIZED_VARS.
31465 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
31468 * ipa-visibility.c (non_local_p): Fix typos.
31469 (localize_node): When localizing symbol in same comdat group,
31470 dissolve the group only when we know external symbols are going
31472 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
31474 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31476 PR tree-optimization/79390
31477 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
31478 order does not result in usable sequence, retry with reversed operand
31484 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
31485 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
31486 op0 instead of fold_convert_loc (loc, type, arg0).
31488 2017-04-12 Jeff Law <law@redhat.com>
31490 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
31491 has a delay slot in the generated code.
31493 * config/cris/cris.md (cris_preferred_reload_class): Return
31494 GENNONACR_REGS rather than GENERAL_REGS.
31496 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31499 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
31500 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
31501 signedness of the result type.
31503 2017-04-12 Richard Biener <rguenther@suse.de>
31504 Jeff Law <law@redhat.com>
31506 PR tree-optimization/80359
31507 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
31508 trim stores to TARGET_MEM_REFs.
31510 2017-04-12 Richard Biener <rguenther@suse.de>
31512 PR tree-optimization/79390
31513 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
31514 threading case even more.
31516 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
31519 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
31520 for quad_address_p for TImode, instead of just not indexed_address.
31522 2017-04-12 Richard Biener <rguenther@suse.de>
31523 Bernd Edlinger <bernd.edlinger@hotmail.de>
31525 PR middle-end/79671
31526 * alias.c (component_uses_parent_alias_set_from): Handle
31527 TYPE_TYPELESS_STORAGE.
31528 (get_alias_set): Likewise.
31529 * tree-core.h (tree_type_common): Add typeless_storage flag.
31530 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
31531 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
31532 for types containing members with TYPE_TYPELESS_STORAGE.
31533 (place_field): Likewise.
31534 (layout_type): Likewise for ARRAY_TYPE.
31535 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
31536 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
31537 TYPE_TYPELESS_STORAGE.
31538 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
31540 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31543 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
31544 first argument to type.
31546 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31550 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
31551 CONST0_RTX (mode) rather than const0_rtx where appropriate.
31552 (rs6000_expand_binop_builtin): Likewise.
31553 (rs6000_expand_ternop_builtin): Likewise; also add missing
31554 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
31555 vshasigma built-ins.
31556 * doc/extend.texi: Document that vec_xxpermdi's third argument
31557 must be a constant.
31559 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
31561 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
31562 Use shift_const cost parameter when calculating gain of STV shifts.
31564 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
31566 PR rtl-optimization/70478
31567 * lra-constraints.c (process_alt_operands): Check memory for
31568 disfavoring memory insn operand.
31570 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31572 PR middle-end/80100
31573 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
31574 left shift in unsigned HOST_WIDE_INT type.
31576 PR rtl-optimization/80385
31577 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
31578 (not (neg X)) into (plus X -1) for complex or non-integral modes.
31581 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
31582 if they have any depend clauses.
31584 2017-04-11 Martin Liska <mliska@suse.cz>
31587 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
31588 * ipa-split.c (split_function): Create a local comdat symbol
31589 if caller is in a comdat group.
31591 2017-04-11 Martin Liska <mliska@suse.cz>
31594 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
31597 2017-04-11 Martin Sebor <msebor@redhat.com>
31599 PR middle-end/80364
31600 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
31601 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
31603 (directive::set_width, directive::set_precision, format_character):
31605 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
31608 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
31611 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
31612 conflict, set target->arch_name instead of target->cpu_name.
31614 2017-04-11 Richard Biener <rguenther@suse.de>
31616 PR tree-optimization/80374
31617 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
31618 build_zero_cst, remove fold_convertible_p check again.
31620 2017-04-11 Martin Liska <mliska@suse.cz>
31623 * ubsan.c (instrument_object_size): Do not instrument register
31626 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31629 * config/i386/i386-builtin-types.def
31630 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
31631 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
31632 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
31633 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
31634 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
31635 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
31636 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
31637 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
31638 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
31639 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
31640 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
31641 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
31642 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
31643 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
31644 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
31645 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
31646 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
31647 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
31648 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
31649 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
31650 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
31651 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
31652 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
31653 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
31654 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
31655 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
31656 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
31657 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
31658 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
31659 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
31660 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
31661 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
31662 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
31663 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
31664 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
31665 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
31666 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
31667 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
31668 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
31669 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
31670 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
31671 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
31672 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
31673 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
31674 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
31675 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
31677 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
31678 flag to second_arg_count, handle 4 argument function type _COUNT
31679 aliases, handle second_arg_count on second argument rather than last.
31681 2017-04-10 Jeff Law <law@redhat.com>
31683 PR tree-optimization/80374
31684 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
31685 record anything if we can not convert integer_zero_node to the
31688 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
31691 * config/rs6000/rs6000.c (rs6000_option_override_internal):
31692 Enhance special handling given to the TARGET_P9_MINMAX option in
31693 relation to certain other options.
31695 2017-04-10 Bin Cheng <bin.cheng@arm.com>
31697 PR tree-optimization/80153
31698 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
31699 remove POINTER_PLUS_EXPR's base part directly, rather than through
31702 2017-04-10 Richard Biener <rguenther@suse.de>
31703 Bin Cheng <bin.cheng@arm.com>
31705 PR tree-optimization/80153
31706 * tree-affine.c (aff_combination_to_tree): Get base pointer from
31707 the first element of pointer type aff_tree. Build result expr in
31709 (add_elt_to_tree): Convert to type unconditionally. Remove other
31710 fold_convert calls.
31711 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
31712 (rewrite_use_nonlinear_expr): Check invariant using iv information.
31714 2017-04-10 Richard Biener <rguenther@suse.de>
31716 * tree-ssa-structalias.c (find_func_aliases): Properly handle
31719 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
31721 PR rtl-optimization/70478
31722 * lra-constraints.c (curr_small_class_check): New.
31723 (update_and_check_small_class_inputs): New.
31724 (process_alt_operands): Update curr_small_class_check. Disfavor
31725 alternative insn memory operands. Check available regs for small
31728 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31731 * config/mips/mips.opt (-mvirt): Update description.
31732 * doc/invoke.texi (-mvirt): Likewise.
31734 2017-04-10 Richard Biener <rguenther@suse.de>
31736 PR middle-end/80362
31737 * fold-const.c (fold_binary_loc): Look at unstripped ops when
31738 looking for NEGATE_EXPR in -A / -B to A / B folding.
31740 2017-04-10 Martin Liska <mliska@suse.cz>
31742 PR gcov-profile/80224
31743 * gcov.c (print_usage): Fix usage string.
31744 (get_gcov_intermediate_filename): Remove.
31745 (output_gcov_file): Use both for normal and intermediate format.
31746 (generate_results): Do not initialize special file for
31747 intermediate format.
31749 2017-04-10 Richard Biener <rguenther@suse.de>
31751 PR tree-optimization/80304
31752 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
31755 2017-04-10 Nathan Sidwell <nathan@acm.org>
31758 * config/rs6000/rs6000.c (rs6000_vector_type): New.
31759 (rs6000_init_builtins): Use it.
31761 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31763 * config/arm/arm.md (<mrc>): Add mode to SET source.
31764 (<mrrc>): Likewise.
31766 2017-04-10 Richard Biener <rguenther@suse.de>
31768 PR middle-end/80344
31769 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
31771 2017-04-10 Jakub Jelinek <jakub@redhat.com>
31774 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
31775 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
31776 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
31777 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
31778 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
31779 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
31780 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
31781 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
31782 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
31783 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
31784 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
31785 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
31786 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
31787 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
31788 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
31789 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
31790 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
31791 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
31792 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
31793 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
31794 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
31795 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
31796 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
31798 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
31800 PR rtl-optimization/70478
31801 * lra-constraints.c: Reverse the last patch.
31803 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
31805 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
31806 Add comment for WCHAR_T.
31808 2017-04-08 Martin Liska <mliska@suse.cz>
31811 2017-04-07 Martin Liska <mliska@suse.cz>
31814 * ipa-split.c (split_function): Add function part to a same comdat
31817 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31820 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
31822 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
31824 * rs6000/rs6000.c (vec_load_pendulum): Rename...
31825 (vec_pairing): ...to this.
31826 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
31827 (rs6000_sched_init): Adjust for name change.
31828 (struct rs6000_sched_context): Likewise.
31829 (rs6000_init_sched_context): Likewise.
31830 (rs6000_set_sched_context): Likewise.
31832 2017-04-07 Jakub Jelinek <jakub@redhat.com>
31838 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
31840 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
31841 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
31842 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
31844 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
31846 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
31848 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31850 PR rtl-optimization/70703
31851 * ira-color.c (update_conflict_hard_regno_costs): Use
31852 int64_t instead of HOST_WIDE_INT.
31854 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31856 PR rtl-optimization/70478
31857 * lra-constraints.c (process_alt_operands): Disfavor alternative
31858 insn memory operands.
31860 2017-04-07 Jeff Law <law@redhat.com>
31862 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
31863 CALL and NOTE_INSN_CALL_ARG_LOCATION.
31865 2017-04-07 Martin Liska <mliska@suse.cz>
31868 * config/aarch64/aarch64.c (aarch64_process_target_attr):
31869 Show error message instead of an ICE.
31871 2017-04-07 Martin Liska <mliska@suse.cz>
31874 * ipa-split.c (split_function): Add function part to a same comdat
31877 2017-04-07 Richard Biener <rguenther@suse.de>
31879 PR middle-end/80341
31880 * tree.c (get_unwidened): Also handle ! for_type case for
31882 * convert.c (do_narrow): Split out from ...
31883 (convert_to_integer_1): ... here. Do not pass final truncation
31884 type to get_unwidened for TRUNC_DIV_EXPR.
31886 2017-04-07 Richard Biener <rguenther@suse.de>
31888 * tree-affine.c (wide_int_ext_for_comb): Take type rather
31890 (aff_combination_const): Adjust.
31891 (aff_combination_scale): Likewise.
31892 (aff_combination_add_elt): Likewise.
31893 (aff_combination_add_cst): Likewise.
31894 (aff_combination_convert): Likewise.
31895 (add_elt_to_tree): Likewise. Remove unused argument.
31896 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
31898 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
31900 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
31902 * config/arm/arm.c (arm_default_short_enums): Use
31903 ARM_DEFAULT_SHORT_ENUMS.
31904 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
31906 2017-04-06 Jakub Jelinek <jakub@redhat.com>
31909 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
31910 members with redundant out-of-class redeclaration.
31912 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31915 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
31916 * config/i386/i386.md (*zero_extendsidi2):
31917 Add (?*x,*x) and (?*v,*v) alternatives.
31919 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31922 * config/i386/i386.c (ix86_expand_builtin)
31923 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
31924 mode from insn data. Convert operands to insn operand mode.
31925 Copy operands that don't satisfy insn predicate to a register.
31927 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
31929 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
31932 2017-04-06 Richard Biener <rguenther@suse.de>
31934 PR tree-optimization/80334
31935 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
31936 preserve alignment of accesses.
31938 2017-04-06 Richard Biener <rguenther@suse.de>
31940 PR tree-optimization/80262
31941 * tree-sra.c (build_ref_for_offset): Preserve address-space
31943 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
31944 Drop useless address-space information on MEM_REF offsets.
31946 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
31948 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
31950 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31952 PR rtl-optimization/70703
31953 * ira-color.c (update_conflict_hard_regno_costs): Use
31954 HOST_WIDE_INT instead of long.
31956 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
31959 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
31960 not defined for x86_64 target. Add -mmmx target option when __SSE2__
31962 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
31963 for x86_64 target. Handle -m3dnowa option.
31965 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31967 PR rtl-optimization/70703
31968 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
31969 (update_conflict_hard_regno_costs): Use long instead of unsigned
31970 arithmetic for cost calculation.
31972 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31973 Bernd Edlinger <bernd.edlinger@hotmail.de>
31976 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
31979 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
31982 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
31983 ptr_mode with Pmode throughout.
31984 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
31985 into probe_stack_range and use DImode.
31987 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
31990 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
31991 call_eh_return is true.
31993 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31995 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
31996 Initialize last_match_fntype_index.
31998 2017-04-05 Jakub Jelinek <jakub@redhat.com>
32001 * tree-nvr.c: Include internal-fn.h.
32002 (pass_return_slot::execute): Ignore internal calls without
32005 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32006 Richard Biener <rguenther@suse.de>
32009 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
32010 captures used multiple times, except for the last use.
32011 * generic-match-head.c: Include gimplify.h.
32013 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32015 PR tree-optimization/79390
32016 * target.h (struct noce_if_info): Declare.
32017 * targhooks.h (default_noce_conversion_profitable_p): Declare.
32018 * target.def (noce_conversion_profitable_p): New target hook.
32019 * ifcvt.h (struct noce_if_info): New type, moved from ...
32020 * ifcvt.c (struct noce_if_info): ... here.
32021 (noce_conversion_profitable_p): Renamed to ...
32022 (default_noce_conversion_profitable_p): ... this. No longer
32024 (noce_try_store_flag_constants, noce_try_addcc,
32025 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
32026 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
32027 instead of noce_conversion_profitable_p.
32028 * config/i386/i386.c: Include ifcvt.h.
32029 (ix86_option_override_internal): Don't override
32030 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
32031 (ix86_noce_conversion_profitable_p): New function.
32032 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
32033 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
32034 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
32035 * doc/tm.texi: Regenerated.
32037 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32039 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
32042 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32045 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
32046 instructions for small multiply cores.
32048 2017-04-04 Jeff Law <law@redhat.com>
32050 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
32052 (mips_expand_vec_perm_const): Initialize elements in orig_perm
32053 that are not set by the loop over the elements.
32055 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32058 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
32059 int mode, convert_modes it to mode as unsigned, otherwise use
32060 lowpart_subreg to mode rather than SImode.
32061 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
32062 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
32063 Use DImode instead of SImode for the shift count operand.
32064 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
32067 2017-04-04 Richard Biener <rguenther@suse.de>
32069 PR middle-end/80281
32070 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
32071 arithmetic done for the negate or the plus. Simplify.
32072 (A - (-B) -> A + B): Likewise.
32073 * fold-const.c (split_tree): Make sure to not negate pointers.
32075 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
32077 PR rtl-optimization/60818
32078 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
32079 a compare of comparisons with the thing compared if this results
32080 in a different machine mode.
32082 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
32084 * alias.c (base_alias_check): Fix typo in comment.
32085 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
32086 * cgraphunit.c (symbol_table::compile): Likewise.
32087 * collect2.c (maybe_run_lto_and_relink): Likewise.
32088 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
32089 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
32090 * config/avr/avr.c (avr_map_op_t): Likewise.
32091 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
32092 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
32093 * config/epiphany/epiphany.md (movcc): Likewise.
32094 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
32095 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
32097 * config/mips/mips.c (mips_save_restore_reg): Likewise.
32098 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
32099 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
32100 * config/sh/sh.c (sh_rtx_costs): Likewise.
32101 * fold-const.c (fold_truth_andor): Likewise.
32102 * genautomata.c (collapse_flag): Likewise.
32103 * gengtype.h (struct type::u::s): Likewise.
32104 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
32105 * input.c (FORMAT_AMOUNT): Likewise.
32106 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
32107 (known_aggs_to_agg_replacement_list): Likewise.
32108 * ipa-inline-analysis.c: Likewise.
32109 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
32110 * ipa-polymorphic-call.c
32111 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
32112 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
32113 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
32115 * modulo-sched.c (apply_reg_moves): Likewise.
32116 * omp-expand.c (build_omp_regions_1): Likewise.
32117 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
32118 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
32119 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
32120 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
32121 * value-prof.c: Likewise.
32122 * var-tracking.c (val_reset): Likewise.
32124 2017-04-03 Richard Biener <rguenther@suse.de>
32126 PR tree-optimization/80275
32127 * fold-const.c (split_address_to_core_and_offset): Handle
32130 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
32132 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
32133 descriptors is at least equal to that of functions.
32135 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
32137 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
32139 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
32142 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
32143 (mov<IMOD4:mode>): New expander.
32144 (*mov<IMOD4:mode>_internal): New insn and split pattern.
32146 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
32148 PR rtl-optimization/79405
32149 * fwprop.c (propagations_left): New variable.
32150 (forward_propagate_into): Decrement it.
32151 (fwprop_init): Initialize it.
32152 (fw_prop): If the variable has reached zero, stop propagating.
32153 (fwprop_addr): Ditto.
32155 2017-03-31 Jakub Jelinek <jakub@redhat.com>
32158 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
32159 a FUNCTION_DECL, pass it as decl instead of origin to
32162 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
32164 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
32167 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
32170 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
32171 TARGET_VSX_SMALL_INTEGER.
32173 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32175 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
32176 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
32178 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
32180 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
32181 extraction from odd-numbered MSA register.
32183 2017-03-31 Jakub Jelinek <jakub@redhat.com>
32185 PR middle-end/80173
32186 * expmed.c (store_bit_field_1): Don't attempt to create
32187 a word subreg out of hard registers wider than word if they
32188 have HARD_REGNO_NREGS of 1 for their mode.
32190 PR middle-end/80163
32191 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
32192 conversions to integer types wider than word and pointer.
32195 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
32196 (rtx_equal_for_cselib_p): Pass 0 to it.
32197 * cselib.c (cselib_hasher::equal): Likewise.
32198 (rtx_equal_for_cselib_1): Add depth argument. If depth
32199 is 128, don't look up VALUE locs and punt. Increment
32200 depth in recursive calls when walking VALUE locs.
32202 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
32204 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
32205 (make_gcov_file_name): Use the canonical path name for generating
32207 (read_line): Fix handling of files with ascii null bytes.
32209 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
32211 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
32212 to initialise a vector register instead
32213 of using a const_int.
32215 2017-03-30 Jakub Jelinek <jakub@redhat.com>
32217 PR translation/80189
32218 * gimplify.c (omp_default_clause): Use %qs instead of %s in
32219 diagnostic messages.
32221 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
32224 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
32225 (dfp_diex_<mode>): Update mode of operand 1.
32226 * doc/extend.texi (dxex, dxexq): Document change to return type.
32227 (diex, diexq): Document change to argument type.
32229 2017-03-30 Martin Jambor <mjambor@suse.cz>
32232 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
32233 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
32234 it reflects the signature changes performed at the callee side.
32235 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
32236 to cgraph_build_function_type_skip_args.
32237 (build_function_decl_skip_args): Adjust call to the above function.
32239 2017-03-30 Jakub Jelinek <jakub@redhat.com>
32242 * config/i386/sse.md
32243 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
32244 register as dest whenever it is a MEM not rtx_equal_p to the
32245 corresponding dup operand, and when forcing into reg move the
32246 reg into the memory afterwards.
32247 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
32248 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
32249 for the force_reg mode.
32250 (avx512vl_vextractf128<mode>): Use register as dest either
32251 always when a MEM, or when it is a MEM not rtx_equal_p to the
32252 corresponding dup operand, or even not when it is a CONST_VECTOR
32253 depending on the mode and lo vs. hi.
32254 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
32256 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
32257 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
32258 Likewise. Require that operands[2] is even.
32259 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
32260 Remove extraneous parens. Require that operands[2] is a multiple
32262 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
32263 operands[0] is a MEM if <mask_applied>, the predicates/constraints
32264 disallow memory then.
32266 2017-03-30 Richard Biener <rguenther@suse.de>
32268 PR tree-optimization/77498
32269 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
32270 to non-constants over backedges.
32272 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
32274 PR rtl-optimization/80233
32275 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
32276 as last_combined_insn. Do not test for BARRIER_P separately.
32278 2017-03-29 Andreas Schwab <schwab@suse.de>
32281 * calls.c (prepare_call_address): Convert funexp to Pmode before
32282 copying to temp reg.
32284 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32286 PR tree-optimization/80158
32287 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
32288 Handle possible future case of more than one alternate
32290 (replace_rhs_if_not_dup): Likewise.
32291 (replace_one_candidate): Likewise.
32293 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
32295 PR rtl-optimization/80193
32296 * ira.c (ira): Do not check allocation for LRA.
32298 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
32300 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
32301 (nvptx_output_simt_exit): Declare.
32302 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
32303 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
32304 (init_softstack_frame): Move initialization of crtl->is_leaf to...
32305 (nvptx_declare_function_name): ...here. Emit declaration of local
32306 memory space buffer for omp_simt_enter insn.
32307 (nvptx_output_unisimt_switch): New.
32308 (nvptx_output_softstack_switch): New.
32309 (nvptx_output_simt_enter): New.
32310 (nvptx_output_simt_exit): New.
32311 * config/nvptx/nvptx.h (struct machine_function): New fields
32312 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
32313 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
32314 (UNSPECV_SIMT_EXIT): Ditto.
32315 (omp_simt_enter_insn): New insn.
32316 (omp_simt_enter): New expansion.
32317 (omp_simt_exit): New insn.
32318 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
32320 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
32321 (expand_GOMP_SIMT_ENTER_ALLOC): New.
32322 (expand_GOMP_SIMT_EXIT): New.
32323 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
32324 (GOMP_SIMT_ENTER_ALLOC): Ditto.
32325 (GOMP_SIMT_EXIT): Ditto.
32326 * target-insns.def (omp_simt_enter): New insn.
32327 (omp_simt_exit): Ditto.
32328 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
32330 (lower_rec_simd_input_clauses): Implement SIMT privatization.
32331 (lower_rec_input_clauses): Likewise.
32332 (lower_lastprivate_clauses): Handle SIMT privatization.
32334 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
32335 (ompdevlow_adjust_simt_enter): New.
32336 (find_simtpriv_var_op): New.
32337 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
32338 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
32340 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
32341 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
32342 (copy_decl_for_dup_finish): Ditto.
32344 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
32346 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
32349 * config/i386/i386.c (ix86_option_override_internal): Always
32350 allow -mpreferred-stack-boundary=3 for 64-bit targets.
32352 2017-03-28 Bin Cheng <bin.cheng@arm.com>
32354 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
32356 2017-03-28 Bin Cheng <bin.cheng@arm.com>
32358 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
32359 mark new edge's irreducible flag accordign to it.
32360 (vect_do_peeling): Check loop preheader edge's irreducible flag
32361 and pass it to function slpeel_add_loop_guard.
32363 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
32365 PR tree-optimization/80218
32366 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
32367 Update block frequencies and counts.
32369 2017-03-28 Richard Biener <rguenther@suse.de>
32371 PR tree-optimization/78644
32372 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
32373 of a simplification result we may not use it at all.
32375 2017-03-28 Richard Biener <rguenther@suse.de>
32378 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
32379 without arguments, generate default definition of a SSA name.
32381 2017-03-28 Richard Biener <rguenther@suse.de>
32383 PR middle-end/80222
32384 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
32385 TYPE_REF_CAN_ALIAS_ALL references.
32386 * fold-const.c (fold_indirect_ref_1): Likewise.
32388 2017-03-28 Martin Liska <mliska@suse.cz>
32391 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
32392 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
32394 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
32395 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
32397 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
32398 (EXTRA_SPECS): Define.
32399 (SUBTARGET_EXTRA_SPECS): Likewise.
32400 (SUBTARGET_CPP_SPEC): Likewise.
32401 * config/arc/elf.h (EXTRA_SPECS): Renamed to
32402 SUBTARGET_EXTRA_SPECS.
32403 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
32405 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
32407 * config/arc/simdext.md (vst64_insn): Update pattern.
32408 (vld32wh_insn): Likewise.
32409 (vld32wl_insn): Likewise.
32410 (vld64_insn): Likewise.
32411 (vld32_insn): Likewise.
32413 2017-03-28 Marek Polacek <polacek@redhat.com>
32416 * fold-const.c (fold_comparison): Use protected_set_expr_location
32417 instead of SET_EXPR_LOCATION.
32419 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
32421 * tree.c (add_expr): Avoid name lookup warning.
32423 2017-03-27 Jeff Law <law@redhat.com>
32425 PR tree-optimization/80216
32426 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
32427 function name. Limit recursion depth.
32428 (record_temporary_equivalences): Corresponding changes.
32430 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
32432 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
32435 2017-03-27 Jakub Jelinek <jakub@redhat.com>
32438 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
32440 * cfgcleanup.c (reg_note_cfa_p): New array.
32441 (insns_have_identical_cfa_notes): New function.
32442 (old_insns_match_p): Don't cross-jump in between /f
32443 and non-/f instructions. If both i1 and i2 are frame related,
32444 verify all CFA notes, their order and content.
32446 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
32449 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
32450 HImode and SImode with zero extend to DImode to one insn.
32451 (bswap<mode>2_extenddi): Likewise.
32452 (bswapsi2_extenddi): Likewise.
32453 (bswaphi2_extendsi): Likewise.
32454 (bswaphi2): Combine bswap HImode and SImode into one insn.
32455 Separate memory insns from swapping register.
32456 (bswapsi2): Likewise.
32457 (bswap<mode>2): Likewise.
32458 (bswaphi2_internal): Delete, no longer used.
32459 (bswapsi2_internal): Likewise.
32460 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
32461 store, and gpr<-gpr swap insns.
32462 (bswap<mode>2_store): Likewise.
32463 (bswaphi2_reg): Register only splitter, combine with the splitter.
32464 (bswaphi2 splitter): Likewise.
32465 (bswapsi2_reg): Likewise.
32466 (bswapsi2 splitter): Likewise.
32467 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
32468 the insns into load, store, and register/register insns.
32469 (bswapdi2_ldbrx): Likewise.
32470 (bswapdi2_load): Likewise.
32471 (bswapdi2_store): Likewise.
32472 (bswapdi2_reg): Likewise.
32474 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
32476 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
32477 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
32479 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
32482 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
32484 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32485 special handling for target option conflicts between dform
32486 options (-mpower9-dform, -mpower9-dform-vector,
32487 -mpower9-dform-scalar) and -mno-direct-move.
32489 2017-03-27 Richard Biener <rguenther@suse.de>
32491 PR tree-optimization/80181
32492 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
32494 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32496 * config/arc/predicates.md (move_double_src_operand): Replace the
32497 call to move_double_src_operand with a call to address_operand.
32499 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32501 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
32502 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
32503 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
32505 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32507 * config/arc/predicates.md (long_immediate_loadstore_operand):
32508 Consider scaled addresses cases.
32510 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32512 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
32513 restored when in interrupt.
32514 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
32515 doesn't have delay slot.
32517 2017-03-27 Richard Biener <rguenther@suse.de>
32520 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
32521 inlined thunk clones.
32523 2017-03-27 Jakub Jelinek <jakub@redhat.com>
32526 * asan.c (instrument_derefs): Copy over last operand from
32527 original COMPONENT_REF to the new COMPONENT_REF with
32528 DECL_BIT_FIELD_REPRESENTATIVE.
32529 * ubsan.c (instrument_object_size): Likewise.
32531 2017-03-27 Richard Biener <rguenther@suse.de>
32533 PR tree-optimization/80170
32534 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
32535 sure DR/SCEV didnt fold in constants we do not see when looking
32536 at the reference base alignment.
32538 2017-03-27 Richard Biener <rguenther@suse.de>
32540 PR middle-end/80171
32541 * gimple-fold.c (fold_ctor_reference): Properly guard against
32542 NULL return value from canonicalize_constructor_val.
32544 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
32547 * config/i386/i386.c (ix86_expand_builtin)
32548 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
32549 flags reg setting and flags reg using instructions.
32550 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
32551 clobbering instructions to zero extend op2.
32553 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
32555 * doc/install.texi (Configuration) <--with-aix-soname>:
32556 Update link to AIX ld.
32558 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
32560 PR rtl-optimization/80160
32561 PR rtl-optimization/80159
32562 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
32563 reg_alternate_class into account.
32565 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
32568 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
32569 to consider in curr_insn_transform.
32571 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32573 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
32574 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
32575 and emit_mode_inner.
32577 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32579 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
32580 argument to the overloaded builtin variants. Use the new flag to
32581 deprecate certain builtin variants.
32582 * config/s390/s390-builtin-types.def: Add new builtin types.
32583 * config/s390/s390-builtins.h: Support new flags field for
32584 overloaded builtins.
32585 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
32586 (s390_macro_to_expand): Enable vector float data type.
32587 (s390_cpu_cpp_builtins_internal): Indicate support of the new
32588 builtins by incrementing the __VEC__ version number.
32589 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
32591 (s390_resolve_overloaded_builtin): Emit error messages depending
32592 on the builtin flags.
32593 * config/s390/s390.c (s390_expand_builtin): Support additional
32594 flags argument. Change error message to match the messages
32595 emitted in s390-c.c.
32596 * config/s390/s390.md: New UNSPEC_* constants.
32597 (op_type): Add new instruction types.
32598 * config/s390/vecintrin.h: Add new builtins and test data class
32600 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
32601 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
32602 (VEC_INEXACT, VEC_NOINEXACT): New constants.
32603 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
32604 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
32605 ("vec_mergel<mode>"): V_HW -> VEC_HW.
32607 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
32608 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
32609 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
32610 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
32612 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
32613 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
32614 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
32615 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
32617 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
32618 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
32619 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
32620 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
32621 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
32622 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
32623 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
32625 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
32626 ("vec_scatter_element<V_HW_4:mode>_DI")
32627 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
32628 ("vec_fpint<mode>", "vflls")
32629 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
32630 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
32631 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
32632 ("*vec_cmphe<mode>_cc"): ... these.
32634 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
32635 mode constant instead of magic value.
32637 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32639 * config/s390/s390.c (s390_expand_vec_compare): Support other
32640 vector floating point modes than just V2DF.
32641 (s390_expand_vcond): Likewise.
32642 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
32643 (s390_cannot_change_mode_class): Prevent mode changes between TF
32644 and V1TF in vector registers.
32645 * config/s390/s390.md (DF, SF): New mode attributes.
32646 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
32647 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
32648 SFmode support for VRs.
32649 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
32651 (VFT, VF_HW): New mode iterators.
32652 (vw, sdx): New mode attributes.
32653 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
32654 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
32655 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
32656 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
32657 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
32658 also the new vector floating point modes. Renaming to ...
32660 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
32661 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
32662 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
32663 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
32664 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
32665 ("vec_unordered<mode>"): ... these.
32667 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
32668 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
32669 ("*vec_extendv2df"): New insn definitions.
32671 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32673 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
32674 ("mulditi3_2", "*muldi3_sign"): New patterns.
32675 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
32676 rename the pattern definition.
32678 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32680 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
32682 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
32684 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32686 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
32687 instruction if possible.
32688 * config/s390/vector.md (vec_halfnumelts): New mode
32690 ("*vec_vllezlf<mode>"): New pattern.
32692 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32694 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
32695 ("popcountv4si2", "popcountv2di2"): Rename to ...
32696 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
32697 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
32699 ("popcount<mode>2_vxe"): New pattern.
32701 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32703 * common/config/s390/s390-common.c (processor_flags_table): Add
32705 * config.gcc: Add arch12.
32706 * config/s390/driver-native.c (s390_host_detect_local_cpu):
32707 Default to arch12 for unknown CPU model numbers.
32708 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
32709 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
32710 PROCESSOR_max sanity check.
32711 * config/s390/s390-opts.h (enum processor_type): Add
32713 * config/s390/s390.c (processor_table): Add arch12.
32714 (s390_expand_builtin): Add check for B_VXE flag.
32715 (s390_issue_rate): Add PROCESSOR_ARCH12.
32716 (s390_get_sched_attrmask): Likewise.
32717 (s390_get_unit_mask): Likewise.
32718 (s390_sched_score): Enable z13 scheduling for arch12.
32719 (s390_sched_reorder): Likewise.
32720 (s390_sched_variable_issue): Likewise.
32721 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
32723 (s390_tune_attr): Use z13 scheduling also for arch12.
32724 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
32725 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
32726 (TARGET_VXE_P): New macros.
32727 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
32728 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
32729 * config/s390/s390.opt: Add arch12 as processor_type.
32731 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32733 * config/s390/s390.md
32734 ("fixuns_truncdddi2", "fixuns_trunctddi2")
32735 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
32736 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
32738 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
32739 Rename expanders to ...
32741 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
32742 ("fixuns_truncdddi2_emu"): ... these.
32744 ("fixuns_trunc<mode>si2_emu"): New expander.
32746 ("*fixuns_truncdfdi2_z13"): Rename to ...
32747 ("*fixuns_truncdfdi2_vx"): ... this.
32749 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32751 * config/s390/2964.md: Remove the single element vector compare
32752 instructions which are no longer used.
32753 * config/s390/s390.c (s390_select_ccmode): Remove handling of
32755 (s390_canonicalize_comparison): Remove handling of DFmode
32757 (s390_expand_vec_compare_scalar): Remove function.
32758 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
32759 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
32761 ("*cmp<mode>_ccs"): Add wfcdb instruction.
32763 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32765 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
32767 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
32768 will anyway by matched by mov<mode>_64dfp.
32770 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32772 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
32773 vlef/vstef. Add missing operand to vleif.
32775 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32777 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
32778 pair for all vector types with 64 bit elements.
32779 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
32780 * config/s390/vector.md (V_HW_64): ... here.
32781 (V_128_NOSINGLE): New mode iterator.
32782 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
32783 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
32784 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
32785 ("*vec_load_pairv2di"): Change to ...
32786 ("*vec_load_pair<mode>"): ... this one.
32788 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32790 * config/s390/constraints.md: Add comments.
32791 (jKK): Reject element sizes > 8 bytes.
32792 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
32794 * config/s390/s390.md: Add the s_operand checks formerly in
32795 s390_split_ok_p to various splitters where they are still
32797 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
32798 for 128 bit vectors. Plus two splitters.
32800 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32802 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
32805 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32808 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
32809 error if the boundary argument is not constant.
32811 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32813 PR rtl-optimization/80112
32814 * loop-doloop.c (doloop_condition_get): Don't check condition
32815 if cmp isn't SET with IF_THEN_ELSE src.
32817 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32819 PR tree-optimization/80158
32820 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
32821 replacing a candidate statement, also replace it for the
32822 candidate's alternate interpretation.
32823 (replace_rhs_if_not_dup): Likewise.
32824 (replace_one_candidate): Likewise.
32826 2017-03-24 Richard Biener <rguenther@suse.de>
32828 PR tree-optimization/80167
32829 * graphite-isl-ast-to-gimple.c
32830 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
32832 (translate_isl_ast_to_gimple::get_rename): Likewise.
32834 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32836 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
32837 handling of certain combinations of target options, including the
32838 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
32839 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
32841 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32844 * config/arm/arm.md (*load_multiple): Add reload_completed to
32845 matching condition.
32847 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32848 Richard Biener <rguenth@suse.de>
32850 PR tree-optimization/79908
32851 PR tree-optimization/80136
32852 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32853 been cast away, gimplify_and_add suffices.
32855 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
32857 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
32859 2017-03-23 Richard Biener <rguenther@suse.de>
32861 PR tree-optimization/80032
32862 * gimplify.c (gimple_push_cleanup): Forced unconditional
32863 cleanups still have to go to the conditional_cleanups
32866 2017-03-22 Jakub Jelinek <jakub@redhat.com>
32868 PR tree-optimization/80072
32869 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
32871 (next_operand_entry_id): Change type to unsigned int.
32872 (sort_by_operand_rank): Make sure to return the right return value
32873 even if unsigned fields are bigger than INT_MAX.
32874 (struct oecount): Change cnt and id type to unsigned int.
32875 (oecount_hasher::equal): Formatting fix.
32876 (oecount_cmp): Make sure to return the right return value
32877 even if unsigned fields are bigger than INT_MAX.
32878 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
32881 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
32882 TREE_READONLY on result if writing it more than once.
32885 * doc/invoke.texi (-fsanitize=thread): Document that with
32886 -fnon-call-exceptions atomics are not able to throw
32890 * tsan.c: Include tree-eh.h.
32891 (instrument_builtin_call): Call maybe_clean_eh_stmt or
32892 maybe_clean_or_replace_eh_stmt where needed.
32893 (instrument_memory_accesses): Add cfg_changed argument.
32894 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
32895 if it returned true.
32896 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
32898 PR rtl-optimization/63191
32899 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
32900 wrapper function, moved the whole old content into ...
32901 (ix86_delegitimize_address_1): ... this. New inline function.
32902 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
32903 true as last argument instead of ix86_delegitimize_address.
32905 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32907 * config/aarch64/aarch64.c (generic_branch_cost): Copy
32908 cortexa57_branch_cost.
32910 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32912 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
32914 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32917 * doc/md.texi (Constraints): Document wA constraint.
32918 * config/rs6000/constraints.md (wA): New.
32919 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
32920 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
32921 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
32922 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
32924 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
32927 * gimplify.c (is_oacc_declared): New function.
32928 (oacc_default_clause): Use it to set default flags for acc declared
32929 variables inside parallel regions.
32930 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
32931 declared variables.
32932 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
32933 declare attribute to any decl as necessary.
32935 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
32938 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
32939 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
32940 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
32941 (arm_arch_lpae): This.
32942 * config/arm/arm.c (arm_arch7ve): Rename into ...
32943 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
32944 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
32947 2017-03-22 Martin Liska <mliska@suse.cz>
32950 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
32951 error message instead of an ICE.
32953 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32955 * doc/extend.texi (6.11 Additional Floating Types): Revise.
32957 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
32959 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
32961 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32964 2017-03-21 Martin Sebor <msebor@redhat.com>
32966 * doc/extend.texi: Use "cannot" instead of "can't."
32967 * doc/hostconfig.texi: Same.
32968 * doc/install.texi: Same.
32969 * doc/invoke.texi: Same.
32970 * doc/loop.texi: Same.
32971 * doc/md.texi: Same.
32972 * doc/objc.texi: Same.
32973 * doc/rtl.texi: Same.
32974 * doc/tm.texi: Same.
32975 * doc/tm.texi.in: Same.
32976 * doc/trouble.texi: Same.
32978 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
32981 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
32982 (collect_checksum_attributes): Set it.
32983 (die_checksum_ordered): Use it.
32985 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32987 PR tree-optimization/79908
32988 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
32989 change: For a VA_ARG whose LHS has been cast away, use
32990 force_gimple_operand to construct the side effects.
32992 2017-03-21 David Malcolm <dmalcolm@redhat.com>
32994 PR translation/80001
32995 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
32996 more amenable to translation.
32997 (oacc_loop_auto_partitions): Likewise.
32999 2017-03-21 Marek Polacek <polacek@redhat.com>
33000 Martin Sebor <msebor@redhat.com>
33002 PR tree-optimization/80109
33003 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
33004 on INTEGRAL_TYPE_P.
33006 2017-03-21 Jakub Jelinek <jakub@redhat.com>
33007 Segher Boessenkool <segher@kernel.crashing.org>
33010 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
33011 check reg_used_between_p between insn and one of succ or succ2
33012 depending on if succ is artificial insn not inserted into insn
33015 2017-03-21 Martin Liska <mliska@suse.cz>
33017 PR gcov-profile/80081
33018 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
33019 * doc/gcc.texi: Include gcov-dump stuff.
33020 * doc/gcov-dump.texi: New file.
33022 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
33024 PR rtl-optimization/79150
33025 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
33026 conditional jump, if the jump is the last insn of the loop.
33028 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33029 Richard Biener <rguenth@suse.de>
33031 PR tree-optimization/79908
33032 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
33033 been cast away, use force_gimple_operand to construct the side
33036 2017-03-21 Martin Liska <mliska@suse.cz>
33038 PR libfortran/79956
33039 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
33042 2017-03-21 Brad Spengler <spender@grsecurity.net>
33045 * plugin.c (htab_hash_plugin): New function.
33046 (add_new_plugin): Use it and adjust.
33047 (parse_plugin_arg_opt): Adjust.
33048 (init_one_plugin): Likewise.
33050 2017-03-21 Richard Biener <rguenther@suse.de>
33052 PR tree-optimization/80032
33053 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
33054 if set force the cleanup to happen unconditionally.
33055 (gimplify_target_expr): Push inserted clobbers with force_uncond
33056 to avoid them being removed by control-dependent DCE.
33058 2017-03-21 Richard Biener <rguenther@suse.de>
33060 PR tree-optimization/80122
33061 * tree-inline.c (copy_bb): Do not expans va-arg packs or
33062 va_arg_pack_len when the inlined call stmt requires pack
33064 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
33066 2017-03-21 Jakub Jelinek <jakub@redhat.com>
33069 * tsan.c (instrument_builtin_call): If the memory model argument
33070 is not a constant, assume it is valid.
33073 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
33076 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
33078 PR rtl-optimization/79910
33079 * combine.c (can_combine_p): Do not allow combining an I0 or I1
33080 if its dest is used by an insn before I2 (other than the combined
33081 insns themselves, which are properly handled already).
33083 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
33086 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
33088 * combine.c (record_used_regs): New static function.
33089 (try_combine): Handle situations where there is an additional
33090 instruction between I2 and I3 which needs to have a LOG_LINK
33094 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
33096 * combine.c (try_combine): Delete redundant i1 test. Call
33097 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
33099 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33102 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
33103 alternatives 13/14.
33105 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33107 PR tree-optimization/80054
33108 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
33109 the optimization if a PHI or any of its arguments is not dominated
33110 by the candidate's basis. Use gphi* rather than gimple* as
33112 (replace_profitable_candidates): Clean up a gimple* variable that
33113 should be a gphi* variable.
33115 2017-03-20 Martin Sebor <msebor@redhat.com>
33118 * doc/extend.texi (attribute constructor): Document present limitation.
33120 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
33123 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
33124 __POWER9_VECTOR__ #ifdef control, change template definition to
33125 use Power9-specific built-in function.
33126 (vec_any_eq): Likewise.
33127 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
33128 to control outcomes from this test.
33129 (vector_ae_<mode>p): For VEC_F modes, likewise.
33131 2017-03-20 Ian Lance Taylor <iant@google.com>
33133 * config/i386/i386.c (ix86_function_regparm): Save an extra
33134 register for -fsplit-stack with DECL_STATIC_CHAIN.
33136 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
33139 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
33140 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
33142 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
33144 * config/riscv/riscv.c (riscv_print_operand): Use "fence
33146 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
33149 2017-03-20 Marek Polacek <polacek@redhat.com>
33152 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
33154 2017-03-20 Richard Biener <rguenther@suse.de>
33156 PR tree-optimization/80113
33157 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
33158 allocate extra SSA name for PHI def.
33159 (add_close_phis_to_outer_loops): Likewise.
33160 (add_close_phis_to_merge_points): Likewise.
33161 (copy_loop_close_phi_args): Likewise.
33162 (copy_cond_phi_nodes): Likewise.
33164 2017-03-20 Martin Liska <mliska@suse.cz>
33166 PR middle-end/79753
33167 * tree-chkp.c (chkp_build_returned_bound): Do not build
33168 returned bounds for a LHS that's not a BOUNDED_P type.
33170 2017-03-20 Martin Liska <mliska@suse.cz>
33174 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
33175 COMPLEX_CST and VECTOR_CST.
33177 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33180 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
33181 target operand. A new splitter adds the clobber statement in case
33182 the target operand is dead anyway.
33184 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
33186 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
33187 to age-old versions of binutils and glibc.
33189 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
33191 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
33193 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
33195 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
33197 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
33199 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
33200 requirement for binutils 2.13.
33202 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
33204 * combine.c (try_combine): Delete redundant i1 test. Call
33205 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
33207 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
33209 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
33210 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
33212 <riscv64-*-elf>: Re-arrange section
33213 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
33214 <riscv32-*-linux>: Likewise.
33215 <riscv64-*-elf>: Likewise
33216 <riscv64-*-linux>: Likewise.
33218 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
33221 * aarch64.opt(verbose-cost-dump): Fix typo.
33223 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
33226 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
33227 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
33229 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
33231 * reload.c (find_reloads): When reloading a nonoffsettable address,
33232 use RELOAD_OTHER for it and its address reloads.
33234 PR rtl-optimization/79910
33235 * combine.c (record_used_regs): New static function.
33236 (try_combine): Handle situations where there is an additional
33237 instruction between I2 and I3 which needs to have a LOG_LINK
33240 2017-03-17 Jeff Law <law@redhat.com>
33242 PR tree-optimization/71437
33243 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
33244 conditional in the hash table first.
33245 (vrp_dom_walker::before_dom_children): Extract condition from
33246 ASSERT_EXPR. Record condition, its inverion and any implied
33247 conditions as well.
33249 2017-03-17 Marek Polacek <polacek@redhat.com>
33250 Markus Trippelsdorf <markus@trippelsdorf.de>
33252 PR tree-optimization/80079
33253 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
33256 2017-03-17 Richard Biener <rguenther@suse.de>
33258 PR middle-end/80075
33259 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
33260 Properly verify the LHS before the RHS possibly claims to be
33262 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
33265 2017-03-17 Martin Jambor <mjambor@suse.cz>
33267 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
33268 (List of -O2 options): Likewise.
33269 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
33272 2017-03-17 Tom de Vries <tom@codesourcery.com>
33274 * gcov-dump.c (print_usage): Print bug_report_url.
33276 2017-03-17 Richard Biener <rguenther@suse.de>
33278 PR middle-end/80050
33279 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
33280 (parser::peek): Likewise.
33282 2017-03-17 Richard Biener <rguenther@suse.de>
33284 PR tree-optimization/80048
33285 * sese.c (free_sese_info): Properly release rename_map and
33286 copied_bb_map elements.
33288 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
33290 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
33291 Add linked-list forward and backlinks. Insert on
33292 construction, remove on destruction.
33293 (class pass_store_merging): Add m_stores_head field.
33294 (pass_store_merging::terminate_and_process_all_chains):
33295 Iterate over m_stores_head list.
33296 (pass_store_merging::terminate_all_aliasing_chains):
33298 (pass_store_merging::execute): Check for debug stmts first.
33299 Push new chains onto the m_stores_head stack.
33301 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
33304 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
33305 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
33306 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
33308 2017-03-16 Jeff Law <law@redhat.com>
33310 PR tree-optimization/71437
33311 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
33312 member function. Implementation moved into after_dom_children
33313 member function and into the threader's thread_outgoing_edges
33315 (dom_opt_dom_walker::after_dom_children): Simplify by moving
33316 some code into new thread_outgoing_edges.
33317 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
33318 definition. Simplify marker handling (do it here). Assume we always
33319 have the available expression and the const/copies tables.
33320 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
33322 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
33323 * tree-vrp.c (equiv_stack): No longer file scoped.
33324 (vrp_dom_walker): New class.
33325 (vrp_dom_walker::before_dom_children): New member function.
33326 (vrp_dom_walker::after_dom_children): Likewise.
33327 (identify_jump_threads): Setup domwalker. Use it rather than
33328 walking edges in a random order by hand. Simplify setup/finalization.
33329 (finalize_jump_threads): Remove.
33330 (vrp_finalize): Do not call identify_jump_threads here.
33331 (execute_vrp): Do it here instead and call thread_through_all_blocks
33334 PR tree-optimization/71437
33335 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
33337 (simplify_stmt_for_jump_threading): Add basic_block argument. All
33339 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
33340 (dom_opt_dom_walker::thread_across_edge): Remove
33341 handle_dominating_asserts argument. All callers changed.
33342 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
33343 changes. Remove calls to lhs_of_dominating_assert. Other
33344 uses of handle_dominating_asserts turn into unconditional code
33345 (simplify_control_stmt_condition_1): Likewise.
33346 (simplify_control_stmt_condition): Likewise.
33347 (thread_through_normal_block, thread_across_edge): Likewise.
33348 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
33349 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
33350 object if it is not an SSA_NAME.
33351 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
33352 before calling into the VRP specific simplifiers.
33353 (identify_jump_threads): Remove handle_dominating_asserts
33356 2017-03-16 Jakub Jelinek <jakub@redhat.com>
33359 * tree-diagnostic.c (default_tree_printer): No longer static.
33360 * tree-diagnostic.h (default_tree_printer): New prototype.
33362 2017-03-16 Tamar Christina <tamar.christina@arm.com>
33364 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
33365 Change ins into fmov.
33367 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33369 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
33370 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
33371 Use h_con constraint for operand 1.
33372 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
33373 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
33375 2017-03-15 Jeff Law <law@redhat.com>
33377 PR tree-optimization/71437
33378 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
33379 (record_temporary_equivalences): Use it.
33381 PR tree-optimization/71437
33382 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
33383 tree-ssa-scopedtables.
33384 (lookup_avail_expr, build_and_record_new_cond): Likewise.
33385 (record_conditions, record_cond, vuse_eq): Likewise.
33386 (record_edge_info): Adjust to API tweak of record_conditions.
33387 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
33388 (record_temporary_equivalences, optimize_stmt): Likewise.
33389 (eliminate_redundant_computations): Likewise.
33390 (record_equivalences_from_stmt): Likewise.
33391 * tree-ssa-scopedtables.c: Include options.h and params.h.
33392 (vuse_eq): New function, moved from tree-ssa-dom.c
33393 (build_and_record_new_cond): Likewise.
33394 (record_conditions): Likewise. Accept vector of conditions rather
33395 than edge_equivalence structure for first argument.
33396 for the first argument.
33397 (avail_exprs_stack::lookup_avail_expr): New member function, moved
33398 from tree-ssa-dom.c.
33399 (avail_exprs_stack::record_cond): Likewise.
33400 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
33401 from tree-ssa-dom.c.
33402 (avail_exprs_stack): Add new member functions lookup_avail_expr
33404 (record_conditions): Declare.
33406 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
33409 * lra-constraints.c (process_alt_operands): Increase reject for
33410 reloading an input/output operand.
33412 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
33415 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
33416 insns to convert from signed/unsigned char/short to IEEE 128-bit
33418 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
33420 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
33423 * config/i386/i386.c (ix86_vector_duplicate_value): Create
33424 subreg of inner mode for values already in registers.
33426 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
33428 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
33429 iteration reg is used after the loop.
33431 2017-03-14 Martin Sebor <msebor@redhat.com>
33433 PR tree-optimization/79800
33434 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
33435 precision in negative-positive range.
33436 (format_floating): Call non-const overload with adjusted precision.
33438 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33441 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
33444 2017-03-14 Martin Sebor <msebor@redhat.com>
33446 PR middle-end/80020
33447 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
33448 * builtins.def (aligned_alloc): Use it.
33451 * Makefile.in (GTFILES): Add calls.c.
33452 * calls.c: Include "gt-calls.h".
33454 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
33456 PR rtl-optimization/79728
33457 * regs.h (struct target_regs): New field
33458 x_contains_allocatable_regs_of_mode.
33459 (contains_allocatable_regs_of_mode): New macro.
33460 * reginfo.c (init_reg_sets_1): Initialize it, and change
33461 contains_reg_of_mode so it includes global regs as well.
33462 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
33463 rather than contains_regs_of_mode.
33465 2017-03-14 Martin Liska <mliska@suse.cz>
33467 * doc/invoke.texi: Document options that can't be combined with
33468 -fcheck-pointer-bounds.
33470 2017-03-14 Martin Liska <mliska@suse.cz>
33472 PR middle-end/79831
33473 * doc/invoke.texi (-Wchkp): Document the option.
33475 2017-03-14 Martin Liska <mliska@suse.cz>
33477 * Makefile.in: Install gcov-dump.
33479 2017-03-14 Martin Liska <mliska@suse.cz>
33481 * multiple_target.c (expand_target_clones): Bail out for
33482 an invalid attribute.
33484 2017-03-14 Richard Biener <rguenther@suse.de>
33486 * alias.c (struct alias_set_entry): Pack properly.
33487 * cfgloop.h (struct loop): Likewise.
33488 * cse.c (struct set): Likewise.
33489 * ipa-utils.c (struct searchc_env): Likewise.
33490 * loop-invariant.c (struct invariant): Likewise.
33491 * lra-remat.c (struct cand): Likewise.
33492 * recog.c (struct change_t): Likewise.
33493 * rtl.h (struct address_info): Likewise.
33494 * symbol-summary.h (function_summary): Likewise.
33495 * tree-loop-distribution.c (struct partition): Likewise.
33496 * tree-object-size.c (struct object_size_info): Likewise.
33497 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
33498 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
33499 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
33500 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
33501 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
33502 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
33503 (struct _stmt_vec_info): Likewise.
33505 2017-03-14 Martin Liska <mliska@suse.cz>
33508 * multiple_target.c (create_dispatcher_calls): Check that
33509 a target can create a function dispatcher.
33511 2017-03-14 Martin Liska <mliska@suse.cz>
33514 * multiple_target.c (expand_target_clones): Drop local.local
33515 flag for default implementation.
33517 2017-03-14 Richard Biener <rguenther@suse.de>
33519 PR tree-optimization/80030
33520 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
33522 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
33524 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
33525 gcc_fallthrough() instead of __attribute__((fallthrough));
33527 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
33529 * doc/gcc.texi: Remove "up" link to (DIR).
33530 * doc/gccint.texi: Ditto.
33532 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
33534 * doc/install.texi (Specific) <avr>: Remove reference to
33537 2017-03-13 Jeff Law <law@redhat.com>
33539 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
33540 attribute rather than comments.
33542 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
33543 match_scratch operand is highest.
33545 2017-03-13 Martin Liska <mliska@suse.cz>
33547 PR middle-end/78339
33548 * ipa-pure-const.c (warn_function_noreturn): If the declarations
33549 is a CHKP clone, use original declaration.
33551 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33553 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
33554 (arc_conditional_register_usage): Use a different allocation order
33555 when optimizing for size.
33556 * common/config/arc/arc-common.c (arc_option_optimization_table):
33557 Section anchors default on when optimizing for size.
33559 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33561 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
33563 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33565 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
33566 * config/arc/arc.md (cpu_facility): Add cd variant.
33567 (*movqi_insn): Add code density variant.
33568 (*movhi_insn): Likewise.
33569 (*movqi_insn): Likewise.
33570 (*addsi3_mixed): Likewise.
33571 (subsi3_insn): Likewise.
33573 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33575 * config/arc/arc.md (movsi_cond_exec): Update constraint.
33577 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33579 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
33580 expressions with MINUS and UNARY ops.
33582 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33585 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
33587 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
33588 between vec_select and vector argument.
33589 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
33590 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
33591 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
33592 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
33593 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
33594 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
33596 2017-03-13 Richard Biener <rguenther@suse.de>
33599 * params.def (vect-max-peeling-for-alignment): Fix typo.
33601 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33603 * doc/install.texi (Specific) <mips-*-*>: Remove description of
33604 issue that only occurred with binutils below 2.18.
33606 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33608 * doc/install.texi (Specific) <cris-axis-elf>: No longer
33609 refer to binutils 2.11/2.12 minimum.
33611 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33613 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
33614 ftp.kernel.org and simplify binutils requirement.
33616 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
33618 * doc/invoke.texi (Warning Options): Fix spelling of link-time
33620 (Optimize Options): Ditto. Also remove redundancy.
33622 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33624 PR translation/79848
33625 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
33627 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
33628 to G_ to avoid double translation.
33630 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33632 PR translation/79923
33633 * auto-profile.c (get_combined_location): Convert leading
33634 character of diagnostics to lower case and remove trailing period.
33635 (read_profile): Likewise for various diagnostics.
33636 * config/arm/arm.c (arm_option_override): Remove trailing period
33637 from various diagnostics.
33638 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
33639 (msp430_expand_delay_cycles): Likewise.
33641 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33644 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
33645 full command-line argument, rather than just "str".
33646 (aarch64_validate_march): Likewise.
33647 (aarch64_validate_mtune): Likewise.
33649 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
33651 PR rtl-optimization/78911
33652 * lra-assigns.c (must_not_spill_p): New function.
33653 (spill_for): Use it.
33655 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33657 PR tree-optimization/79981
33658 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
33659 ATOMIC_COMPARE_EXCHANGE ifn result.
33660 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
33661 IFN_ATOMIC_COMPARE_EXCHANGE.
33663 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33666 * opts.c (parse_sanitizer_options): Add missing question mark to
33667 "did you mean" message.
33669 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33671 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
33673 (VMULEUH_UNS): Likewise.
33674 (VMULOUB_UNS): Likewise.
33675 (VMULOUH_UNS): Likewise.
33676 * config/rs6000/rs6000.c (builtin_function_type): Remove
33677 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
33679 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33682 * read-rtl-function.c (function_reader::read_rtx_operand): Update
33683 x with result of extra_parsing_for_operand_code_0.
33684 (function_reader::extra_parsing_for_operand_code_0): Convert
33685 return type from void to rtx, returning x. When reading
33686 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
33687 larger size containing struct block_symbol.
33689 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
33691 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
33692 -mfloat128-hardware without -m64.
33694 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
33697 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
33698 entries to the case statement that marks unsigned arguments to
33699 overloaded functions.
33701 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33703 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
33704 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
33706 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
33709 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
33710 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
33712 2017-03-10 Martin Liska <mliska@suse.cz>
33716 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
33717 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
33720 2017-03-10 Olivier Hainque <hainque@adacore.com>
33722 * tree-switch-conversion (array_value_type): Start by resetting
33723 candidate type to it's main variant.
33725 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33727 PR rtl-optimization/79909
33728 * combine.c (try_combine): Use simplify_replace_rtx on individual
33729 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
33730 of the whole CALL_INSN_FUNCTION_USAGE.
33732 PR tree-optimization/79972
33733 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
33734 get_range_info on SSA_NAMEs. Formatting fixes.
33736 2017-03-10 Richard Biener <rguenther@suse.de>
33737 Jakub Jelinek <jakub@redhat.com>
33739 PR tree-optimization/77975
33740 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
33741 edge to be constant.
33742 (get_val_for): For constant x return it. Formatting fix.
33743 (loop_niter_by_eval): Avoid pointless looping if the next iteration
33744 would use the same bases as the current one.
33746 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33748 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
33749 instead of vec_select for V1TImode.
33750 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
33752 (VSX_LE_128): Add V1TI to this mode iterator.
33753 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
33754 (*vsx_le_perm_store_<mode>): Likewise.
33755 (pre-reload splitter for VSX stores): Likewise.
33756 (post-reload splitter for VSX stores): Likewise.
33757 (*vsx_xxpermdi2_le_<mode>): Likewise.
33758 (*vsx_lxvd2x2_le_<mode>): Likewise.
33759 (*vsx_stxvd2x2_le_<mode>): Likewise.
33761 2017-03-09 Michael Eager <eager@eagercon.com>
33763 Correct failures with --enable-checking=yes,rtl.
33765 * config/microblaze/microblaze.c (microblaze_expand_shift):
33766 Replace GET_CODE test with CONST_INT_P and INTVAL test with
33767 test for const0_rtx.
33768 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
33769 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
33771 2017-03-09 Richard Biener <rguenther@suse.de>
33773 PR tree-optimization/79977
33774 * graphite-scop-detection.c (scop_detection::merge_sese):
33775 Handle the case of extra exits to blocks dominating the entry.
33777 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
33779 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
33782 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
33784 PR rtl-optimization/79949
33785 * lra-constraints.c (process_alt_operands): Check memory when
33786 trying to predict a cycle. Print about the overall increase.
33788 2017-03-09 Richard Biener <rguenther@suse.de>
33790 PR middle-end/79971
33791 * gimple-expr.c (useless_type_conversion_p): Preserve
33792 TYPE_SATURATING for fixed-point types.
33794 2017-03-09 Richard Biener <rguenther@suse.de>
33797 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
33798 alignment of BLKmode params.
33800 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33803 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
33804 (VALL_NO_V2Q): Likewise.
33806 * config/aarch64/aarch64-simd.md
33807 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
33809 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
33810 VALL_NO_V2Q mode iterator.
33811 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
33813 2017-03-09 Martin Liska <mliska@suse.cz>
33815 PR tree-optimization/79631
33816 * tree-chkp-opt.c (chkp_is_constant_addr): Call
33817 tree_int_cst_sign_bit just for INTEGER constants.
33819 2017-03-09 Martin Liska <mliska@suse.cz>
33823 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
33826 2017-03-09 Marek Polacek <polacek@redhat.com>
33829 * tree.c (inchash::add_expr): Handle TREE_VEC.
33831 2017-03-09 Martin Liska <mliska@suse.cz>
33834 (chkp_narrow_size_and_offset): New function.
33835 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
33836 (void chkp_parse_bit_field_ref): New function.
33837 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
33838 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
33840 2017-03-09 Martin Liska <mliska@suse.cz>
33843 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
33844 (chkp_find_bounds_1): Remove gcc_unreachable.
33846 2017-03-09 Jakub Jelinek <jakub@redhat.com>
33849 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
33850 BUILT_IN_SYNC*, determine the access type from the size suffix and
33851 always build a MEM_REF with that type. Handle forgotten
33852 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
33855 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
33856 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
33857 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
33858 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
33859 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
33860 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
33861 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
33862 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
33863 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
33864 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
33865 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
33866 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
33867 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
33868 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
33869 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
33870 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
33871 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
33872 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
33873 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
33874 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
33875 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
33876 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
33877 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
33878 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
33879 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
33880 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
33881 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
33882 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
33883 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
33884 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
33885 definitions outside of __OPTIMIZE__ guarded section.
33888 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
33889 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
33890 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
33891 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
33894 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33896 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
33897 ("vfenez<mode>"): Add missing constraints.
33899 2017-03-08 Martin Sebor <msebor@redhat.com>
33902 * config/nds32/nds32.c (nds32_option_override):
33903 Fix misspelled diagnostic.
33905 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33908 * gimplify.c (gimplify_omp_for): Replace index var in outer
33909 taskloop statement with an artificial variable and add
33910 OMP_CLAUSE_PRIVATE clause for it.
33912 2017-03-08 Richard Biener <rguenther@suse.de>
33914 PR tree-optimization/79955
33915 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
33916 for accesses that are completely outside of the variable.
33918 2017-03-08 Andrew Haley <aph@redhat.com>
33920 PR tree-optimization/79943
33921 * tree-ssa-loop-split.c (compute_new_first_bound): When
33922 calculating the new upper bound, (END-BEG) should be added, not
33925 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33927 * config/avr/avr.md (setmemhi): Make sure match_dup
33928 operand number comes before match_scratch.
33930 2017-03-08 Richard Biener <rguenther@suse.de>
33932 PR tree-optimization/79920
33933 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
33934 with ncopies == 1 to ...
33935 (vect_transform_slp_perm_load): ... here. Properly compute
33936 all element loads by iterating VF times over the group. Do
33937 not handle ncopies (computed in a broken way) in
33938 vect_create_mask_and_perm.
33940 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33943 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
33944 is a uniform vector, use uniform_vector_p return value instead of
33945 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
33947 2017-03-07 Marek Polacek <polacek@redhat.com>
33949 PR middle-end/79809
33950 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
33951 (alloca_call_type): Likewise.
33953 2017-03-07 Martin Liska <mliska@suse.cz>
33955 * gcov.c (process_args): Put comment to correct location.
33957 2017-03-07 Martin Liska <mliska@suse.cz>
33959 PR middle-end/68270
33960 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
33961 Use array_at_struct_end_p instead of DECL_CHAIN (field).
33962 (chkp_narrow_bounds_for_field): Likewise.
33963 (chkp_parse_array_and_component_ref): Pass one more argument to
33966 2017-03-07 Richard Biener <rguenther@suse.de>
33968 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
33971 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
33973 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
33974 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
33976 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33979 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
33980 to end of description.
33981 (PARAM_MAX_STORES_TO_MERGE): Likewise.
33983 2017-03-07 Jakub Jelinek <jakub@redhat.com>
33985 PR rtl-optimization/79901
33986 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
33988 (*avx512f_<code><mode>3<mask_name>): ... this.
33989 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
33990 iterator instead of VI8_AVX2_AVX512BW.
33992 PR rtl-optimization/79901
33993 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
33994 min/max expander, expand it using expand_vec_cond_expr.
33997 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
34000 2017-03-06 Jakub Jelinek <jakub@redhat.com>
34003 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
34004 to void * for PCH reasons.
34005 * dwarf2out.c (output_loc_operands, output_die): Cast
34006 v.val_vec.array to unsigned char *.
34008 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
34011 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
34014 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
34016 PR rtl-optimization/79571
34017 * lra-constraints.c (process_alt_operands): Calculate static
34018 reject and subtract it from overall when only addresses will be
34021 2017-03-06 Julia Koval <julia.koval@intel.com>
34024 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
34025 incoming stack boundary to 128 for 64-bit targets.
34027 2017-03-06 Richard Biener <rguenther@suse.de>
34029 PR tree-optimization/79894
34030 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
34031 to NULL after folding it.
34033 2017-03-06 Richard Biener <rguenther@suse.de>
34035 PR tree-optimization/79824
34036 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
34037 check disabling peeling for gaps.
34039 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
34041 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
34042 attributes): Document gettimeofday.
34044 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
34046 * config/s390/s390.c (s390_option_override_internal): Set
34047 PARAM_MIN_VECT_LOOP_BOUND
34049 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
34051 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
34052 * config/s390/s390.md: Likewise.
34054 2017-03-06 Jakub Jelinek <jakub@redhat.com>
34057 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
34058 (<avx2_avx512>_perm<mode>): Rename to ...
34059 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
34061 (<avx512>_perm<mode>_mask): Rename to ...
34062 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
34064 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
34065 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
34066 instead of VI8F_256_512.
34067 (avx512f_perm<mode>): New define_expand.
34068 (avx512f_perm<mode>_mask): Likewise.
34069 (avx512f_perm<mode>_1<mask_name>): New define_insn.
34070 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
34072 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34074 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
34075 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
34077 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
34079 2017-03-06 Martin Liska <mliska@suse.cz>
34082 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
34083 when having a SSA NAME w/o VAR_DECL assigned to it.
34085 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34087 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
34088 msa_dpsub_<su>_d): Fix MODE for vec_select.
34090 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34092 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
34094 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
34096 2017-03-06 Richard Biener <rguenther@suse.de>
34098 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
34099 * plugin.c (register_plugin_info): Likewise.
34100 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
34102 2017-03-05 Jakub Jelinek <jakub@redhat.com>
34104 * config/i386/sse.md (sse_storehps, sse_storelps,
34105 avx_<castmode><avxsizesuffix>_<castmode>,
34106 avx512f_<castmode><avxsizesuffix>_<castmode>,
34107 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
34108 in condition that at least one operand is not a MEM.
34110 2017-03-03 Jakub Jelinek <jakub@redhat.com>
34112 PR middle-end/79805
34113 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
34114 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
34116 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
34117 gimple_call_nothrow_p flag based on whether original builtin can throw.
34118 If it can, emit following stmts on the fallthrough edge.
34119 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
34120 don't create new bb if inserting just debug stmts on the edge, try to
34121 insert them on the fallthru bb or just reset debug stmts.
34123 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
34126 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
34127 restore recog_data (including the operand rtxes inside it) around
34128 the call to get_insn_template.
34130 2017-03-03 Martin Sebor <msebor@redhat.com>
34132 PR tree-optimization/79699
34133 * context.c (context::~context): Free MPFR caches to avoid
34134 a memory leak on program exit.
34136 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34138 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
34139 Use wide_int::ulow () instead of .elt (0).
34141 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
34143 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
34144 (*pushxf): Limit oF constraint to 32bit targets and add oC
34145 constraint for 64bit targets.
34146 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
34147 (*pushdf): Change rmF constraint to rmC.
34149 2017-03-03 Martin Liska <mliska@suse.cz>
34151 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
34152 Remove unused variable.
34154 2017-03-03 Jakub Jelinek <jakub@redhat.com>
34157 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
34158 is a memory operand, increase num_memory.
34159 (ix86_expand_args_builtin): Likewise.
34161 2017-03-03 Jan Hubicka <jh@suse.cz>
34164 * ipa-devirt.c (maybe_record_node): Properly handle
34165 __cxa_pure_virtual visibility.
34167 2017-03-03 Martin Liska <mliska@suse.cz>
34169 PR tree-optimization/79803
34170 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
34172 (pass_loop_prefetch::execute): Disabled optimization if an
34173 assumption about L1 cache size is not met.
34175 2017-03-03 Martin Liska <mliska@suse.cz>
34177 PR rtl-optimization/79574
34178 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
34179 (hash_scan_set): Likewise.
34180 (dump_hash_table): Likewise.
34181 (hoist_code): Likewise.
34183 2017-03-03 Richard Biener <rguenther@suse.de>
34185 * fixed-value.c (fixed_from_string): Restore use of elt (1)
34186 in place of uhigh ().
34187 (fixed_convert_from_real): Likewise.
34189 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
34192 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
34194 2017-03-03 Richard Biener <rguenther@suse.de>
34196 PR middle-end/79818
34197 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
34198 TYPE_OVERFLOW_UNDEFINED check.
34200 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34202 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
34204 (vector_ae_<mode>_p): Likewise.
34205 (vector_nez_<mode>_p): Likewise.
34206 (vector_ne_v2di_p): Likewise.
34207 (vector_ae_v2di_p): Likewise.
34208 (vector_ne_<mode>_p): Likewise.
34209 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
34211 (vsx_tsqrt<mode>2_fe): Likewise.
34213 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
34216 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
34218 2017-03-02 Jakub Jelinek <jakub@redhat.com>
34220 PR rtl-optimization/79780
34221 * cprop.c (one_cprop_pass): When second and further conditional trap
34222 in a single basic block is turned into an unconditional trap, turn it
34223 into a deleted note to avoid RTL verification failures.
34225 2017-03-02 Richard Biener <rguenther@suse.de>
34227 * fold-const.c (const_binop): Use ulow () instead of elt (0).
34229 2017-03-02 Richard Biener <rguenther@suse.de>
34231 PR tree-optimization/79345
34233 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
34234 param and abort the walk, returning -1 if it is hit.
34235 (walk_aliased_vdefs): Take a limit param and pass it on.
34236 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
34237 defaulting to 0 and return a signed int.
34238 * tree-ssa-uninit.c (struct check_defs_data): New struct.
34239 (check_defs): New helper.
34240 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
34241 about uninitialized memory.
34242 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
34243 bogus uninitialized warning.
34244 (fixed_convert_from_real): Likewise.
34246 2017-03-02 Bin Cheng <bin.cheng@arm.com>
34248 PR tree-optimization/66768
34249 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
34250 iv_use if base object can't be determined.
34252 2017-03-02 Jakub Jelinek <jakub@redhat.com>
34254 PR tree-optimization/79345
34255 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
34256 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
34257 (get_pattern_stats): Initialize it.
34258 * genemit.c (gen_expand): Verify match_scratch numbers come after
34259 match_operand/match_dup numbers.
34260 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
34261 match_scratch numbers.
34262 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
34264 * config/s390/s390.md (trunctdsd2): Likewise.
34266 2017-03-02 Richard Biener <rguenther@suse.de>
34268 * wide-int.h (wide_int_storage::operator=): Implement in terms
34271 2017-03-02 Richard Biener <rguenther@suse.de>
34273 PR tree-optimization/79777
34274 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
34275 the to insert expression to sth existing.
34277 2017-03-01 Martin Sebor <msebor@redhat.com>
34279 PR middle-end/79692
34280 * gimple-ssa-sprintf.c
34281 (directive::known_width_and_precision): New function.
34282 (format_integer): Use it.
34283 (get_mpfr_format_length): Consider the full range of precision
34284 when computing %g output with the # flag. Set the likely byte
34285 count to 3 rather than 1 when precision is indeterminate.
34286 (format_floating): Correct the lower bound of precision.
34288 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34290 * doc/invoke.texi: Document default code model for 64-bit Linux.
34292 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34295 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
34296 udiv rather than div since input pattern is unsigned.
34298 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
34300 * config/i386/i386.c (print_reg): Warn for values of
34301 unsupported size in integer register.
34303 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
34306 * config/rs6000/predicates.md (current_file_function_operand): Do
34307 not allow self calls to be local if the function is replaceable.
34309 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
34312 * config/rs6000/altivec.h (vec_ctz and others): Change the
34313 preprocessor macro that controls conditional compilation from
34314 _ARCH_PWR9 to __POWER9_VECTOR__.
34315 (vec_all_ne): Change parameterization of __altivec_scalar_pred
34316 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
34317 control (instead of _ARCH_PWR9 control) so that template
34318 definition uses power9-specific function.
34319 (vec_any_eq): Likewise.
34320 (vec_all_ne): Change macro definition to use a power9-specific
34321 expansion under #ifdef __POWER9_VECTOR__ control (instead of
34322 _ARCH_PWR9 control).
34323 (vec_any_eq) Likewise.
34324 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
34325 expansion for CMPNEF to remove support for xvcmpnesp instruction.
34326 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
34327 support for xvcmpnedp instruction.
34328 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
34329 macro expansion so that Power9 implementation of vec_all_ne does
34330 not use the AltiVec predicate framework.
34331 (VCMPNEH_P): Likewise.
34332 (VCMPNEW_P): Likewise.
34333 (VCMPNED_P): Likewise.
34334 (VCMPNEFP_P): Likewise.
34335 (VCMPNEDP_P): Likewise.
34336 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
34337 implementation of vec_any_eq to not use AltiVec predicate
34339 (VCMPAEH_P): Likewise.
34340 (VCMPAEW_P): Likewise.
34341 (VCMPAED_P): Likewise.
34342 (VCMPAEFP_P): Likewise.
34343 (VCMPAEDP_P): Likewise.
34344 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
34345 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
34346 not use the AltiVec predicate framework.
34347 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
34348 of vec_any_eq to not use AltiVec predicate framework.
34349 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34350 support for predefined __POWER9_VECTOR__ macro to indicate that
34351 Power9 instruction selection is enabled.
34352 (altivec_overloaded_builtins): Remove extraneous
34353 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
34354 function argument types RS6000_BTI_bool_V16QI and
34355 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
34356 entry for overloaded function argument types RS6000_BTI_bool_V4SI
34357 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
34358 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
34359 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
34360 Power9 for implementations of vec_cmpne. Change the signature for
34361 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
34362 (representing vec_all_ne) to remove the previously described first
34363 argument of type RS6000_BTI_INTSI, as this was an artifact of
34364 reliance on the AltiVec predicate framework, which is no longer
34365 used in the implementation of these functions. Add
34366 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
34367 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
34368 since, unlike the AltiVec predicate framework implementation, we
34369 do not share function descriptors between vec_alle and vec_anyeq.
34370 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
34371 set of modes that receive special treatment even when
34372 TARGET_P9_VECTOR is true. The special treatment emits code that
34373 does not depend on Power9 instructions.
34374 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
34375 define_expand to not rely on AltiVec predicate framework.
34376 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
34378 (vector_ne_v2di_p): Change this define_expand to not rely on
34379 AltiVec predicate framework.
34380 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
34382 (vector_ne_<mode>_p): Change this define_expand to not rely on
34383 AltiVec predicate framework.
34384 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
34386 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
34387 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
34388 define_insn pattern.
34389 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
34390 define_insn pattern because the xvcmpne<VSs>. instruction is not
34392 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
34393 instruction is not supported.
34395 2017-03-01 Jakub Jelinek <jakub@redhat.com>
34397 * config/nvptx/nvptx.c: Include intl.h.
34399 2017-03-01 Martin Jambor <mjambor@suse.cz>
34402 * ipa-prop.h (ipa_bits): Removed field known.
34403 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
34404 to pointers. Adjusted their comments to warn about their sharing.
34405 (ipcp_transformation_summary): Change bits to a vector of pointers.
34406 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
34407 (ipa_get_ipa_bits_for_value): Declare.
34408 * tree-vrp.h (value_range): Mark as GTY((for_user)).
34409 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
34410 (ipa_bits_hash_table): Likewise.
34411 (ipa_vr_ggc_hash_traits): Likewise.
34412 (ipa_vr_hash_table): Likewise.
34413 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
34414 being pointers and vr_known being removed.
34415 (ipa_set_jf_unknown): Likewise.
34416 (ipa_get_ipa_bits_for_value): New function.
34417 (ipa_set_jfunc_bits): Likewise.
34418 (ipa_get_value_range): New overloaded functions.
34419 (ipa_set_jfunc_vr): Likewise.
34420 (ipa_compute_jump_functions_for_edge): Use the above functions to
34421 construct bits and vr parts of jump functions.
34422 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
34423 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
34425 (ipcp_grow_transformations_if_necessary): Also allocate
34426 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
34428 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
34429 them. Fix too long lines.
34430 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
34431 vr_known being removed.
34432 (ipa_read_jump_function): Use new setter functions to construct bits
34433 and vr parts of jump functions or set them to NULL.
34434 (write_ipcp_transformation_info): Adjust for bits being pointers.
34435 (read_ipcp_transformation_info): Likewise.
34436 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
34438 Include gt-ipa-prop.h.
34439 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
34441 (ipcp_store_bits_results): Likewise.
34442 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
34443 Do not write to existing jump functions but use a temporary instead.
34445 2017-03-01 Jakub Jelinek <jakub@redhat.com>
34448 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
34449 attempt to use its first operand as BIT_FIELD_REF base.
34451 2017-03-01 Richard Biener <rguenther@suse.de>
34453 PR middle-end/79721
34454 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
34455 interpolating formula in wrapping arithmetic.
34456 (chrec_apply): Convert chrec_evaluate return value to wanted type.
34458 2017-03-01 Jakub Jelinek <jakub@redhat.com>
34460 PR tree-optimization/79734
34461 * tree-vect-generic.c (expand_vector_condition): Optimize
34462 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
34463 Handle VEC_COND_EXPR where comparison has different inner width from
34464 type's inner width.
34466 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
34468 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
34469 markup, and similar issues. Remove @opindex entries for things
34470 that aren't options. Add missing -mmpy-option entries.
34472 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34474 PR tree-optimization/79737
34475 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
34476 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
34477 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
34478 instead of byte_size. Formatting fix.
34479 (shift_bytes_in_array_right): Formatting fix.
34481 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
34484 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
34485 condition on optimize for the leaf function test.
34487 2017-02-28 Martin Liska <mliska@suse.cz>
34490 * read-rtl-function.c (function_reader::handle_unknown_directive):
34491 Bail out when one uses -flto.
34493 2017-02-28 Martin Liska <mliska@suse.cz>
34495 * common.opt: Replace space with tabular for options of <number>
34497 * config/i386/i386.opt: Show <number> value for
34498 -mlarge-data-threshold.
34499 * opts.c (print_filtered_help): Do not display number in hexadecimal
34502 2017-02-28 Martin Liska <mliska@suse.cz>
34504 * common.opt: Fix --help=option -Q for options which are of
34507 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
34509 * config/i386/i386.c (print_reg): Error out for values
34510 of 8-bit size in invalid integer register.
34512 2017-02-28 Martin Sebor <msebor@redhat.com>
34514 PR tree-optimization/79691
34515 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
34517 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34520 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
34521 gcc_unreachable with output_operand_lossage.
34523 2017-02-28 Richard Biener <rguenther@suse.de>
34525 PR tree-optimization/79740
34526 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
34528 (visit_nary_op): Insert the nary into the hashtable if we
34529 pattern-matched sth.
34530 * tree-ssa-pre.c (eliminate_insert): Robustify.
34532 2017-02-28 Richard Biener <rguenther@suse.de>
34534 PR middle-end/79731
34535 * fold-const.c (decode_field_reference): Reject out-of-bound
34538 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34540 * config/i386/i386.c: Include intl.h.
34541 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
34542 instead of just cond ? "..." : "...".
34543 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
34544 * coverage.c (read_counts_file): Likewise.
34545 * omp-offload.c: Include intl.h.
34546 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
34547 of just cond ? "..." : "...".
34548 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
34549 of just cond ? "..." : "...".
34551 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
34554 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
34556 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
34557 'tune for' CPU name.
34558 * config/arm/arm-cpu-data.h: Regenerated.
34560 2017-02-28 Richard Biener <rguenther@suse.de>
34562 PR tree-optimization/79732
34563 * tree-inline.c (expand_call_inline): Do not shadow var.
34565 2017-02-28 Richard Biener <rguenther@suse.de>
34567 PR tree-optimization/79723
34568 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
34569 address-space properly.
34571 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
34573 * doc/optinfo.texi (Optimization groups): Fix option used for
34575 * doc/invoke.texi (-fopt-info): Document "omp".
34576 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
34577 (OPTGROUP_ALL): Add OPTGROUP_OMP.
34578 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
34579 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
34580 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
34582 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
34584 * dumpfile.c (optgroup_options): Instead of "openmp", associate
34585 OPTGROUP_OMP with "omp".
34587 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
34590 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
34591 for arithmetic shift of unsigned V2DI.
34593 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
34595 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
34596 arc/linux.h headers.
34597 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
34598 (LINK_SPEC): Likewise.
34599 (ARC_TLS_EXTRA_START_SPEC): Likewise.
34600 (EXTRA_SPECS): Likewise.
34601 (STARTFILE_SPEC): Likewise.
34602 (ENDFILE_SPEC): Likewise.
34603 (LIB_SPEC): Likewise.
34604 (TARGET_SDATA_DEFAULT): Likewise.
34605 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
34606 (MULTILIB_DEFAULTS): Likewise.
34607 (DWARF2_UNWIND_INFO): Likewise.
34608 * config/arc/big.h: New file.
34609 * config/arc/elf.h: Likewise.
34610 * config/arc/linux.h: Likewise.
34611 * config/arc/t-uClibc: Remove.
34613 2017-02-27 Bin Cheng <bin.cheng@arm.com>
34615 PR tree-optimization/77536
34616 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
34617 (tree_transform_and_unroll_loop): Use above function to compute the
34618 estimated niter of unrolled loop and use it when scaling profile.
34619 Also use count info rather than frequency if it's non-zero.
34620 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
34621 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
34622 (vect_transform_loop): Call above function.
34624 2017-02-27 Richard Biener <rguenther@suse.de>
34626 PR tree-optimization/45397
34627 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
34628 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
34629 (visit_nary_op): Add pattern matching for CSEing sign-changed
34630 or truncated operations with wider ones.
34632 2017-02-27 Richard Biener <rguenther@suse.de>
34634 PR tree-optimization/79690
34635 * tree-vect-stmts.c (vectorizable_store): Use vector type
34636 built from the DR with address-space.
34638 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
34640 * doc/invoke.texi (Optimize Options): Refine the description
34641 of asan-use-after-return.
34643 2017-02-25 Alan Modra <amodra@gmail.com>
34645 PR rtl-optimization/79584
34646 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
34647 base, not ad->base_term, the reg within base. Remove assertion
34648 that ad->base == ad->base_term. Replace gen_int_mode using
34649 bogus mode with const0_rtx.
34651 2017-02-25 Jakub Jelinek <jakub@redhat.com>
34653 PR middle-end/79396
34654 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
34655 FMA_EXPR like tcc_binary or tcc_unary.
34657 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
34660 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
34662 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
34663 (output_loc_operands): Handle DW_OP_call_ref and
34664 DW_OP_GNU_variable_value.
34665 (struct variable_value_struct): New type.
34666 (struct variable_value_hasher): Likewise.
34667 (variable_value_hash): New variable.
34668 (string_types): Remove.
34669 (copy_loc_descr): New function.
34670 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
34671 (prepend_loc_descr_to_each): New function.
34672 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
34673 instead of add_loc_descr_to_each if the first argument is single
34674 location list and the second has multiple.
34675 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
34676 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
34677 when looking for variable value which doesn't have other location info.
34678 (loc_list_from_tree): Formatting fix.
34679 (gen_array_type_die): Simplify DW_AT_string_length handling.
34680 (adjust_string_types): Remove.
34681 (gen_subprogram_die): Don't call adjust_string_types nor test/set
34682 string_types. Call resolve_variable_values.
34683 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
34684 (resolve_addr_in_expr): Likewise. Add A argument.
34685 (copy_deref_exprloc): Remove deref argument. Adjust for the
34686 original expression being DW_OP_GNU_variable_value with optionally
34687 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
34688 optionally after it.
34689 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
34690 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
34691 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
34692 (variable_value_hasher::hash, variable_value_hasher::equal): New
34694 (resolve_variable_value_in_expr, resolve_variable_value,
34695 resolve_variable_values, note_variable_value_in_expr,
34696 note_variable_value): New functions.
34697 (dwarf2out_early_finish): Call note_variable_value on all toplevel
34700 2017-02-24 Jakub Jelinek <jakub@redhat.com>
34703 * opts.h (handle_generated_option): Add GENERATED_P argument.
34704 * opts-common.c (handle_option): Adjust function comment.
34705 (handle_generated_option): Add GENERATED_P argument, pass it to
34707 (control_warning_option): Pass false to handle_generated_option
34709 * opts.c (maybe_default_option): Pass true to handle_generated_option
34711 * optc-gen.awk: Likewise.
34713 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34715 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
34716 a REG, look at the REG it is a SUBREG of.
34717 (splitter for cmpeqsi_t): Ditto.
34719 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34721 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
34722 the special USEs with the pattern of the insn, not the insn itself.
34724 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
34727 * doc/invoke.texi: Document -mload-store-pairs.
34729 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34730 Sandra Loosemore <sandra@codesourcery.com>
34732 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
34733 argument isn't a CONST_INT.
34734 (nios2_alternate_compare_const): Assert op is a CONST_INT.
34735 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
34736 (nios2_validate_compare): Bypass alternate compare logic if *op2
34737 is not a CONST_INT.
34738 (ldstwm_operation_p): Return false if first_base is not a REG or
34739 if first_offset is not a CONST_INT.
34741 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34743 * config/cris/cris.md: Use correct operand in a define_peephole2.
34745 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34747 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
34749 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34751 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
34752 this_insn if it is an INSN or JUMP_INSN.
34753 (force_offsettable): Look at base, not at addr.
34754 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
34755 on things that aren't necessarily CONST_INTs.
34757 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
34759 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
34760 -mfpmath=sse is the default also for x86-32 targets with SSE2
34761 instruction set when @option{-ffast-math} is enabled
34763 2017-02-24 Jeff Law <law@redhat.com>
34765 PR rtl-optimizatoin/79286
34766 * ira.c (update_equiv_regs): Drop may_trap_p exception to
34769 2017-02-24 Richard Biener <rguenther@suse.de>
34771 PR tree-optimization/79389
34772 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
34775 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
34777 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
34778 function comment to reflect reality.
34779 (loop_exits_before_overflow): Fix typo in function description.
34781 2017-02-24 Richard Biener <rguenther@suse.de>
34783 PR tree-optimization/79389
34784 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
34785 properly that a threading opportunity exists. Detect conditional
34786 copy/constant propagation opportunities.
34788 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
34790 * config/visium/visium.md (type): Add trap.
34791 (b): New mode attribute.
34792 (*btst): Rename into...
34793 (*btst<mode>): ...this and adjust.
34794 (*cbranchsi4_btst_insn): Rename into...
34795 (*cbranch<mode>4_btst_insn): ...this and adjust.
34796 (trap): New define_insn.
34798 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34800 PR tree-optimization/79389
34801 * ifcvt.c (struct noce_if_info): Add rev_cond field.
34802 (noce_reversed_cond_code): New function.
34803 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
34804 reversed_comparison_code. Formatting fix.
34805 (noce_try_store_flag): Test rev_cond != NULL in addition to
34806 reversed_comparison_code.
34807 (noce_try_store_flag_constants): Likewise.
34808 (noce_try_store_flag_mask): Likewise.
34809 (noce_try_addcc): Use rev_cond if non-NULL instead of
34810 reversed_comparison_code.
34811 (noce_try_cmove_arith): Likewise. Formatting fixes.
34812 (noce_try_minmax, noce_try_abs): Clear rev_cond.
34813 (noce_find_if_block): Initialize rev_cond.
34814 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
34815 instead of false as last argument never attempt to reverse it
34818 2017-02-23 Bin Cheng <bin.cheng@arm.com>
34820 PR tree-optimization/79663
34821 * tree-predcom.c (combine_chains): Process refs in reverse order
34822 only for ZERO length chains, and add explaining comment.
34824 2017-02-23 Jeff Law <law@redhat.com>
34826 PR tree-optimization/79578
34827 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
34828 in call to operand_equal_p.
34830 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
34833 * config/i386/cpuid.h: Fix another undefined behavior.
34835 2017-02-23 Richard Biener <rguenther@suse.de>
34837 PR tree-optimization/79683
34838 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
34839 vector types for data-refs.
34841 2017-02-23 Martin Liska <mliska@suse.cz>
34843 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
34845 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34847 PR middle-end/79665
34848 * internal-fn.c (get_range_pos_neg): Moved to ...
34849 * tree.c (get_range_pos_neg): ... here. No longer static.
34850 * tree.h (get_range_pos_neg): New prototype.
34851 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
34852 are known to be in between 0 and signed maximum inclusive, try to
34853 expand both unsigned and signed divmod and use the cheaper one from
34856 2017-02-22 Jeff Law <law@redhat.com>
34858 PR tree-optimization/79578
34859 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
34860 to compare base operands.
34862 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
34865 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
34866 gpc_reg_operand instead of fpr_reg_operand.
34868 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
34870 * config/mips/mips.c (mips_return_in_memory): Force FP
34871 vector types to be returned in memory for o32 ABI.
34873 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34875 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
34876 instead of DW_TAG_member for static data member declarations and don't
34877 set no_linkage_name for static inline data members.
34878 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
34881 2017-02-22 Martin Liska <mliska@suse.cz>
34883 * doc/invoke.texi: Replace inequality signs with square brackets
34886 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
34889 * lra-constraints.c (simplify_operand_subreg): Handle
34890 WORD_REGISTER_OPERATIONS targets.
34892 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34895 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
34896 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
34897 elimination by swapping fld*.
34899 2017-02-22 Richard Biener <rguenther@suse.de>
34901 PR tree-optimization/79673
34902 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
34903 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
34904 irrelevant address-space qualifiers and avoiding a
34905 ADDR_SPACE_CONVERT_EXPR from fold_convert.
34907 2017-02-22 Richard Biener <rguenther@suse.de>
34909 PR tree-optimization/79666
34910 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
34911 to not symbolically negate if that may introduce undefined
34914 2017-02-22 Martin Liska <mliska@suse.cz>
34917 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
34918 * data-streamer-out.c (streamer_write_gcov_count_stream):
34920 * value-prof.c (stream_out_histogram_value): Make assert more
34921 precise based on type of counter.
34923 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
34926 * config/i386/i386.md (standard_x87sse_constant_load splitter):
34927 Use nonimmediate_operand instead of memory_operand for operand 1.
34928 (float-extend standard_x87sse_constant_load splitter): Ditto.
34930 2017-02-21 Jeff Law <law@redhat.com>
34932 PR tree-optimization/79621
34933 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
34934 blocks with edges to themselves.
34936 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34939 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
34940 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
34941 Use gimple_call_builtin_p.
34944 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
34945 on temporarily removed DEBUG_INSNs.
34947 PR tree-optimization/79649
34948 * tree-loop-distribution.c (classify_partition): Give up on
34949 non-generic address space loads/stores.
34951 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
34953 * doc/loop.texi (Loop manipulation): Remove nonexistent
34954 tree_ssa_loop_version from the documentation.
34955 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
34957 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34960 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
34961 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
34962 * config/rs6000/rs6000.c: Include except.h.
34963 (rs6000_expand_split_stack_prologue): Call
34964 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
34966 2017-02-21 Martin Jambor <mjambor@suse.cz>
34969 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
34970 have been analyzed.
34972 2017-02-21 Martin Jambor <mjambor@suse.cz>
34974 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
34975 for backward compatibility only.
34976 * doc/invoke.texi (Option Summary): Remove all references to
34977 -fipa-cp-alignment.
34979 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
34983 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34985 * lra-constraints.c (curr_insn_transform): Handle
34986 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34988 2017-02-21 Martin Liska <mliska@suse.cz>
34990 * config/i386/i386.opt: Replace -masm-dialect with -masm.
34992 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
34994 PR translation/79638
34995 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
34997 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
35000 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
35001 (arm_function_ok_for_sibcall): Return false for an indirect call by
35002 descriptor if all the argument registers are used.
35003 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
35004 alignment of the function.
35006 2017-02-21 Jakub Jelinek <jakub@redhat.com>
35008 PR tree-optimization/61441
35009 * simplify-rtx.c (simplify_const_unary_operation): For
35010 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
35011 the sNaN unmodified.
35013 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
35015 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
35016 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
35017 instead of SYSTEM_HEADER_DIR.
35019 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
35020 Martin Liška <mliska@suse.cz>
35022 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
35023 Fix typos and grammar, use active voice, and clarify.
35025 2017-02-20 Marek Polacek <polacek@redhat.com>
35027 PR middle-end/79537
35028 * gimplify.c (gimplify_expr): Handle unused *&&L;.
35031 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
35033 2017-02-20 Jakub Jelinek <jakub@redhat.com>
35036 * config/i386/i386.c (ix86_expand_builtin): Handle
35037 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
35038 ix86_builtins_isa[fcode].isa as a requirement of those
35039 flags and any other flag in the bitmask.
35040 (ix86_init_mmx_sse_builtins): Use 0 instead of
35041 ~OPTION_MASK_ISA_64BIT as mask.
35042 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
35043 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
35044 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
35045 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
35047 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35050 * lra-constraints.c (split_reg): Check requested split mode
35051 is supported by the register.
35053 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35055 * lra-constraints.c (simplify_operand_subreg): Remove early
35058 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35061 * lra-constraints.c (curr_insn_transform): Tighten condition
35062 for converting SUBREG reloads from OP_OUT to OP_INOUT.
35064 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35067 * lra-constraints.c (curr_insn_transform): Handle
35068 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
35070 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
35073 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
35075 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
35077 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
35080 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
35083 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
35085 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
35088 2017-02-18 Jakub Jelinek <jakub@redhat.com>
35090 * final.c (last_columnnum, override_columnnum): New variables.
35091 (final_start_function): Set last_columnnum, pass it to begin_prologue
35092 hook and pass 0 to dwarf2out_begin_prologue.
35093 (final_scan_insn): Update override_columnnum. Pass last_columnnum
35094 to source_line debug hook.
35095 (notice_source_line): Compute last_columnnum and for debug_column_info
35096 return true on column changes.
35097 * debug.h (struct gcc_debug_hooks): Add column argument to
35098 source_line and begin_prologue hooks.
35099 (debug_nothing_int_charstar_int_bool): Remove prototype.
35100 (debug_nothing_int_int_charstar,
35101 debug_nothing_int_int_charstar_int_bool): New prototypes.
35102 (dwarf2out_begin_prologue): Add column argument.
35103 * debug.c (do_nothing_debug_hooks): Adjust source_line and
35104 begin_prologue hooks.
35105 (debug_nothing_int_charstar_int_bool): Remove.
35106 (debug_nothing_int_int_charstar,
35107 debug_nothing_int_int_charstar_int_bool): New functions.
35108 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
35109 through to dwarf2out_source_line.
35110 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
35111 (dwarf2out_source_line): Add column argument, emit it if requested.
35112 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
35114 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
35115 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
35116 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
35117 through to dwarf2out_begin_prologue.
35118 (vmsdbgout_source_line): Add column argument, pass it through to
35119 dwarf2out_source_line.
35120 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
35121 dbxout_source_line caller.
35122 (dbxout_source_line): Add column argument.
35124 * common.opt (gno-column-info, gcolumn-info): New options.
35125 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
35126 (check_die): Also test for multiple DW_AT_decl_column attributes.
35127 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
35128 DW_AT_decl_column if requested.
35129 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
35131 (gen_variable_die): Likewise.
35132 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
35133 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
35136 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
35137 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
35138 (ix86_handle_option): Handle OPT_m3dnowa.
35139 * doc/invoke.texi (-m3dnowa): Document.
35140 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
35141 -m3dnowa instead of -m3dnow -march=athlon.
35144 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
35145 instead of gcc_assert for K, r and R code checks. Formatting fixes.
35147 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35150 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
35151 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
35152 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
35153 generator for vsx_xxpermdi_<mode>_be.
35154 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
35155 force big-endian semantics.
35156 (vsx_xxpermdi_<mode>_be): New define_expand with same
35157 implementation as previous version of vsx_xxpermdi_<mode>.
35159 2017-02-17 Jakub Jelinek <jakub@redhat.com>
35161 PR tree-optimization/79327
35162 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
35163 variable, its initialization and use.
35165 2017-02-17 Julia Koval <julia.koval@intel.com>
35167 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
35168 (OPTION_MASK_ISA_PKU_UNSET): New.
35169 (ix86_handle_option): Handle -mrdpid.
35170 * config/i386/cpuid.h (bit_RDPID): New.
35171 * config/i386/driver-i386.c (host_detect_local_cpu):
35172 Detect RDPID feature.
35173 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
35174 * config/i386/i386-c.c (ix86_target_macros_internal):
35176 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
35177 (ix86_valid_target_attribute_inner_p): Add "rdpid".
35178 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
35179 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
35180 * config/i386/i386.md (define_insn "rdpid"): New.
35181 * config/i386/i386.opt Add -mrdpid.
35182 * config/i386/immintrin.h (_rdpid_u32): New.
35184 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
35186 PR rtl-optimization/79541
35187 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
35188 instead of transforming it into USE.
35190 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
35192 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
35193 If HONOR_SNANS (SFmode) force the input to a register.
35194 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
35195 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
35196 an frsp or similar insn.
35198 2017-02-17 Martin Liska <mliska@suse.cz>
35200 PR rtl-optimization/79577
35201 * params.def (selsched-max-sched-times): Increase minimum to 1.
35203 2017-02-17 Martin Liska <mliska@suse.cz>
35205 PR rtl-optimization/79574
35206 * gcse.c (want_to_gcse_p): Prevent integer overflow.
35208 2017-02-17 Martin Liska <mliska@suse.cz>
35210 PR tree-optimization/79529
35211 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
35212 ssa_defined_default_def_p to handle cases which are implicitly
35214 * tree-ssa.c (ssa_defined_default_def_p): New function.
35215 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
35216 which are implicitly defined.
35217 * tree-ssa.h (ssa_defined_default_def_p): Declare.
35219 2017-02-17 Richard Biener <rguenther@suse.de>
35221 PR middle-end/79576
35222 * params.def (max-ssa-name-query-depth): Limit to 10.
35224 2017-02-17 Richard Biener <rguenther@suse.de>
35226 PR tree-optimization/79552
35227 * tree-ssa-structalias.c (visit_loadstore): Properly verify
35230 2017-02-17 Richard Biener <rguenther@suse.de>
35233 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
35235 2017-02-17 Marek Polacek <polacek@redhat.com>
35237 PR middle-end/79536
35238 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
35239 (fold_negate_expr): New wrapper.
35241 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
35243 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
35244 Correct terminology and de-emphasize pre-standard behavior.
35246 2017-02-16 Alan Modra <amodra@gmail.com>
35248 PR rtl-optimization/79286
35249 * ira.c (def_dominates_uses): New function.
35250 (update_equiv_regs): Don't create an equivalence for insns that
35251 may trap where the register def does not dominate the use.
35253 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
35255 PR rtl-optimization/78127
35256 * lra.c (lra): Call lra_eliminate before finish the loop after
35259 2017-02-16 Richard Biener <rguenther@suse.de>
35261 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
35263 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
35264 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
35265 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
35266 (isl_val_int_from_wi): New function.
35267 (extract_affine_gmp): Rename to ...
35268 (extract_affine_wi): ... this, take a widest_int.
35269 (extract_affine_int): Just wrap extract_affine_wi.
35270 (add_param_constraints): Use isl_val_int_from_wi.
35271 (add_loop_constraints): Likewise, and extract_affine_wi.
35273 2017-02-15 Jeff Law <law@redhat.com>
35275 PR middle-end/79521
35276 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
35277 ira_init_register_move_cost_if_necessary.
35279 2017-02-15 Martin Sebor <msebor@redhat.com>
35281 PR middle-end/32003
35282 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
35283 removed in a prior commit.
35285 2017-02-15 Bin Cheng <bin.cheng@arm.com>
35287 PR tree-optimization/79347
35288 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
35289 counters during peeling.
35291 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
35293 * Makefile.in (site.exp): Remove "set ISLVER".
35295 2017-02-15 Jakub Jelinek <jakub@redhat.com>
35298 * real.c (real_from_integer): Call real_convert even for decimal.
35300 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
35303 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
35305 2017-02-14 Andrew Pinski <apinski@cavium.com>
35307 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
35308 cores and change the partno/implementer to be correct.
35309 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
35310 the 'B" as the implementer.
35311 * config/aarch64/aarch64-tune.md: Regenerate.
35313 2017-02-14 Carl Love <cel@us.ibm.com>
35315 * config/rs6000/rs6000.c: Add case statement entry to make the
35316 xvcvuxdsp built-in argument unsigned.
35317 * config/rs6000/vsx.md: Fix the source and return operand types so they
35318 match the instruction definitions from the ISA document. Fix typo
35319 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
35322 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
35325 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
35326 member early_clobber_alts.
35327 * lra-lives.c (reg_early_clobber_p): New.
35328 (process_bb_lives): Use it.
35329 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
35330 (debug_operand_data): Initialize early_clobber_alts.
35331 (setup_operand_alternative): Set up early_clobber_alts.
35332 (collect_non_operand_hard_regs): Ditto. Pass early clobber
35333 alternatives to new_insn_reg.
35334 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
35336 (lra_update_insn_regno_info): Pass the new arg.
35338 2017-02-14 Jakub Jelinek <jakub@redhat.com>
35340 PR middle-end/79505
35341 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
35342 (new_oacc_loop_raw): Don't clear already cleared fields.
35345 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
35346 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
35347 _mm512_prefetch_i64gather_ps): New inline functions and macros.
35349 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
35352 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
35354 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
35357 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
35358 the extra instruction to the right place to store 128-bit constant
35361 2017-02-14 Martin Sebor <msebor@redhat.com>
35363 PR middle-end/79448
35364 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
35365 warning for strings of unknown length.
35367 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
35369 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
35371 2017-02-14 Jeff Law <law@redhat.com>
35374 * ira-costs.c (scan_one_insn): Initialize register move costs
35375 for pseudos seen in USE/CLOBBER insns.
35377 PR tree-optimization/79095
35378 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
35379 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
35380 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
35381 if the operands are known to be not equal, then the resulting range
35383 (intersect_ranges): If the new range is ~[0,0] and the old range is
35384 wide, then prefer ~[0,0].
35385 * tree-vrp.c (overflow_comparison_p_1): New function.
35386 (overflow_comparison_p): New function.
35387 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
35388 if NAME is used in an overflow test.
35389 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
35390 overflow check that can be expressed as an equality test, then adjust
35391 ops to be that equality test.
35393 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35395 * config/s390/s390-builtin-types.def: Remove flags argument.
35396 * config/s390/s390.c (s390_init_builtins): Likewise.
35398 2017-02-14 Martin Liska <mliska@suse.cz>
35400 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
35401 vector. Fix trailing white spaces.
35403 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
35405 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
35408 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35410 PR rtl-optimization/68664
35411 * config/arm/arm.c (arm_sched_can_speculate_insn):
35412 New function. Declare prototype.
35413 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
35415 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35417 PR rtl-optimization/68664
35418 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
35420 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
35422 2017-02-14 Amit Pawar <amit.pawar@amd.com>
35424 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
35425 max skip bytes for function, loop and jump.
35427 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
35429 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
35430 ABS_EXPR for gimple dump.
35432 2017-02-14 Jakub Jelinek <jakub@redhat.com>
35435 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
35437 PR tree-optimization/79408
35438 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
35439 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
35440 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
35441 also if rhs1 is INTEGER_CST.
35443 2017-02-14 Richard Biener <rguenther@suse.de>
35445 PR middle-end/79432
35446 * tree-into-ssa.c (insert_phi_nodes): When the function can
35447 have abnormal edges rewrite SSA names with broken use-def
35448 dominance out of SSA and register them for PHI insertion.
35450 2017-02-13 Martin Sebor <msebor@redhat.com>
35452 PR middle-end/79496
35453 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
35454 clearing info.nowrite flag when snprintf size argument is a range.
35456 2017-02-13 Jakub Jelinek <jakub@redhat.com>
35458 * cprop.c (cprop_jump): Add missing space in string literal.
35459 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
35460 (get_constraint_for_component_ref): Likewise.
35461 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
35462 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
35463 * lra-constraints.c (process_alt_operands): Likewise.
35464 * ipa-inline.c (inline_small_functions): Likewise.
35465 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
35466 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
35467 * trans-mem.c (diagnose_tm_1_op): Likewise.
35468 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
35469 (grid_parallel_clauses_gridifiable): Likewise.
35471 * config/nvptx/mkoffload.c (process): Add space in between
35474 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
35475 "MOD4_SSE_REGS" and "ALL_REGS".
35477 * spellcheck.c (test_data): Add , in between "foo" and "food".
35479 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35482 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
35483 boundary crossing check and subsequent code generation agree.
35485 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35487 * config/aarch64/aarch64.c (has_memory_op): Delete.
35488 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
35491 2017-02-13 Jakub Jelinek <jakub@redhat.com>
35493 PR rtl-optimization/79388
35494 PR rtl-optimization/79450
35495 * combine.c (distribute_notes): When removing TEM_INSN for which
35496 corresponding dest has last value recorded, invalidate that last
35499 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35501 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
35502 of explicit '@'. Add missing assembly comment marker on branch costs
35505 2017-02-13 Nathan Sidwell <nathan@acm.org>
35507 * gengtype-lex.l (<in_struct>): Add '/'.
35509 2017-02-13 Martin Liska <mliska@suse.cz>
35512 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
35514 2017-02-13 Richard Biener <rguenther@suse.de>
35516 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
35518 * configure: Re-generate.
35519 * config.in: Likewise.
35520 * graphite-dependences.c: Simplify as if
35521 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
35522 * graphite-isl-ast-to-gimple.c: Likewise.
35523 * graphite-optimize-isl.c: Likewise.
35524 * graphite-poly.c: Likewise.
35525 * graphite-sese-to-poly.c: Likewise.
35526 * graphite.h: Likewise.
35527 * toplev.c: Include isl/version.h and use isl_version () for
35528 printing the ISL version.
35529 * doc/install.texi: Update ISL requirement.
35531 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
35533 * doc/standards.texi (Standards): Update reference to
35536 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
35538 * doc/extend.texi (Named Address Spaces): sourceware.org now
35540 * doc/install.texi (Binaries): Ditto.
35543 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
35545 * doc/cpp.texi: Replace "stringify"/"stringification" with C
35546 standard terminology "stringize"/"stringizing" throughout.
35547 * doc/cppinternals.texi: Likewise.
35549 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
35551 * doc/extend.texi: Fix some spelling mistakes and typos.
35552 * doc/invoke.texi: Likewise.
35554 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
35557 * params.def (inline-min-speedup) Change from 10 to 8.
35559 2017-02-11 Jakub Jelinek <jakub@redhat.com>
35561 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
35564 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
35567 * ipa-inline-analysis.c (get_minimal_bb): New function.
35568 (record_modified): Use it.
35569 (remap_edge_change_prob): Handle also ancestor functions.
35571 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
35573 * doc/contrib.texi (Contributors): Remove broken link into
35574 the Mauve CVS repository.
35576 2017-02-11 Jakub Jelinek <jakub@redhat.com>
35578 PR middle-end/79454
35579 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
35580 result computation whenever lhs doesn't have vector mode, not
35581 just when it has BLKmode.
35583 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35585 * doc/makefile.texi (profiledbootstrap): Refer to the
35586 installation instructions only in textual form.
35588 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35591 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
35593 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35595 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
35596 (Specific): Update mingw-w64 reference.
35598 (Specific): Remove broken link to Renesas RX processor.
35600 2017-02-10 Richard Biener <rguenther@suse.de>
35602 * toplev.c (process_options): Do not mention obsolete graphite
35603 options when printing sorry message about missing graphite support.
35604 Mention -floop-nest-optimize.
35606 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
35608 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
35609 (vtst_p16): Likewise.
35610 (vtstq_p8): Likewise.
35611 (vtstq_p16): Likewise.
35613 (vtstq_p64): Likewise.
35614 * config/arm/arm_neon.h (vgetq_lane_p64): New.
35615 (vset_lane_p64): New.
35616 (vsetq_lane_p64): New.
35618 2017-02-10 Jakub Jelinek <jakub@redhat.com>
35620 PR tree-optimization/79411
35621 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
35622 stmt operands are SSA_NAMEs used in abnormal phis.
35623 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
35626 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35629 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
35632 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35634 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
35636 2017-02-09 Jakub Jelinek <jakub@redhat.com>
35638 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
35642 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
35643 not arbitrary TREE_CONSTANT.
35646 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
35647 "omp declare target link" attribute unless is_global_var.
35648 * omp-offload.c (find_link_var_op): Likewise.
35650 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
35651 Chung-Lin Tang <cltang@codesourcery.com>
35653 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
35655 (gimplify_adjust_omp_clauses): Don't delete TILE.
35656 (gimplify_omp_for): Deal with TILE.
35657 * internal-fn.c (expand_GOACC_TILE): New function.
35658 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
35660 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
35661 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
35663 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
35664 avoid DIV for outermost collapse var.
35665 (expand_oacc_for): Insert tile element loop as needed. Adjust.
35666 Remove out of date comments, fix whitespace.
35667 * omp-general.c (omp_extract_for_data): Deal with tiling.
35668 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
35669 adjust OLF_DIM_BASE value.
35670 (struct omp_for_data): Add tiling field.
35671 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
35672 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
35673 for auto loops. Remove default auto determining, moved to
35674 oacc_loop_fixed_partitions.
35675 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
35676 stmts, add e_mask field.
35677 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
35678 (oacc_thread_numbers): Use oacc_dim_call.
35679 (oacc_xform_tile): New.
35680 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
35681 (finish_oacc_loop): Adjust for ifns vector.
35682 (oacc_loop_discover_walk): Append loop abstraction sites to list,
35683 add case for GOACC_TILE fns.
35684 (oacc_loop_xform_loop): Delete.
35685 (oacc_loop_process): Iterate over call list directly, and add
35686 handling for GOACC_TILE fns.
35687 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
35689 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
35690 vector partitioning to outer loops. Assign 2 partitions to loops
35691 when available. Add TILE handling.
35692 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
35693 (execite_oacc_device_lower): Process GOACC_TILE fns,
35694 ignore unknown specs.
35695 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
35696 * tree.c (omp_clause_num_ops): Adjust TILE ops.
35697 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
35699 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
35701 * configure.ac (ACX_BUGURL): Update.
35702 * configure: Regenerate.
35704 2017-02-09 Richard Biener <rguenther@suse.de>
35706 PR tree-optimization/69823
35707 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
35708 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
35710 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35712 * config/arc/arc-c.def: Add __NPS400__ definition.
35713 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
35714 (TARGET_NPS400): Define.
35716 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35718 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
35720 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
35721 pointer, arch_info.
35722 (arc_cpu_types): Fill the arch_info field with a pointer into the
35723 arc_arch_types table.
35724 (arc_selected_cpu): Declare.
35725 * config/arc/arc.c (arc_selected_cpu): Make global.
35726 (arc_selected_arch): Delete.
35727 (arc_base_cpu): Delete.
35728 (arc_override_options): Remove references to deleted variables,
35729 update access to arch information.
35730 (ARC_OPT): Update access to arch information.
35731 (ARC_OPTX): Likewise.
35732 * config/arc/arc.h (arc_base_cpu): Remove declaration.
35733 (TARGET_ARC600): Update access to arch information.
35734 (TARGET_ARC601): Likewise.
35735 (TARGET_ARC700): Likewise.
35736 (TARGET_EM): Likewise.
35737 (TARGET_HS): Likewise.
35738 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
35741 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
35744 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
35745 condition/operands for integer GE/LE/GEU/LEU operations.
35747 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
35749 PR translation/79397
35750 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
35753 2017-02-08 Martin Jambor <mjambor@suse.cz>
35756 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
35757 whether allocation happened.
35758 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
35759 nothing was allocated.
35761 2017-02-08 Jakub Jelinek <jakub@redhat.com>
35763 PR tree-optimization/79408
35764 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
35765 constant, but SSA_NAME with a known integer range, use the minimum
35766 of that range instead of op1 to determine if modulo can be replaced
35767 with its first operand.
35769 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35771 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
35773 2017-02-08 Richard Biener <rguenther@suse.de>
35775 PR tree-optimization/71824
35776 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35777 Check all loops contained in the merged region.
35779 2017-02-07 Andrew Pinski <apinski@cavium.com>
35781 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
35783 2017-02-07 Andrew Pinski <apinski@cavium.com>
35785 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
35786 (thunderxt88): Likewise.
35787 (thunderxt81): Disable LSE and change v8.1 to v8.
35788 (thunderxt83): Likewise.
35790 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35791 Richard Biener <rguenther@suse.de>
35793 PR middle-end/79399
35794 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
35795 type from int to size_t.
35796 * ira-costs.c (struct_costs_size): Change type from int to size_t.
35798 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35800 PR rtl-optimization/79386
35801 * cprop.c (bypass_conditional_jumps): Initialize
35802 bypass_last_basic_block already before splitting bbs after
35803 unconditional traps...
35804 (bypass_conditional_jumps): ... rather than here.
35807 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
35808 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
35809 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
35810 fix -masm=intel patterns.
35812 2017-02-07 Richard Biener <rguenther@suse.de>
35814 PR tree-optimization/79256
35815 PR middle-end/79278
35816 * builtins.c (get_object_alignment_2): Use min_align_of_type
35817 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
35818 and ADJUST_FIELD_ALIGN.
35820 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
35822 * doc/tm.texi: Regenerate.
35823 * stor-layout.c (layout_decl): Adjust.
35824 (update_alignment_for_field): Likewise.
35825 (place_field): Likewise.
35826 (min_align_of_type): Likewise.
35827 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
35828 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
35829 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
35830 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
35831 * config/frv/frv.c (frv_adjust_field_align): Likewise.
35832 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
35833 * config/i386/i386.c (x86_field_alignment): Likewise.
35834 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
35835 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
35836 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
35837 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
35838 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
35839 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
35843 2017-01-30 Richard Biener <rguenther@suse.de>
35845 PR tree-optimization/79256
35846 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35847 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35850 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
35852 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
35853 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
35854 builtins to SImode and emit a zero-extend, if necessary.
35856 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35858 * docs/invoke.texi (RISC-V Options): Alphabetize.
35860 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35862 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
35865 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35867 * config/riscv/riscv.c: New file.
35868 * common/config/riscv/riscv-common.c: Likewise.
35869 * config.gcc: Likewise.
35870 * config/riscv/constraints.md: Likewise.
35871 * config/riscv/elf.h: Likewise.
35872 * config/riscv/generic.md: Likewise.
35873 * config/riscv/linux.h: Likewise.
35874 * config/riscv/multilib-generator: Likewise.
35875 * config/riscv/peephole.md: Likewise.
35876 * config/riscv/pic.md: Likewise.
35877 * config/riscv/predicates.md: Likewise.
35878 * config/riscv/riscv-builtins.c: Likewise.
35879 * config/riscv/riscv-c.c: Likewise.
35880 * config/riscv/riscv-ftypes.def: Likewise.
35881 * config/riscv/riscv-modes.def: Likewise.
35882 * config/riscv/riscv-opts.h: Likewise.
35883 * config/riscv/riscv-protos.h: Likewise.
35884 * config/riscv/riscv.h: Likewise.
35885 * config/riscv/riscv.md: Likewise.
35886 * config/riscv/riscv.opt: Likewise.
35887 * config/riscv/sync.md: Likewise.
35888 * config/riscv/t-elf-multilib: Likewise.
35889 * config/riscv/t-linux: Likewise.
35890 * config/riscv/t-linux-multilib: Likewise.
35891 * config/riscv/t-riscv: Likewise.
35892 * configure.ac: Likewise.
35893 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
35894 Waterman as RISC-V maintainers.
35895 * doc/install.texi: Add RISC-V entries.
35896 * doc/invoke.texi: Add RISC-V options section.
35897 * doc/md.texi: Add RISC-V constraints section.
35898 * configure: Regenerated.
35900 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
35903 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
35904 false values to be constant vectors with all 0 or all 1 bits set.
35905 (vcondu<mode><mode>): Likewise.
35906 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
35908 (fpmask_comparison_operator): Update comment.
35909 (vecint_comparison_operator): New predicate.
35910 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
35911 vector conditionals when the true and false values are constant
35912 vectors with all 0 bits or all 1 bits set.
35914 2017-02-06 Martin Sebor <msebor@redhat.com>
35916 PR tree-optimization/79376
35917 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
35919 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
35921 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
35922 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
35923 to simplify split condition.
35925 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35927 * omp-expand.c (oxpand_omp_atomic_fetch_op,
35928 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
35931 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
35933 PR rtl-optimization/68664
35934 * target.def (can_speculate_insn): New hook.
35935 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
35936 * doc/tm.texi: Regenerate.
35937 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
35938 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
35939 (rs6000_sched_can_speculate_insn): New function.
35941 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35943 PR tree-optimization/79284
35944 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
35945 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
35946 vectorizable_mask_load_store, vectorizable_operation,
35947 vect_is_simple_cond, get_same_sized_vectype): Use it instead
35948 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
35949 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
35950 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
35951 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
35952 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
35953 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
35954 is_gimple_assign (stmt). Replace another such test with
35955 is_gimple_assign (stmt).
35957 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35960 * config/avr/avr.c (rtl-iter.h): Include it.
35961 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
35962 (avr_legitimate_combined_insn): ...and implementation.
35964 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
35966 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
35967 * config/s390/s390.c (s390_const_operand_ok)
35968 (s390_canonicalize_comparison, s390_extract_part)
35969 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
35970 (s390_contiguous_bitmask_p, s390_rtx_costs)
35971 (legitimize_pic_address): Likewise.
35972 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
35973 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
35974 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
35975 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
35976 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
35978 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35980 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
35981 REGNO($0) == REGNO($1).
35983 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35985 * config/s390/linux.h(SIZE_TYPE): Add comment.
35987 2017-02-06 Julian Brown <julian@codesourcery.com>
35988 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35989 Virendra Pathak <virendra.pathak@broadcom.com>
35991 * config/aarch64/aarch64-cores.def: Change the scheduler
35993 * config/aarch64/aarch64.md: Include thunderx2t99.md.
35994 * config/aarch64/thunderx2t99.md: New file.
35996 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35998 * doc/standards.texi (Go Language): Update link to language
36001 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
36003 * tree-eh.c (lower_resx): Sanitize profile.
36004 (cleanup_empty_eh_move_lp): Likewise.
36006 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
36009 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
36011 * cfgloopmanip.h (loop_version): Update prototype.
36012 * modulo-sched.c (sms_schedule): Update call of loop_version.
36013 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
36014 * tree-parloops.c (gen_parallel_loop): Likewise.
36015 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
36016 * tree-ssa-loop-split.c (split_loop): Likewise.
36017 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
36018 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
36020 2017-02-05 Martin Liska <mliska@suse.cz>
36023 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
36025 (print_operand_address): Initialize a struct to zero.
36027 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36029 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
36030 garbage collector only in textual form.
36032 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36034 * doc/extend.texi (x86 specific memory model extensions for
36035 transactional memory): Simplify a phrase.
36037 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
36040 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
36041 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
36042 (atomic_storedi_1): Likewise.
36044 2017-02-04 Jakub Jelinek <jakub@redhat.com>
36046 PR tree-optimization/79338
36047 * tree-parloops.c (gather_scalar_reductions): Don't call
36048 vect_analyze_loop_form for loop->inner before destroying loop's
36051 2017-02-03 Martin Sebor <msebor@redhat.com>
36053 PR tree-optimization/79327
36054 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
36055 when precision has resulted in leading zeros.
36056 (format_integer): Adjust the likely counter to assume an unknown
36057 argument that may be zero is non-zero.
36059 2017-02-03 Jason Merrill <jason@redhat.com>
36062 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
36063 avoid copying non-taken branch.
36065 2017-02-03 Jakub Jelinek <jakub@redhat.com>
36067 PR tree-optimization/79340
36068 * tree-vect-loop.c (vectorizable_reduction): Release
36069 vec_defs elements after safe_splicing them into other vectors.
36072 PR tree-optimization/79327
36073 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
36074 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
36076 (format_integer): Use wide_int_to_tree instead of build_int_cst
36077 + to_?hwi. If argmin is NULL, just set argmin and argmax to
36078 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
36079 of shortest and longest sequence.
36081 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
36083 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
36084 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
36086 2017-02-03 Walter Lee <walt@tilera.com>
36089 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
36090 after initial stackframe link reg save.
36091 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
36093 2017-02-03 Jakub Jelinek <jakub@redhat.com>
36096 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
36097 wu for stxssp alternative.
36099 2017-02-03 Martin Sebor <msebor@redhat.com>
36101 PR tree-optimization/79352
36102 * gimple-fold.c (get_range_strlen): Add argument.
36103 (get_range_strlen): Change return type to bool.
36104 (get_maxval_strlen): Pass in a dummy argument.
36105 * gimple-fold.h (get_range_strlen): Change return type to bool.
36106 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
36107 * tree.h (array_at_struct_end_p): Add argument.
36108 * tree.c (array_at_struct_end_p): Handle it.
36110 2017-02-03 Martin Liska <mliska@suse.cz>
36113 * multiple_target.c (create_dispatcher_calls): Redirect edge
36114 from a caller of a dispatcher.
36115 (expand_target_clones): Make the clones local.
36116 (ipa_target_clone): Do both target clones and resolvers.
36117 (ipa_dispatcher_calls): Remove the pass.
36118 (pass_dispatcher_calls::gate): Likewise.
36119 (make_pass_dispatcher_calls): Likewise.
36120 * passes.def (pass_target_clone): Put as very first IPA early
36123 2017-02-03 Martin Liska <mliska@suse.cz>
36125 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
36126 in case of a function with ifunc attribute.
36128 2017-02-03 Martin Liska <mliska@suse.cz>
36130 * cgraph.c (cgraph_node::dump): Dump function version info.
36131 * symtab.c (symtab_node::dump_base): Add missing new line.
36133 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
36135 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
36136 (ifcombine_ifandif): Use it.
36138 2017-02-03 Martin Liska <mliska@suse.cz>
36140 * doc/invoke.texi: Document default value for
36141 use-after-scope-direct-emission-threshold.
36143 2017-02-03 Martin Liska <mliska@suse.cz>
36145 PR tree-optimization/79339
36146 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
36147 (format_floating): Likewise.
36149 2017-02-03 Martin Liska <mliska@suse.cz>
36152 * ipa-prop.c (ipa_node_params_t::insert): Remove current
36154 (ipa_node_params_t::remove): Likewise.
36155 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
36156 initialization from removed ipa_node_params_t::insert.
36157 (ipa_node_params::~ipa_node_params): Move from removed
36158 ipa_node_params_t::release.
36159 * symbol-summary.h (symbol_summary::m_released): New member.
36160 Do not release a summary twice. Do not allow to call finalizer
36161 for types of a summary that live in GGC memory.
36163 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
36165 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
36168 2017-02-02 Martin Sebor <msebor@redhat.com>
36170 PR middle-end/79275
36171 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
36172 (format_string): Tighten up the range of output for non-constant
36173 strings and correct the expected range for wide non-constant strings.
36175 2017-02-02 Martin Sebor <msebor@redhat.com>
36177 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
36179 PR middle-end/32003
36180 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
36182 (-fdump-tree-@var): Add to index and document how to come up
36183 with pass-specific option and dump file names.
36184 (-fdump-passes): Clarify where to look for output.
36186 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
36188 PR middle-end/77445
36189 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
36190 statistics of the analyzed path; allow threading for speed when
36191 any of BBs along the path are optimized for speed.
36193 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
36195 PR middle-end/78468
36196 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
36197 settings of the virtual registers.
36200 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
36202 * explow.c (get_dynamic_stack_size): Take known alignment of stack
36203 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
36206 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36208 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
36209 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
36211 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36213 * config/s390/s390.md: Add missing comments with the expanded
36215 * config/s390/vector.md: Likewise.
36216 * config/s390/vx-builtins.md: Likewise.
36218 2017-02-02 Jakub Jelinek <jakub@redhat.com>
36221 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
36222 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
36223 conditions on a single line.
36225 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36227 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
36228 __S390_VX__ to __VX__.
36230 2017-02-01 Andrew Pinski <apinski@cavium.com>
36232 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
36233 stmt_info to record_stmt_cost.
36234 (vect_get_known_peeling_cost): Pass stmt_info if known to
36236 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
36237 cpu_vector_cost field into
36238 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
36239 field into vec_int_stmt_cost and vec_fp_stmt_cost.
36240 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
36241 splitting of scalar_stmt_cost and vec_stmt_cost.
36242 (thunderx_vector_cost): Likewise.
36243 (cortexa57_vector_cost): LIkewise.
36244 (exynosm1_vector_cost): Likewise.
36245 (xgene1_vector_cost): Likewise.
36246 (thunderx2t99_vector_cost): Improve after the splitting of the two
36248 (aarch64_builtin_vectorization_cost): Update for the splitting of
36249 scalar_stmt_cost and vec_stmt_cost.
36251 2017-02-01 Torvald Riegel <triegel@redhat.com>
36252 Richard Henderson <rth@redhat.com>
36254 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
36255 conditional on existance of a fast atomic load.
36256 * optabs-query.c (can_atomic_load_p): New function.
36257 * optabs-query.h (can_atomic_load_p): Declare it.
36258 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
36259 no fast atomic load is available for the particular size of access.
36260 (expand_atomic_compare_and_swap): Likewise.
36261 (expand_atomic_load): Likewise.
36262 (expand_atomic_store): Likewise.
36263 (expand_atomic_fetch_op): Likewise.
36264 * testsuite/lib/target-supports.exp
36265 (check_effective_target_sync_int_128): Remove x86 because it provides
36266 no fast atomic load.
36267 (check_effective_target_sync_int_128_runtime): Likewise.
36269 2017-02-01 Richard Biener <rguenther@suse.de>
36271 * graphite.c: Include tree-vectorizer.h for find_loop_location.
36272 (graphite_transform_loops): Provide opt-info for optimized nests.
36273 * tree-parloop.c (parallelize_loops): Provide opt-info for
36274 parallelized loops.
36276 2017-02-01 Richard Biener <rguenther@suse.de>
36278 PR middle-end/79315
36279 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
36280 was not set before.
36282 2017-02-01 Richard Biener <rguenther@suse.de>
36284 PR tree-optimization/71824
36285 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
36286 Verify the loops are valid in the merged SESE region.
36287 (scop_detection::can_represent_loop_1): Check analyzing the
36288 evolution of the number of iterations in the region succeeds.
36290 2017-01-31 Ian Lance Taylor <iant@golang.org>
36292 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
36293 REG_ARGS_SIZE note to 32-bit push insns and call insn.
36295 2017-01-31 David Malcolm <dmalcolm@redhat.com>
36297 PR preprocessor/79210
36298 * input.c (get_substring_ranges_for_loc): Replace line_width
36299 assertion with error-handling.
36301 2017-01-31 Richard Biener <rguenther@suse.de>
36303 PR tree-optimization/77318
36304 * graphite-sese-to-poly.c (extract_affine): Fix assert.
36305 (create_pw_aff_from_tree): Take loop parameter.
36306 (add_condition_to_pbb): Pass loop of the condition to
36307 create_pw_aff_from_tree.
36309 2017-01-31 Jakub Jelinek <jakub@redhat.com>
36311 * config/s390/s390.c (s390_asan_shadow_offset): New function.
36312 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
36314 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
36318 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
36320 (convert_int_to_float128): Likewise.
36321 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
36322 (convert_int_to_float128): Likewise.
36323 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
36324 (UNSPEC_IEEE128_CONVERT): Likewise.
36325 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
36326 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
36327 Use local variables for IBM extended format.
36328 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
36329 (fix_trunc<mode>si2_fprs): Likewise.
36330 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
36331 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
36332 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
36333 to know that we can now have integers of all sizes in vector
36335 (fix<uns>_<mode>di2_hw): Likewise.
36336 (float<uns>_<mode>si2_hw): Likewise.
36337 (fix_<mode>si2_hw): Likewise.
36338 (fixuns_<mode>si2_hw): Likewise.
36339 (float<uns>_<mode>di2_hw): Likewise.
36340 (float_<mode>di2_hw): Likewise.
36341 (float_<mode>si2_hw): Likewise.
36342 (floatuns_<mode>di2_hw): Likewise.
36343 (floatuns_<mode>si2_hw): Likewise.
36344 (xscvqp<su>wz_<mode>): Delete, no longer used.
36345 (xscvqp<su>dz_<mode>): Likewise.
36346 (xscv<su>dqp_<mode>): Likewise.
36347 (ieee128_mfvsrd_64bit): Likewise.
36348 (ieee128_mfvsrd_32bit): Likewise.
36349 (ieee128_mfvsrwz): Likewise.
36350 (ieee128_mtvsrw): Likewise.
36351 (ieee128_mtvsrd_64bit): Likewise.
36352 (ieee128_mtvsrd_32bit): Likewise.
36354 2017-01-31 Martin Liska <mliska@suse.cz>
36357 * ipa-prop.c (ipa_free_all_node_params): Call release method
36358 instead of ~sumbol_summary to not to trigger double times
36361 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
36363 PR tree-optimization/71691
36364 * bitmap.h (class auto_bitmap): New.
36365 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
36366 is_maybe_undefined instead of ssa_undefined_value_p.
36368 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36370 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
36371 __S390_ARCH_LEVEL__ to __ARCH__.
36373 2017-01-31 Jakub Jelinek <jakub@redhat.com>
36375 PR tree-optimization/79267
36376 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
36377 if should_remove_lhs_p is true.
36379 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
36382 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
36383 (add_alignment_attribute): New.
36384 (base_type_die): Add alignment attribute.
36385 (subrange_type_die): Likewise.
36386 (modified_type_die): Likewise.
36387 (gen_array_type_die): Likewise.
36388 (gen_descr_array_type_die: Likewise.
36389 (gen_enumeration_type_die): Likewise.
36390 (gen_subprogram_die): Likewise.
36391 (gen_variable_die): Likewise.
36392 (gen_field_die): Likewise.
36393 (gen_ptr_to_mbr_type_die): Likewise.
36394 (gen_struct_or_union_type_die): Likewise.
36395 (gen_subroutine_type_die): Likewise.
36396 (gen_typedef_die): Likewise.
36397 (base_type_cmp): Compare alignment attribute.
36399 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36402 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
36403 (setb_unsigned) New pattern for setb with CCUNS.
36404 * config/rs6000/rs6000.c (expand_block_compare): Use a different
36405 subfc./subfe sequence to avoid overflow problems. Generate a
36406 shorter sequence with cmpld/setb for power9.
36407 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
36408 for generating subfc. instruction.
36409 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
36410 now uses this instruction.
36412 2017-01-30 Ian Lance Taylor <iant@google.com>
36415 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
36416 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
36418 2017-01-30 Martin Sebor <msebor@redhat.com>
36420 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
36421 Move constant to the right of a relational operator.
36422 (get_mpfr_format_length, format_character, format_string): Ditto.
36423 (should_warn_p, maybe_warn): Same.
36425 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
36427 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
36430 * asan.c (get_translation_unit_decl): Remove function.
36431 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
36433 2017-01-30 Martin Liska <mliska@suse.cz>
36435 PR gcov-profile/79259
36436 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
36437 -fprofile-generate.
36439 2017-01-30 Martin Liska <mliska@suse.cz>
36442 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
36443 Initialize variables with NULL value.
36445 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
36448 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
36450 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
36452 2017-01-30 Richard Biener <rguenther@suse.de>
36454 PR tree-optimization/79276
36455 * tree-vrp.c (process_assert_insertions): Properly adjust common
36456 when removing a duplicate.
36458 2017-01-30 Richard Biener <rguenther@suse.de>
36460 PR tree-optimization/79256
36461 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
36462 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
36464 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
36466 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
36469 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
36470 ("*r<noxa>sbg_<mode>_sll_bitmask")
36471 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
36472 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
36473 Use contiguous_bitmask_nowrap_operand.
36475 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36478 * config/rs6000/altivec.h (vec_xl): Revise #define.
36479 (vec_xst): Likewise.
36481 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
36483 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
36485 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
36487 PR rtl-optimization/79194
36488 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
36489 traps before call to bypass_conditional_jumps.
36491 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
36493 PR tree-optimization/71374
36494 * lra-constraints.c (check_conflict_input_operands): New.
36495 (match_reload): Use it.
36497 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
36500 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
36501 account to calculate conflict_set.
36503 2017-01-27 Bin Cheng <bin.cheng@arm.com>
36505 PR rtl-optimization/78559
36506 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
36507 other_insn in combine.
36509 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
36511 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
36512 uint16_type_node for BT_UINT16.
36514 2017-01-27 David Malcolm <dmalcolm@redhat.com>
36516 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
36517 "RTL Tests" to menu.
36518 (GIMPLE Tests): New node.
36519 (RTL Tests): New node.
36521 2017-01-27 Richard Biener <rguenther@suse.de>
36523 PR tree-optimization/79245
36524 * tree-loop-distribution.c (distribute_loop): Apply cost
36525 modeling also to detected patterns.
36527 2017-01-27 Richard Biener <rguenther@suse.de>
36529 PR tree-optimization/71433
36530 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
36531 (compare_assert_loc): New function.
36532 (process_assert_insertions): Sort and optimize assert locations
36533 to remove duplicates and push down identical assertions on
36534 edges to their destination block.
36536 2017-01-27 Richard Biener <rguenther@suse.de>
36538 PR tree-optimization/79244
36539 * tree-vrp.c (remove_range_assertions): Forcefully propagate
36540 out SSA names even if abnormal.
36542 2017-01-27 Jakub Jelinek <jakub@redhat.com>
36544 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
36545 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
36546 instead of MPFR_RNDN.
36548 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
36551 * arm.c (arm_option_override): Don't call build_target_option_node
36552 until after doing all option overrides.
36553 (arm_valid_target_attribute_tree): Likewise.
36555 2017-01-27 Martin Liska <mliska@suse.cz>
36557 * doc/invoke.texi (-fprofile-arcs): Document profiling support
36558 for {cd}tors and C++ {cd}tors.
36560 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
36562 * config/s390/s390.md ("*setmem_long_and")
36563 ("*setmem_long_and_31z"): Use zero_extend instead of and.
36565 2017-01-26 Martin Sebor <msebor@redhat.com>
36567 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
36570 2017-01-26 Martin Sebor <msebor@redhat.com>
36572 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
36573 HAVE_DFmode before using XFmode or DFmode.
36574 (parse_directive): Avoid using the z length modifier to avoid
36575 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
36577 PR middle-end/78703
36578 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
36579 to accept adjustment as an array.
36580 (get_int_range): New function.
36581 (struct directive): Make width and prec arrays.
36582 (directive::set_width, directive::set_precision): Call get_int_range.
36583 (format_integer, format_floating): Handle width and precision ranges.
36584 (format_string, parse_directive): Same.
36586 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36589 * dwarf2out.c (generate_skeleton_bottom_up): For children with
36590 comdat_type_p set, just clone them, but keep the children in the
36594 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
36595 which have direct callers with -fvar-tracking-assignments enabled
36597 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
36598 inside of type units.
36600 2017-01-26 Martin Sebor <msebor@redhat.com>
36602 PR middle-end/78703
36603 * gimple-ssa-sprintf.c (struct result_range): Add likely and
36605 (struct format_result): Replace number_chars, number_chars_min,
36606 and number_chars_max with a single member of struct result_range.
36608 (format_result::operator+=): Adjust.
36609 (struct fmtresult): Remove bounded. Handle likely and unlikely
36611 (fmtresult::adjust_for_width_or_precision): New function.
36612 (fmtresult:type_max_digits): New function.
36613 (bytes_remaining): Handle likely and unlikely counters.
36614 (min_bytes_remaining): Remove.
36615 (format_percent): Simplify.
36616 (format_integer, format_floating): Set likely and unlikely counters.
36617 (get_string_length, format_character, format_string): Same.
36618 (format_plain, should_warn_p): New function.
36619 (maybe_warn): Call should_warn_p. Update diagnostic messages
36620 and handle those for all directives, including plain strings.
36621 (format_directive): Handle likely and unlikely counters.
36622 Remove unnecessary quoting from diagnostics. Add an informational
36624 (add_bytes): Remove.
36625 (pass_sprintf_length::compute_format_length): Simplify.
36626 (try_substitute_return_value): Handle likely and unlikely counters.
36628 2017-01-26 Carl Love <cel@us.ibm.com>
36630 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
36631 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
36633 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
36636 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
36637 endianess for subregs into account.
36638 * lra-constraints.c (lra_constraints): Do risky transformations
36639 always on the first iteration.
36640 * lra-lives.c (check_pseudos_live_through_calls): Add arg
36641 last_call_used_reg_set.
36642 (process_bb_lives): Define and use last_call_used_reg_set.
36643 * lra.c (lra): Always continue after lra_constraints on the first
36646 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
36648 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
36650 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
36652 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36654 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
36655 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
36656 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
36657 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
36658 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
36659 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
36660 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
36661 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
36662 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
36664 2017-01-26 Marek Polacek <polacek@redhat.com>
36667 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
36668 for the third operand.
36670 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36672 PR middle-end/79236
36673 * omp-low.c (struct omp_context): Add simt_stmt field.
36674 (scan_omp_for): Return omp_context *.
36675 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
36676 context to the _simt_ SIMD stmt.
36677 (lower_omp_for): For combined SIMD with sibling _simt_
36678 SIMD, make sure to use the same decls in _looptemp_
36679 clauses as in the sibling.
36681 2017-01-26 David Sherwood <david.sherwood@arm.com>
36683 PR middle-end/79212
36684 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
36687 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36690 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
36691 emit fld b; fld a; if possible.
36693 * brig-builtins.def: Update copyright years.
36694 * config/arm/arm_acle_builtins.def: Update copyright years.
36696 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
36699 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
36700 constraint instead of o for the stxsd instruction.
36702 2017-01-25 Carl Love <cel@us.ibm.com>
36704 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
36705 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
36707 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
36709 * doc/invoke.texi (C++ Dialect Options): Fix typo.
36711 2017-01-25 Richard Biener <rguenther@suse.de>
36713 PR tree-optimization/69264
36714 * target.def (vector_alignment_reachable): Improve documentation.
36715 * doc/tm.texi: Regenerate.
36716 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
36718 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
36719 earlier changes with respect to TYPE_USER_ALIGN.
36720 (vector_alignment_reachable_p): Likewise. Improve dumping.
36722 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36725 * config/arm/arm.md (xordi3): Force constant operand into a register
36728 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36730 * doc/invoke.texi (-fstore-merging): Correct default optimization
36731 levels at which it is enabled.
36732 (-O): Move -fstore-merging from list to...
36735 2017-01-25 Richard Biener <rguenther@suse.de>
36738 * omp-expand.c: Include debug.h.
36739 (expand_omp_taskreg): Make sure to generate early debug before
36740 outlining anything from a function.
36741 (expand_omp_target): Likewise.
36742 (grid_expand_target_grid_body): Likewise.
36744 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
36747 * asan.c (get_translation_unit_decl): New function.
36748 (asan_add_global): Extract modules file name from globals
36749 TRANSLATION_UNIT_DECL name.
36751 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
36754 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
36755 for long calls with APCS frame and VFP.
36757 2017-01-24 David Malcolm <dmalcolm@redhat.com>
36759 * cfg.c (original_copy_tables_initialized_p): New function.
36760 * cfg.h (original_copy_tables_initialized_p): New decl.
36761 * cfgrtl.c (relink_block_chain): Guard the call to
36762 free_original_copy_tables with a call to
36763 original_copy_tables_initialized_p.
36764 * cgraph.h (symtab_node::native_rtl_p): New decl.
36765 * cgraphunit.c (symtab_node::native_rtl_p): New function.
36766 (symtab_node::needed_p): Don't assert for early assembly output
36767 for __RTL functions.
36768 (cgraph_node::finalize_function): Set "force_output" for __RTL
36770 (cgraph_node::analyze): Bail out early for __RTL functions.
36771 (analyze_functions): Update assertion to support __RTL functions.
36772 (cgraph_node::expand): Bail out early for __RTL functions.
36773 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
36775 * function.h (struct function): Update comment for field
36777 * gimple-expr.c: Include "tree-pass.h".
36778 (gimple_has_body_p): Return false for __RTL functions.
36779 * Makefile.in (OBJS): Add run-rtl-passes.o.
36780 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
36782 (gcc::pass_manager::get_clean_slate): New accessor.
36783 * passes.c: Include "insn-addr.h".
36784 (should_skip_pass_p): Add logging. Update logic for running
36785 "expand" to be compatible with both __GIMPLE and __RTL. Guard
36786 property-provider override so it is only done for gimple passes.
36788 (skip_pass): New function.
36789 (execute_one_pass): Call skip_pass when skipping passes.
36790 * read-md.c (md_reader::read_char): Support filtering
36791 the input to a subset of line numbers.
36792 (md_reader::md_reader): Initialize fields
36793 m_first_line and m_last_line.
36794 (md_reader::read_file_fragment): New function.
36795 * read-md.h (md_reader::read_file_fragment): New decl.
36796 (md_reader::m_first_line): New field.
36797 (md_reader::m_last_line): New field.
36798 * read-rtl-function.c (function_reader::create_function): Only
36799 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
36800 curr_properties. Set DECL_INITIAL to a dummy block.
36801 (read_rtl_function_body_from_file_range): New function.
36802 * read-rtl-function.h (read_rtl_function_body_from_file_range):
36804 * run-rtl-passes.c: New file.
36805 * run-rtl-passes.h: New file.
36807 2017-01-24 Jeff Law <law@redhat.com>
36809 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
36812 2017-01-24 Bin Cheng <bin.cheng@arm.com>
36814 PR tree-optimization/79159
36815 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
36816 (record_nonwrapping_iv): Improve boundary using above function if no
36817 value range information.
36819 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
36820 Martin Jambor <mjambor@suse.cz>
36822 * brig-builtins.def: New file.
36823 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
36824 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
36825 (DEF_HSAIL_SAT_BUILTIN): Likewise.
36826 (DEF_HSAIL_INTR_BUILTIN): Likewise.
36827 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
36828 * builtin-types.def (BT_INT8): New.
36829 (BT_INT16): Likewise.
36830 (BT_UINT8): Likewise.
36831 (BT_UINT16): Likewise.
36832 (BT_FN_ULONG): Likewise.
36833 (BT_FN_UINT_INT): Likewise.
36834 (BT_FN_UINT_ULONG): Likewise.
36835 (BT_FN_UINT_LONG): Likewise.
36836 (BT_FN_UINT_PTR): Likewise.
36837 (BT_FN_ULONG_PTR): Likewise.
36838 (BT_FN_INT8_FLOAT): Likewise.
36839 (BT_FN_INT16_FLOAT): Likewise.
36840 (BT_FN_UINT32_FLOAT): Likewise.
36841 (BT_FN_UINT16_FLOAT): Likewise.
36842 (BT_FN_UINT8_FLOAT): Likewise.
36843 (BT_FN_UINT64_FLOAT): Likewise.
36844 (BT_FN_UINT16_UINT32): Likewise.
36845 (BT_FN_UINT32_UINT16): Likewise.
36846 (BT_FN_UINT16_UINT16_UINT16): Likewise.
36847 (BT_FN_INT_PTR_INT): Likewise.
36848 (BT_FN_UINT_PTR_UINT): Likewise.
36849 (BT_FN_LONG_PTR_LONG): Likewise.
36850 (BT_FN_ULONG_PTR_ULONG): Likewise.
36851 (BT_FN_VOID_UINT64_UINT64): Likewise.
36852 (BT_FN_UINT8_UINT8_UINT8): Likewise.
36853 (BT_FN_INT8_INT8_INT8): Likewise.
36854 (BT_FN_INT16_INT16_INT16): Likewise.
36855 (BT_FN_INT_INT_INT): Likewise.
36856 (BT_FN_UINT_FLOAT_UINT): Likewise.
36857 (BT_FN_FLOAT_UINT_UINT): Likewise.
36858 (BT_FN_ULONG_UINT_UINT): Likewise.
36859 (BT_FN_ULONG_UINT_PTR): Likewise.
36860 (BT_FN_ULONG_ULONG_ULONG): Likewise.
36861 (BT_FN_UINT_UINT_UINT): Likewise.
36862 (BT_FN_VOID_UINT_PTR): Likewise.
36863 (BT_FN_UINT_UINT_PTR: Likewise.
36864 (BT_FN_UINT32_UINT64_PTR): Likewise.
36865 (BT_FN_INT_INT_UINT_UINT): Likewise.
36866 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
36867 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
36868 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
36869 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
36870 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
36871 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
36872 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
36873 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
36874 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
36875 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
36876 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
36877 * doc/frontends.texi: List BRIG FE.
36878 * doc/install.texi (Testing): Add BRIG tesring requirements.
36879 * doc/invoke.texi (Overall Options): Mention BRIG.
36880 * doc/standards.texi (Standards): Doucment BRIG HSA version.
36882 2017-01-24 Richard Biener <rguenther@suse.de>
36884 PR translation/79208
36885 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
36887 2017-01-24 Martin Jambor <mjambor@suse.cz>
36890 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
36891 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
36892 and known_contexts.
36894 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
36896 PR middle-end/79123
36897 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
36898 casts from signed to unsigned really don't have a range.
36900 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
36902 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
36903 GMP_RNDx for compatiblity.
36905 2017-01-24 Martin Liska <mliska@suse.cz>
36908 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
36909 that would prevent us to call alloca with -1 as argument.
36911 2017-01-24 Jakub Jelinek <jakub@redhat.com>
36913 * dwarf2out.c (output_compilation_unit_header, output_file_names):
36914 Avoid -Wformat-security warning.
36916 2017-01-23 Andrew Pinski <apinski@cavium.com>
36918 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
36921 2017-01-23 Martin Sebor <msebor@redhat.com>
36923 PR middle-end/78703
36924 * gimple-ssa-sprintf.c (warn_level): New global.
36925 (format_integer): Use it here and throughout the rest of the file.
36926 Use the same switch to compute sign as base.
36927 (maybe_warn): New function.
36928 (format_directive): Factor out warnings into maybe_warn.
36929 Add debugging output. Use warn_level.
36930 (add_bytes): Use warn_level.
36931 (pass_sprintf_length::compute_format_length): Add debugging output.
36932 (try_substitute_return_value): Same.
36933 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
36935 PR middle-end/78703
36936 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
36937 (struct fmtresult, format_integer, format_floating): Adjust.
36938 (fmtresult::fmtresult): Set max correctly in two argument ctor.
36939 (get_string_length, format_string,format_directive): Same.
36940 (pass_sprintf_length::compute_format_length): Same.
36941 (try_substitute_return_value): Simplify slightly.
36943 PR middle-end/78703
36944 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
36945 (fmtresult::operator+=): Outlined.
36946 (struct fmtresult): Add ctors.
36947 (struct conversion_spec): Rename...
36948 (struct directive): ...to this. Add and remove data members.
36949 (directive::set_width, directive::set_precision): New functions.
36950 (format_percent): Use fmtresult ctor.
36951 (get_width_and_precision): Remove.
36952 (format_integer): Make naming changes. Avoid computing width and
36954 (format_floating): Same. Adjust indentation.
36955 (format_character, format_none): New functions.
36956 (format_string): Moved character handling to format_character.
36957 (format_directive): Remove arguments, change return type.
36958 (parse_directive): New function.
36959 (pass_sprintf_length::compute_format_length): Move directive
36960 parsing to parse_directive.
36962 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36964 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
36965 (assign_assembler_name_if_needed): ... this.
36966 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
36967 (assign_assembler_name_if_needed): ... this.
36968 (free_lang_data_in_cgraph): Adjust callers.
36969 * cgraphunit.c (cgraph_node::analyze): Likewise.
36970 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
36973 2017-01-23 Richard Biener <rguenther@suse.de>
36975 PR tree-optimization/79088
36976 PR tree-optimization/79188
36977 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
36978 resetting loop bounds after last path deletion. Reset loop
36979 bounds of the target loop, make code match the comments.
36980 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
36981 Make sure loops need no fixups.
36983 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
36985 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
36986 exponent support with double type for first argument.
36987 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
36988 type returned by __builtin_vec_extract_sig,
36989 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
36990 functions from "vector int" to "vector unsigned int" or from
36991 "vector long long int" to "vector unsigned long long int".
36992 Changed type returned by __builtin_vec_extract_exp,
36993 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
36994 functions from "vector int" to "vector unsigned int" or from
36995 "vector long long int" to "vector unsigned long long int".
36996 Changed return type of __builtin_vec_test_data_class,
36997 __builtin_vec_test_data_class_sp, and
36998 __builtin_vec_test_data_class_dp from "vector int" to
36999 "vector bool int" or from "vector long long int" to "vector bool
37000 long long int" and changed second argument type from "unsigned
37001 int" to "int". Added new overloaded function forms "vector float
37002 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
37003 "vector float __builtin_vec_insert_exp_sp (vector float, vector
37004 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
37005 double, vector unsigned long long int)" and "vector double
37006 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
37007 long int)". Changed return type of
37008 __builtin_scalar_test_data_class and
37009 __builtin_scalar_test_data_class_sp and
37010 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
37011 int" and changed second argument from "unsigned int" to "int".
37012 Changed type returned by __builtin_scalar_test_neg,
37013 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
37014 from "int" to "bool int". Added new overloaded function form
37015 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
37016 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
37017 exponent double-precision with floating point first argument.
37018 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
37019 documentation of scalar_test_data_class, scalar_test_neg,
37020 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
37021 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
37022 vec_test_data_class built-in functions to reflect refinements in
37023 their type signatures.
37025 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
37027 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
37029 (aarch64_elf_asm_destructor): Likewise.
37031 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
37033 PR rtl-optimization/78634
37034 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
37035 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
37036 * ifcvt.c (noce_try_cmove): Add missing cost check.
37038 PR rtl-optimization/71724
37039 * combine.c (if_then_else_cond): Look for situations where it is
37040 beneficial to undo the work of one of the recursive calls.
37042 2017-01-23 Bin Cheng <bin.cheng@arm.com>
37044 PR tree-optimization/70754
37045 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
37046 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
37047 combined stmt before it if not NULL.
37048 (combine_chains): Process refs reversely and compute dominance point
37051 2017-01-23 Martin Liska <mliska@suse.cz>
37053 PR tree-optimization/79196
37054 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
37055 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
37057 (strlen_optimize_stmt): Call the renamed function.
37059 2017-01-23 Michael Matz <matz@suse.de>
37061 PR tree-optimization/78384
37062 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
37064 2017-01-23 Richard Biener <rguenther@suse.de>
37066 PR tree-optimization/79186
37067 * tree-vrp.c (register_new_assert_for): Make sure we've seen
37068 both incoming edges before moving an assert.
37070 2017-01-23 Martin Jambor <mjambor@suse.cz>
37072 * ipa-prop.c (load_from_param_1): Removed.
37073 (load_from_unmodified_param): Bits from load_from_param_1 put back
37075 (load_from_param): Removed.
37076 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
37077 with stmt. Reverted back to use of load_from_unmodified_param.
37079 2017-01-23 Martin Jambor <mjambor@suse.cz>
37082 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
37083 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
37084 field a pointer to garbage collected vector, mark lattices and
37085 ipcp_orig_node with GTY((skip)).
37086 (ipa_get_param_count): Adjust to descriptors being a pointer.
37087 (ipa_get_param): Likewise.
37088 (ipa_get_type): Likewise.
37089 (ipa_get_param_move_cost): Likewise.
37090 (ipa_set_param_used): Likewise.
37091 (ipa_get_controlled_uses): Likewise.
37092 (ipa_set_controlled_uses): Likewise.
37093 (ipa_is_param_used): Likewise.
37094 (ipa_node_params_t): Move into garbage collector. New methods insert
37096 (ipa_node_params_sum): Annotate wth GTY(()).
37097 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
37099 (ipa_load_from_parm_agg): Adjust declaration.
37100 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
37101 * ipa-profile.c (ipa_profile): Likewise.
37102 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
37103 (ipa_populate_param_decls): Make descriptors parameter garbage
37105 (ipa_dump_param): Adjust to descriptors being a pointer.
37106 (ipa_alloc_node_params): Likewise.
37107 (ipa_initialize_node_params): Likewise.
37108 (load_from_param_1): Make descriptors parameter garbage collected.
37109 (load_from_unmodified_param): Likewise.
37110 (load_from_param): Likewise.
37111 (ipa_load_from_parm_agg): Likewise.
37112 (ipa_node_params::~ipa_node_params): Removed.
37113 (ipa_free_all_node_params): Remove call to delete operator.
37114 (ipa_node_params_t::insert): New.
37115 (ipa_node_params_t::remove): Likewise.
37116 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
37117 copy known_csts and known_contexts vectors.
37118 (ipa_read_node_info): Adjust to descriptors being a pointer.
37119 (ipcp_modif_dom_walker): Make m_descriptors field garbage
37121 (ipcp_transform_function): Make descriptors variable garbage
37124 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
37126 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
37127 * config/i386/avx512dqintrin.h: Ditto.
37128 * config/i386/avx512fintrin.h: Ditto.
37129 * config/i386/i386.c: Handle new builtins.
37130 * config/i386/i386-builtin.def: Add new builtins.
37131 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
37132 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
37134 2017-01-23 Jakub Jelinek <jakub@redhat.com>
37135 Martin Liska <mliska@suse.cz>
37137 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
37138 * asan.c (asan_expand_poison_ifn): Support stores and use
37139 appropriate ASAN report function.
37140 * internal-fn.c (expand_ASAN_POISON_USE): New function.
37141 * internal-fn.def (ASAN_POISON_USE): Declare.
37142 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
37143 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
37144 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
37145 ASAN_POISON calls w/o LHS.
37146 * tree-ssa.c (execute_update_addresses_taken): Create clobber
37147 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
37148 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
37149 * gimplify.c (asan_poison_variables): Add attribute
37150 use_after_scope_memory to variables that really needs to live
37152 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
37153 having the attribute.
37155 2017-01-23 Martin Liska <mliska@suse.cz>
37157 * asan.c (create_asan_shadow_var): New function.
37158 (asan_expand_poison_ifn): Likewise.
37159 * asan.h (asan_expand_poison_ifn): New declaration.
37160 * internal-fn.c (expand_ASAN_POISON): Likewise.
37161 * internal-fn.def (ASAN_POISON): New builtin.
37162 * sanopt.c (pass_sanopt::execute): Expand
37163 asan_expand_poison_ifn.
37164 * tree-inline.c (copy_decl_for_dup_finish): Make function
37166 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
37167 * tree-ssa.c (is_asan_mark_p): New function.
37168 (execute_update_addresses_taken): Rewrite local variables
37169 (identified just by use-after-scope as addressable) into SSA.
37171 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
37173 * doc/install.texi (Specific): opensource.apple.com uses https
37174 now. Remove trailing slash.
37176 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
37178 * README.Portability: Remove note on an Irix compatibility issue.
37180 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
37182 * gcov.c (INCLUDE_ALGORITHM): Define.
37183 (INCLUDE_VECTOR): Define.
37184 No longer include <vector> and <algorithm> directly.
37186 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
37188 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
37190 * doc/invoke.texi (Code Gen Options): Ditto.
37192 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
37195 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
37197 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
37199 rtl-optimization/79125
37200 * cprop.c (local_cprop_pass): Handle cases where we make an
37201 unconditional trap.
37203 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
37207 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
37208 read from, for big endian.
37210 2017-01-20 Jiong Wang <jiong.wang@arm.com>
37212 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
37213 register pauth builtins for LP64 only.
37215 2017-01-20 Marek Polacek <polacek@redhat.com>
37218 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
37221 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
37223 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
37225 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
37226 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
37227 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
37229 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37232 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
37233 in big-endian mode when they are not a single duplicated value.
37235 2017-01-20 Richard Biener <rguenther@suse.de>
37237 * BASE-VER: Bump to 7.0.1.
37239 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
37241 * omp-low.c (omplow_simd_context): New struct. Use it...
37242 (lower_rec_simd_input_clauses): ...here and...
37243 (lower_rec_input_clauses): ...here to hold common data. Adjust all
37244 references to idx, lane, max_vf, is_simt.
37246 2017-01-20 Graham Markall <graham.markall@embecosm.com>
37248 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
37251 2017-01-20 Martin Jambor <mjambor@suse.cz>
37253 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
37254 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
37256 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
37257 (GTFILES): Rename hsa.c to hsa-common.c.
37258 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
37259 * hsa-dump.c: Likewise.
37260 * hsa-gen.c: Likewise.
37261 * hsa-regalloc.c: Likewise.
37262 * ipa-hsa.c: Likewise.
37263 * omp-expand.c: Likewise.
37264 * omp-low.c: Likewise.
37265 * toplev.c: Likewise.
37267 2017-01-20 Marek Polacek <polacek@redhat.com>
37270 * doc/invoke.texi: Document -Wduplicated-branches.
37271 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
37272 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
37273 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
37274 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
37275 return 0 only when not OEP_LEXICOGRAPHIC.
37276 (fold_build_cleanup_point_expr): Use the expression
37277 location when building CLEANUP_POINT_EXPR.
37278 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
37279 * tree.c (add_expr): Handle error_mark_node.
37281 2017-01-20 Martin Liska <mliska@suse.cz>
37284 * tree-profile.c (init_ic_make_global_vars): Do not call
37286 (gimple_init_gcov_profiler): Likewise.
37288 2017-01-20 Martin Liska <mliska@suse.cz>
37291 * cgraph.h (maybe_create_reference): Remove argument and
37293 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
37295 * ipa-cp.c (create_specialized_node): Likewise.
37296 * symtab.c (symtab_node::maybe_create_reference): Handle
37297 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
37299 2017-01-20 Martin Liska <mliska@suse.cz>
37301 * read-rtl-function.c (function_reader::create_function): Use
37302 build_decl instread of build_decl_stat.
37304 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
37306 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
37307 * config/i386/avx512dqintrin.h: Ditto.
37308 * config/i386/avx512fintrin.h: Ditto.
37309 * config/i386/i386-builtin-types.def: Add new types.
37310 * config/i386/i386.c: Handle new types.
37311 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
37312 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
37313 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
37314 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
37315 (__builtin_ia32_kshiftridi): New.
37316 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
37318 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
37322 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
37323 define to rs6000_init_stack_protect_guard.
37324 (rs6000_init_stack_protect_guard): New function.
37326 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
37327 Yunqiang Su <yunqiang.su@imgtec.com>
37329 * config.gcc (supported_defaults): Add madd4.
37330 (with_madd4): Add validation.
37331 (all_defaults): Add madd4.
37332 * config/mips/mips.opt (mmadd4): New option.
37333 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
37335 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
37337 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
37338 (ISA_HAS_FUSED_MADD4): Likewise.
37339 * doc/invoke.texi (-mmadd4): Document the new option.
37340 * doc/install.texi (--with-madd4): Document the new option.
37342 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37344 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
37345 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
37346 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
37347 (aarch64_init_pauth_hint_builtins): New.
37348 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
37349 (aarch64_expand_builtin): Expand new builtins.
37351 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37353 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
37354 * combine-stack-adj.c (no_unhandled_cfa): Handle
37355 REG_CFA_TOGGLE_RA_MANGLE.
37356 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
37357 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
37358 info for return address signing.
37359 (aarch64_expand_epilogue): Likewise.
37361 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37363 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
37364 * config/aarch64/aarch64-protos.h
37365 (aarch64_return_address_signing_enabled): New declaration.
37366 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
37368 (aarch64_expand_prologue): Sign return address before it's pushed onto
37370 (aarch64_expand_epilogue): Authenticate return address fetched from
37372 (aarch64_override_options): Sanity check for ILP32 and ISA level.
37373 (aarch64_attributes): New function attributes for "sign-return-address".
37374 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
37375 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
37376 ("*do_return"): Generate combined instructions according to key index.
37377 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
37378 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
37380 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
37381 * config/aarch64/aarch64.opt (msign-return-address=): New.
37382 * doc/extend.texi (AArch64 Function Attributes): Documents
37383 "sign-return-address=".
37384 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
37386 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
37388 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
37389 overall option summary.
37391 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37393 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
37394 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
37395 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
37396 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
37398 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
37400 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
37401 -mpower9-minmax by default for -mcpu=power9.
37402 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
37403 128-bit floating point.
37405 2017-01-20 Alan Modra <amodra@gmail.com>
37407 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
37408 optimizing for size.
37410 2017-01-20 Alan Modra <amodra@gmail.com>
37413 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
37414 for strcmp and strncmp from corresponding builtin decl.
37416 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
37418 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
37419 instead of i386/rtems-64.h.
37420 * config/i386/rtems-64.h: Remove.
37422 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
37426 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
37428 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
37430 2017-01-19 Tamar Christina <tamar.christina@arm.com>
37432 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
37433 Change int to HOST_WIDE_INT.
37434 * config/aarch64/aarch64-protos.h
37435 (aarch64_simd_gen_const_vector_dup): Likewise.
37436 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
37438 2017-01-19 David Malcolm <dmalcolm@redhat.com>
37440 * langhooks-def.h (lhd_type_for_size): New decl.
37441 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
37442 * langhooks.c (lhd_type_for_size): New function, taken from
37445 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
37447 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
37448 define_bypass for CR latency.
37449 (power9-cracked-alu): Update bypass latency and remove power9-branch.
37450 (power9-alu2): Add define_bypass for CR latency.
37452 (power9-mul): Update insn latency.
37453 (power9-mul-compare): Update insn latency, bypass latency and remove
37456 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37458 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
37460 * config/aarch64/aarch64.md
37461 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
37462 aarch64_nopcrelative_literal_loads.
37463 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
37465 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
37467 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
37468 TARGET_LOONGSON_3A.
37469 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
37471 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
37474 * config.gcc (supported_defaults): Add lxc1-sxc1.
37475 (with_lxc1_sxc1): Add validation.
37476 (all_defaults): Add lxc1-sxc1.
37477 * config/mips/mips.opt (mlxc1-sxc1): New option.
37478 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
37480 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
37481 __mips_no_lxc1_sxc1.
37482 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
37483 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
37484 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
37486 2017-01-19 Richard Biener <rguenther@suse.de>
37488 PR tree-optimization/72488
37489 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
37490 sure to restore SSA info.
37491 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
37493 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
37495 PR rtl-optimization/79121
37496 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
37497 of the inner type when shifting an extended value.
37499 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
37502 * symtab.c (symtab_node::equal_address_to): Fix comparing of
37503 interposable aliases.
37505 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
37508 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
37509 Use the evmergelohi instruction.
37510 (mov_si<mode>_e500_subreg4_2_le): Likewise.
37511 (mov_sitf_e500_subreg8_2_be): Likewise.
37512 (mov_sitf_e500_subreg12_2_le): Likewise.
37513 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
37514 (mov_si<mode>_e500_subreg4_2_be): Likewise.
37515 (mov_sitf_e500_subreg8_2_le): Likewise.
37516 (mov_sitf_e500_subreg12_2_be): Likewise.
37518 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37520 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
37521 attribute from vecsimple to vecperm.
37522 (altivec_vbpermq2): Likewise.
37524 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37527 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
37529 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37530 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
37531 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
37532 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
37533 case where N arg is SIZE_MAX.
37534 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
37535 (cmpstrsi): Add pattern.
37537 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
37539 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
37540 __builtin_vec_revb builtins.
37541 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
37542 built-in functions to support generation of the ISA 3.0 XXBR<x>
37543 vector byte reverse instructions.
37544 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
37545 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
37546 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
37547 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
37548 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
37549 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
37550 (P9V_BUILTIN_VEC_REVB): Likewise.
37551 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
37552 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
37553 (p9_xxbrq_v16qi): Likewise.
37554 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
37555 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
37556 (p9_xxbrh_v8hi): Likewise.
37557 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
37558 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
37559 vec_revb built-in functions.
37561 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
37563 PR rtl-optimization/78952
37564 * config/i386/i386.md (any_extract): New code iterator.
37565 (*insvqi_2): Use any_extract for source operand.
37566 (*insvqi_3): Use any_shiftrt for source operand.
37568 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
37570 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
37572 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
37574 2017-01-18 Matthias Klose <doko@ubuntu.com>
37576 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
37578 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37580 * config/rs6000/altivec.h (vec_bperm): Change #define.
37581 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
37582 (altivec_vbpermq2): New define_insn.
37583 (altivec_vbpermd): Likewise.
37584 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
37585 function interface.
37586 (VBPERMD): Likewise.
37587 (VBPERM): New polymorphic function interface.
37588 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
37589 Add entries for P9V_BUILTIN_VEC_VBPERM.
37590 * doc/extend.texi: Add interfaces for vec_bperm.
37592 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37594 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
37595 first letter of error messages.
37596 (s390_resolve_overloaded_builtin): Likewise.
37597 * config/s390/s390.c (s390_expand_builtin): Likewise.
37598 (s390_invalid_arg_for_unprototyped_fn): Likewise.
37599 (s390_valid_target_attribute_inner_p): Likewise.
37600 * config/s390/s390.md ("tabort"): Likewise.
37602 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
37604 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
37605 (ISA_AVOID_DIV_HILO): New macro.
37606 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
37607 (ISA_HAS_DDIV): Likewise.
37609 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37611 * doc/invoke.texi (fabi-version): Correct number of occurrences.
37613 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37615 * doc/invoke.texi (fabi-version): Spelling fix.
37617 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37620 * doc/invoke.texi (fabi-version): Mention mangling fix for
37623 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37626 * doc/invoke.texi (fabi-version): Document discriminator mangling.
37628 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
37631 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
37632 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
37634 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
37636 (stack_protect_test): Ditto.
37637 * config/rs6000/rs6000.opt (mstack-protector-guard=,
37638 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
37640 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
37641 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
37642 -mstack-protector-guard-offset=.
37643 (RS/6000 and PowerPC Options): Ditto.
37645 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37647 * config/i386/i386.h (MASK_CLASS_P): New define.
37648 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
37649 there are no registers from different register sets also when
37650 mask registers are used. Update function comment.
37651 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
37652 to (*k/*r) and (*k/*km) alternatives.
37654 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
37656 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
37657 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
37658 (EH_RETURN_HANDLER_RTX): New define.
37659 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
37660 Force frame pointer in EH return functions.
37661 (aarch64_expand_epilogue): Add barrier for eh_return.
37662 (aarch64_final_eh_return_addr): Remove.
37663 (aarch64_eh_return_handler_rtx): New function.
37664 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
37666 (aarch64_eh_return_handler_rtx): New prototype.
37668 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37670 * config/rs6000/altivec.h (vec_rlmi): New #define.
37671 (vec_vrlnm): Likewise.
37672 (vec_rlnm): Likewise.
37673 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
37674 (UNSPEC_VRLNM): Likewise.
37675 (VIlong): New mode iterator.
37676 (altivec_vrl<VI_char>mi): New define_insn.
37677 (altivec_vrl<VI_char>nm): Likewise.
37678 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
37680 (VRLDNM): Likewise.
37681 (RLNM): New polymorphic function entry.
37682 (VRLWMI): New monomorphic function entry.
37683 (VRLDMI): Likewise.
37684 (RLMI): New polymorphic function entry.
37685 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
37686 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
37687 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
37690 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37693 * dwarf2out.c (field_byte_offset): Restore the
37694 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
37695 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
37696 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
37699 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37702 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
37704 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37707 * dwarf2out.c (add_data_member_location_attribute): For constant
37708 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
37709 instead of DW_AT_data_member_location, DW_AT_bit_offset and
37710 DW_AT_byte_size attributes.
37712 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37714 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
37715 after forcing to constant memory when the code model is medium.
37717 2017-01-17 Julia Koval <julia.koval@intel.com>
37720 * config/i386/avx512fintrin.h
37721 (_mm512_i32gather_ps): Change __addr type to void const*.
37722 (_mm512_mask_i32gather_ps): Ditto.
37723 (_mm512_i32gather_pd): Ditto.
37724 (_mm512_mask_i32gather_pd): Ditto.
37725 (_mm512_i64gather_ps): Ditto.
37726 (_mm512_mask_i64gather_ps): Ditto.
37727 (_mm512_i64gather_pd): Ditto.
37728 (_mm512_mask_i64gather_pd): Ditto.
37729 (_mm512_i32gather_epi32): Ditto.
37730 (_mm512_mask_i32gather_epi32): Ditto.
37731 (_mm512_i32gather_epi64): Ditto.
37732 (_mm512_mask_i32gather_epi64): Ditto.
37733 (_mm512_i64gather_epi32): Ditto.
37734 (_mm512_mask_i64gather_epi32): Ditto.
37735 (_mm512_i64gather_epi64): Ditto.
37736 (_mm512_mask_i64gather_epi64): Ditto.
37737 (_mm512_i32scatter_ps): Change __addr type to void*.
37738 (_mm512_mask_i32scatter_ps): Ditto.
37739 (_mm512_i32scatter_pd): Ditto.
37740 (_mm512_mask_i32scatter_pd): Ditto.
37741 (_mm512_i64scatter_ps): Ditto.
37742 (_mm512_mask_i64scatter_ps): Ditto.
37743 (_mm512_i64scatter_pd): Ditto.
37744 (_mm512_mask_i64scatter_pd): Ditto.
37745 (_mm512_i32scatter_epi32): Ditto.
37746 (_mm512_mask_i32scatter_epi32): Ditto.
37747 (_mm512_i32scatter_epi64): Ditto.
37748 (_mm512_mask_i32scatter_epi64): Ditto.
37749 (_mm512_i64scatter_epi32): Ditto.
37750 (_mm512_mask_i64scatter_epi32): Ditto.
37751 (_mm512_i64scatter_epi64): Ditto.
37752 (_mm512_mask_i64scatter_epi64): Ditto.
37753 * config/i386/avx512pfintrin.h
37754 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
37755 (_mm512_mask_prefetch_i32gather_ps): Ditto.
37756 (_mm512_mask_prefetch_i64gather_pd): Ditto.
37757 (_mm512_mask_prefetch_i64gather_ps): Ditto.
37758 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
37759 (_mm512_prefetch_i32scatter_ps): Ditto.
37760 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
37761 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
37762 (_mm512_prefetch_i64scatter_pd): Ditto.
37763 (_mm512_prefetch_i64scatter_ps): Ditto.
37764 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
37765 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
37766 * config/i386/avx512vlintrin.h
37767 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
37768 (_mm_mmask_i32gather_ps): Ditto.
37769 (_mm256_mmask_i32gather_pd): Ditto.
37770 (_mm_mmask_i32gather_pd): Ditto.
37771 (_mm256_mmask_i64gather_ps): Ditto.
37772 (_mm_mmask_i64gather_ps): Ditto.
37773 (_mm256_mmask_i64gather_pd): Ditto.
37774 (_mm_mmask_i64gather_pd): Ditto.
37775 (_mm256_mmask_i32gather_epi32): Ditto.
37776 (_mm_mmask_i32gather_epi32): Ditto.
37777 (_mm256_mmask_i32gather_epi64): Ditto.
37778 (_mm_mmask_i32gather_epi64): Ditto.
37779 (_mm256_mmask_i64gather_epi32): Ditto.
37780 (_mm_mmask_i64gather_epi32): Ditto.
37781 (_mm256_mmask_i64gather_epi64): Ditto.
37782 (_mm_mmask_i64gather_epi64): Ditto.
37783 (_mm256_i32scatter_ps): Change __addr type to void*.
37784 (_mm256_mask_i32scatter_ps): Ditto.
37785 (_mm_i32scatter_ps): Ditto.
37786 (_mm_mask_i32scatter_ps): Ditto.
37787 (_mm256_i32scatter_pd): Ditto.
37788 (_mm256_mask_i32scatter_pd): Ditto.
37789 (_mm_i32scatter_pd): Ditto.
37790 (_mm_mask_i32scatter_pd): Ditto.
37791 (_mm256_i64scatter_ps): Ditto.
37792 (_mm256_mask_i64scatter_ps): Ditto.
37793 (_mm_i64scatter_ps): Ditto.
37794 (_mm_mask_i64scatter_ps): Ditto.
37795 (_mm256_i64scatter_pd): Ditto.
37796 (_mm256_mask_i64scatter_pd): Ditto.
37797 (_mm_i64scatter_pd): Ditto.
37798 (_mm_mask_i64scatter_pd): Ditto.
37799 (_mm256_i32scatter_epi32): Ditto.
37800 (_mm256_mask_i32scatter_epi32): Ditto.
37801 (_mm_i32scatter_epi32): Ditto.
37802 (_mm_mask_i32scatter_epi32): Ditto.
37803 (_mm256_i32scatter_epi64): Ditto.
37804 (_mm256_mask_i32scatter_epi64): Ditto.
37805 (_mm_i32scatter_epi64): Ditto.
37806 (_mm_mask_i32scatter_epi64): Ditto.
37807 (_mm256_i64scatter_epi32): Ditto.
37808 (_mm256_mask_i64scatter_epi32): Ditto.
37809 (_mm_i64scatter_epi32): Ditto.
37810 (_mm_mask_i64scatter_epi32): Ditto.
37811 (_mm256_i64scatter_epi64): Ditto.
37812 (_mm256_mask_i64scatter_epi64): Ditto.
37813 (_mm_i64scatter_epi64): Ditto.
37814 (_mm_mask_i64scatter_epi64): Ditto.
37815 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
37816 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
37817 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
37818 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
37819 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
37820 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
37821 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
37822 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
37823 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
37824 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
37825 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
37826 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
37827 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
37828 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
37829 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
37830 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
37831 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
37832 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
37833 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
37834 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
37835 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
37836 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
37837 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
37838 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
37839 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
37840 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
37841 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
37842 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
37843 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
37844 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
37845 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
37846 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
37847 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
37848 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
37849 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
37850 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
37851 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
37852 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
37853 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
37854 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
37855 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
37856 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
37857 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
37858 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
37859 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
37860 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
37861 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
37862 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
37863 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
37864 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
37865 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
37866 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
37867 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
37868 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
37869 definitions accordingly.
37871 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
37872 Kuan-Lin Chen <kuanlinchentw@gmail.com>
37875 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
37878 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
37881 * ira-conflicts.c (ira_build_conflicts): Update total conflict
37882 hard regs for inner regno.
37884 2017-01-17 Martin Liska <mliska@suse.cz>
37887 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
37888 assumption and add comment.
37890 2017-01-17 Nathan Sidwell <nathan@acm.org>
37892 * ipa-visibility.c (localize_node): New function, broken out of ...
37893 (function_and_variable_visibility): ... here. Call it.
37895 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
37897 PR middle-end/77445
37898 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
37899 correctly set frequency of oudgoing edge.
37900 (duplicate_thread_path): Fix profile updating.
37902 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37905 * configure.ac: Add GCC_BASE_VER.
37906 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
37907 version from BASE-VER file.
37908 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
37909 (gcc.o): Depend on $(BASEVER).
37910 * common.opt (dumpfullversion): New option.
37911 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
37912 * doc/invoke.texi: Document -dumpfullversion.
37913 * doc/install.texi: Document --with-gcc-major-version-only.
37914 * configure: Regenerated.
37916 2017-01-17 Richard Biener <rguenther@suse.de>
37918 PR tree-optimization/71433
37919 * tree-vrp.c (register_new_assert_for): Merge same asserts
37920 on all incoming edges.
37921 (process_assert_insertions_for): Handle insertions at the
37924 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
37926 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
37927 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
37929 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
37932 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
37935 2017-01-17 Alan Modra <amodra@gmail.com>
37938 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
37939 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
37940 symbolic stack limit when pic.
37942 2017-01-16 Martin Sebor <msebor@redhat.com>
37944 PR tree-optimization/78608
37945 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
37947 2017-01-16 Jeff Law <law@redhat.com>
37950 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
37951 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
37952 for several include directories that may be relative to sysroot.
37953 * config/i386/x-mingw32 (gplus_includedir): Define.
37954 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
37955 (native_system_includedir): Likewise.
37956 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
37957 override if TARGET_SYSTEM_ROOT is defined.
37958 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
37960 PR tree-optimization/79090
37961 PR tree-optimization/33562
37962 PR tree-optimization/61912
37963 PR tree-optimization/77485
37964 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
37965 and computed trims into the dump file.
37967 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37969 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
37971 2017-01-16 Jakub Jelinek <jakub@redhat.com>
37974 * gimplify.c (gimplify_init_constructor): If want_value and
37975 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
37979 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
37980 sequence. Formatting fixes.
37981 (doloop_optimize): Formatting fixes.
37984 * gcc.c (debug_level_greater_than_spec_func): New function.
37985 (static_spec_functions): Add debug-level-gt spec function.
37986 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
37988 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37989 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37990 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
37991 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
37992 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
37993 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
37995 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
37997 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
37998 QImode fixups to general and mask registers only.
38000 2017-01-16 Carl Love <cel@us.ibm.com>
38002 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
38003 for built-in functions
38004 vector signed char vec_nabs (vector signed char)
38005 vector signed short vec_nabs (vector signed short)
38006 vector signed int vec_nabs (vector signed int)
38007 vector signed long long vec_nabs (vector signed long long)
38008 vector float vec_nabs (vector float)
38009 vector double vec_nabs (vector double)
38010 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
38012 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
38013 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
38014 * doc/extend.texi: Update the documentation file for the new built-in
38017 2017-01-16 Martin Sebor <msebor@redhat.com>
38019 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
38022 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38024 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
38025 UNSPEC_VSX__XXSPLTD to require special splat handling.
38027 2017-01-16 David Malcolm <dmalcolm@redhat.com>
38030 * system.h: Poison strndup.
38032 2017-01-16 Alan Modra <amodra@gmail.com>
38035 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
38038 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
38040 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
38042 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
38044 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
38045 call recog here. Assert that INSN_CODE (insn) is non-negative.
38047 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
38050 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
38052 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
38053 in the currently scheduled RTL fragment.
38055 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
38057 PR rtl-optimization/78751
38058 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
38061 2017-01-14 Jeff Law <law@redhat.com>
38063 PR tree-optimization/79090
38064 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
38065 variable length stores.
38066 (compute_trims): Delete dead assignment to *trim_tail.
38067 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
38070 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
38072 PR rtl-optimization/78626
38073 PR rtl-optimization/78727
38074 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
38075 of a block, and split such blocks after everything else is finished.
38077 2017-01-14 Alan Modra <amodra@gmail.com>
38080 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
38081 target legitimate_combined_insn.
38082 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
38083 (rs6000_legitimate_combined_insn): New function.
38084 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
38086 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
38087 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
38088 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
38090 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
38092 * doc/frontends.texi (G++ and GCC): Remove references to Java.
38094 2017-01-13 Jeff Law <law@redhat.com>
38096 PR tree-optimization/33562
38097 PR tree-optimization/61912
38098 PR tree-optimization/77485
38099 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
38101 (delete_dead_assignment): Likewise.
38102 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
38103 statement to delete_dead_call and delete_dead_assignment.
38105 2017-01-13 David Malcolm <dmalcolm@redhat.com>
38108 * substring-locations.c (format_warning_va): Strengthen case 1 so
38109 that both endpoints of the substring must be within the format
38110 range for just the substring to be printed.
38112 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
38114 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
38115 * config/i386/i386.c (ix86_target_string): Add missing options
38116 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
38117 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
38118 flags_other and ix86_target_other to flags2_other. Display unknown
38120 (ix86_valid_target_attribute_inner_p): Add missing options and
38121 reorder options by implied ISAs, as in ix86_target_string.
38123 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
38125 * hash-table.h (hash_table::too_empty_p): New function.
38126 (hash_table::expand): Use it.
38127 (hash_table::traverse): Likewise.
38128 (hash_table::empty_slot): Use sizeof (value_type) instead of
38129 sizeof (PTR) to convert bytes to elements. Shrink the table
38130 if the current size is excessive for the current number of
38133 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
38135 * ira-costs.c (record_reg_classes): Break from the inner loop
38136 early once alt_fail is known to be true. Update outer loop
38137 handling accordingly.
38139 2017-01-13 Jeff Law <law@redhat.com>
38141 * tree-ssa-dse.c (decrement_count): New function.
38142 (increment_start_addr, maybe_trim_memstar_call): Likewise.
38143 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
38144 when we know the partially dead statement is a mem* function.
38146 PR tree-optimization/61912
38147 PR tree-optimization/77485
38148 * tree-ssa-dse.c: Include expr.h.
38149 (maybe_trim_constructor_store): New function.
38150 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
38152 PR tree-optimization/33562
38153 PR tree-optimization/61912
38154 PR tree-optimization/77485
38155 * doc/invoke.texi: Document new dse-max-object-size param.
38156 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
38157 * tree-ssa-dse.c: Include params.h.
38158 (dse_store_status): New enum.
38159 (initialize_ao_ref_for_dse): New, partially extracted from
38161 (valid_ao_ref_for_dse, normalize_ref): New.
38162 (setup_live_bytes_from_ref, compute_trims): Likewise.
38163 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
38164 (maybe_trim_partially_dead_store): Likewise.
38165 (maybe_trim_complex_store): Likewise.
38166 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
38167 Track what bytes live from the original store. Return tri-state
38168 for dead, partially dead or live.
38169 (dse_dom_walker): Add constructor, destructor and new private members.
38170 (delete_dead_call, delete_dead_assignment): New extracted from
38172 (dse_optimize_stmt): Make a member of dse_dom_walker.
38173 Use initialize_ao_ref_for_dse.
38175 PR tree-optimization/33562
38176 PR tree-optimization/61912
38177 PR tree-optimization/77485
38178 * sbitmap.h (bitmap_count_bits): Prototype.
38179 (bitmap_clear_range, bitmap_set_range): Likewise.
38180 * sbitmap.c (bitmap_clear_range): New function.
38181 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
38183 2017-01-13 Martin Liska <mliska@suse.cz>
38186 * function.c (set_cfun): Add new argument force.
38187 * function.h (set_cfun): Likewise.
38188 * ipa-inline-transform.c (inline_call): Use the function when
38189 strict alising from is dropped for function we inline to.
38191 2017-01-13 Richard Biener <rguenther@suse.de>
38193 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
38194 for dumping GIMPLE INTEGER_CSTs.
38196 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
38198 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
38199 to 201112L since C++17.
38201 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
38204 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
38205 if -fsanitize=kernel-address is present.
38207 2017-01-13 Richard Biener <rguenther@suse.de>
38209 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
38210 as _Literal ( type ) number in case usual suffixes do not
38211 preserve all information.
38213 2017-01-13 Richard Biener <rguenther@suse.de>
38215 PR tree-optimization/77283
38216 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
38217 and ssa-iterators.h.
38218 (is_feasible_trace): Implement a cost model based on joiner
38221 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
38224 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
38225 char or short to __float128/_Float128 directly.
38227 2017-01-12 Martin Sebor <msebor@redhat.com>
38229 to -Wformat-overflow.
38230 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
38231 (min_bytes_remaining): Same.
38232 (get_string_length): Same.
38233 (format_string): Same.
38234 (format_directive): Same.
38236 (pass_sprintf_length::handle_gimple_call): Same.
38238 2017-01-12 Jakub Jelinek <jakub@redhat.com>
38240 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
38241 info.nowrite calls with no lhs that can't throw. Return bool
38242 whether gsi_remove has been called or not.
38243 (pass_sprintf_length::handle_gimple_call): Return bool whether
38244 try_substitute_return_value called gsi_remove. Formatting fix.
38245 (pass_sprintf_length::execute): Don't use gsi_remove if
38246 handle_gimple_call returned true.
38249 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
38250 be removed due to side-effects, don't remove following barrier nor
38251 turn the successor edge into fallthru edge.
38253 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38256 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
38257 element-reversing loads and stores as not swappable.
38259 2017-01-12 Nathan Sidwell <nathan@acm.org>
38260 Nicolai Stange <nicstange@gmail.com>
38262 * combine.c (try_combine): Don't ignore result of overlap checking
38263 loop. Combine overlap & asm check into single loop.
38265 2017-01-12 Richard Biener <rguenther@suse.de>
38267 * tree-pretty-print.c (dump_generic_node): Provide -gimple
38268 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
38270 2017-01-12 Richard Biener <rguenther@suse.de>
38272 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
38273 and TS_TARGET_OPTION directly derive from TS_BASE.
38274 * tree-core.h (tree_optimization_option): Derive from tree_base.
38275 (tree_target_option): Likewise.
38277 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
38279 * config/i386/i386.c (memory_address_length): Increase len
38280 only when rip_relative_addr_p returns false.
38282 2017-01-11 Julia Koval <julia.koval@intel.com>
38284 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
38285 (OPTION_MASK_ISA_SGX_SET): New.
38286 (ix86_handle_option): Handle OPT_msgx.
38287 * config.gcc: Added sgxintrin.h.
38288 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
38289 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
38290 * config/i386/i386.c (ix86_target_string): Add -msgx.
38292 (ix86_option_override_internal): Handle new options.
38293 (ix86_valid_target_attribute_inner_p): Add sgx.
38294 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
38295 * config/i386/i386.opt: Add msgx.
38296 * config/i386/sgxintrin.h: New file.
38297 * config/i386/x86intrin.h: Add sgxintrin.h.
38299 2017-01-11 Jakub Jelinek <jakub@redhat.com>
38302 * fold-const.c (maybe_nonzero_address): Return 1 for function
38304 (tree_single_nonzero_warnv_p): Don't handle function local objects
38308 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
38311 2017-01-11 David Malcolm <dmalcolm@redhat.com>
38314 * opts.c: Include "spellcheck.h"
38315 (struct string_fragment): New struct.
38316 (struct edit_distance_traits<const string_fragment &>): New
38318 (get_closest_sanitizer_option): New function.
38319 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
38321 2017-01-11 Jakub Jelinek <jakub@redhat.com>
38323 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
38325 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
38326 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
38327 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
38328 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
38329 for initial die_offset if dwarf_split_debug_info.
38330 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
38331 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
38333 (output_skeleton_debug_sections): Formatting fix. Use
38334 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
38335 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
38337 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
38339 * config/arm/cortex-a53.md: Add bypasses for
38340 cortex_a53_r2f_cvt.
38341 (cortex_a53_r2f): Only use for transfers.
38342 (cortex_a53_f2r): Likewise.
38343 (cortex_a53_r2f_cvt): Add reservation for conversions.
38344 (cortex_a53_f2r_cvt): Likewise.
38346 2017-01-11 Tamar Christina <tamar.christina@arm.com>
38348 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
38349 to all inlined functions, change static to extern.
38351 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
38354 * config/arm/arm.c (legitimize_pic_address): Handle reference to
38356 (arm_assemble_integer): Likewise.
38358 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
38360 * config.gcc: Use new awk script to check CPU, FPU and architecture
38361 parameters for --with-... options.
38362 * config/arm/parsecpu.awk: New file
38363 * config/arm/arm-cpus.in: New file.
38364 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
38366 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
38368 * config/arm/t-arm: Update dependency rules.
38369 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
38370 of processing .def files.
38371 * config/arm/genopt.sh: Deleted.
38372 * config/arm/gentune.sh: Deleted.
38373 * config/arm/arm-cores.def: Deleted.
38374 * config/arm/arm-arches.def: Deleted.
38375 * config/arm/arm-fpus.def: Deleted.
38376 * config/arm/arm-tune.md: Regenerated.
38377 * config/arm/arm-tables.opt: Regenerated.
38378 * config/arm/arm-cpu.h: New generated file.
38379 * config/arm/arm-cpu-data.h: New generated file.
38380 * config/arm/arm-cpu-cdata.h: New generated file.
38382 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
38385 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
38387 (input_varpool_node): Unpack dynamically_initialized bit.
38389 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
38391 PR rtl-optimization/79032
38392 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
38393 the alignment of the adjusted memory reference against that of MODE,
38394 instead of the alignment of the original memory reference.
38396 2017-01-11 Martin Jambor <mjambor@suse.cz>
38398 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
38400 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
38401 decorated functions.
38403 2017-01-11 Richard Biener <rguenther@suse.de>
38405 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
38406 set range/nonnull info for PHI results. Do not set it on
38407 stmts marked for removal.
38409 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
38411 * expr.c (store_field): In the bitfield case, fetch the return value
38412 from the registers before applying a single big-endian adjustment.
38413 Always do a final load for a BLKmode value not larger than a word.
38415 2017-01-10 David Malcolm <dmalcolm@redhat.com>
38418 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
38419 that we correctly handle column numbers greater than
38420 LINE_MAP_MAX_COLUMN_NUMBER.
38422 2017-01-10 Martin Sebor <msebor@redhat.com>
38424 PR middle-end/78245
38425 * gimple-ssa-sprintf.c (get_destination_size): Call
38426 {init,fini}object_sizes.
38427 * tree-object-size.c (addr_object_size): Adjust.
38428 (pass_through_call): Adjust.
38429 (pass_object_sizes::execute): Adjust.
38430 * tree-object-size.h (fini_object_sizes): Declare.
38432 2017-01-10 Martin Sebor <msebor@redhat.com>
38434 PR tree-optimization/78775
38435 * builtins.c (get_size_range): Move...
38436 * calls.c: ...to here.
38437 (alloc_max_size): Accept zero argument.
38438 (operand_signed_p): Remove.
38439 (maybe_warn_alloc_args_overflow): Call get_size_range.
38440 * calls.h (get_size_range): Declare.
38442 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
38444 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
38445 from TI's devices.csv file as of September 2016.
38446 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
38448 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
38450 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
38451 * doc/invoke.texi: Likewise.
38452 * doc/md.texi: Likewise.
38453 * doc/objc.texi: Likewise.
38455 2017-01-10 Joshua Conner <joshconner@google.com>
38457 * config/arm/fuchsia-elf.h: New file.
38458 * config/fuchsia.h: New file.
38459 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
38460 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
38462 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
38464 2016-01-10 Richard Biener <rguenther@suse.de>
38466 PR tree-optimization/79034
38467 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
38468 Propagate out degenerate PHIs in the joiner.
38470 2017-01-10 Martin Liska <mliska@suse.cz>
38472 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
38473 (sort_congruence_classes_by_decl_uid): Likewise.
38474 (sort_congruence_class_groups_by_decl_uid): Likewise.
38475 (sem_item_optimizer::merge_classes): Sort class, groups in these
38476 classes and members in the groups by DECL_UID of declarations.
38477 This would make merge operations stable.
38479 2017-01-10 Martin Liska <mliska@suse.cz>
38481 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
38482 usage of m_classes_vec.
38483 (sem_item_optimizer::~sem_item_optimizer): Likewise.
38484 (sem_item_optimizer::get_group_by_hash): Likewise.
38485 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
38486 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
38487 (sem_item_optimizer::verify_classes): Likewise.
38488 (sem_item_optimizer::process_cong_reduction): Likewise.
38489 (sem_item_optimizer::dump_cong_classes): Likewise.
38490 (sem_item_optimizer::merge_classes): Likewise.
38491 * ipa-icf.h (congruence_class_hash): Rename from
38492 congruence_class_group_hash. Remove declaration of m_classes_vec.
38494 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
38496 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
38497 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
38498 * config.gcc: Add avx512vpopcntdqintrin.h.
38499 * config/i386/avx512vpopcntdqintrin.h: New.
38500 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
38501 * config/i386/i386-builtin-types.def: Add new types.
38502 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
38503 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
38504 __builtin_ia32_vpopcountq_v8di_mask): New.
38505 * config/i386/i386-c.c (ix86_target_macros_internal): Define
38506 __AVX512VPOPCNTDQ__.
38507 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
38508 (PTA_AVX512VPOPCNTDQ): Define.
38509 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
38510 TARGET_AVX512VPOPCNTDQ_P): Define.
38511 * config/i386/i386.opt: Add mavx512vpopcntdq.
38512 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
38513 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
38515 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38517 PR middle-end/77484
38518 * predict.def (PRED_CALL): Set to 67.
38520 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
38522 * expr.c (store_field): In the bitfield case, if the value comes from
38523 a function call and is of an aggregate type returned in registers, do
38524 not modify the field mode; extract the value in all cases if the mode
38525 is BLKmode and the size is not larger than a word.
38527 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
38530 * config/i386/cpuid.h: Fix undefined behavior.
38532 2017-01-04 Jeff Law <law@redhat.com>
38534 PR tree-optimization/79007
38535 PR tree-optimization/67955
38536 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
38537 conservative for pt.null when flag_non_call_exceptions is on.
38539 2017-01-09 Jakub Jelinek <jakub@redhat.com>
38541 PR translation/79019
38542 PR translation/79020
38543 * params.def (PARAM_INLINE_MIN_SPEEDUP,
38544 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
38545 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
38547 * config/avr/avr.opt (maccumulate-args): Likewise.
38548 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
38549 * common.opt (freport-bug): Likewise.
38550 * cif-code.def (CIF_FINAL_ERROR): Likewise.
38551 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
38552 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
38553 translatable string.
38554 * config/i386/i386.c (function_value_32): Likewise.
38555 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
38556 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
38558 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
38559 * common/config/msp430/msp430-common.c (msp430_handle_option):
38561 * symtab.c (symtab_node::verify_base): Likewise.
38562 * opts.c (set_debug_level): Likewise.
38563 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
38564 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
38565 missing whitespace to translatable strings.
38566 * config/avr/avr.md (bswapsi2): Fix typo in comment.
38567 * config/sh/superh.h: Likewise.
38568 * config/i386/xopintrin.h: Likewise.
38569 * config/i386/znver1.md: Likewise.
38570 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
38571 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
38572 * double-int.h (struct double_int): Likewise.
38573 * double-int.c (div_and_round_double): Likewise.
38574 * wide-int.cc: Likewise.
38575 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
38576 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
38577 * cfgcleanup.c (crossjumps_occured): Renamed to ...
38578 (crossjumps_occurred): ... this.
38579 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
38582 PR tree-optimization/78899
38583 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
38584 returning bool return struct loop *, NULL for failure and the new
38586 (versionable_outer_loop_p): Don't version outer loop if it has
38587 dont_vectorized bit set.
38588 (tree_if_conversion): When versioning outer loop, ensure
38589 tree_if_conversion is performed also on the inner loop of the
38590 non-vectorizable outer loop copy.
38591 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
38592 LOOP_VECTORIZED in inner loop of the scalar outer loop and
38593 prevent vectorization of it.
38594 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
38595 the outer loop vectorization of the non-scalar version is attempted
38596 before vectorization of the inner loop in scalar version. If
38597 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
38598 vectorization of its inner loop.
38599 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
38600 has 2 inner loops, rename also on edges from bb whose single pred
38601 is outer_loop->header. Fix typo in function comment.
38603 2017-01-09 Martin Sebor <msebor@redhat.com>
38606 * asan.c (asan_emit_stack_protection): Increase local buffer size
38607 to avoid snprintf truncation warning.
38609 2017-01-09 Andrew Pinski <apinski@cavium.com>
38611 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
38612 to reference thunderx2t99 for the tuning structure
38613 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
38615 (thunderx2t99_extra_costs): This.
38616 * config/aarch64/aarch64-tune.md: Regenerate.
38617 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
38618 (vulcan_addrcost_table): This.
38619 (vulcan_regmove_cost): Rename to ...
38620 (thunderx2t99_regmove_cost): This.
38621 (vulcan_vector_cost): Rename to ...
38622 (thunderx2t99_vector_cost): this.
38623 (vulcan_branch_cost): Rename to ...
38624 (thunderx2t99_branch_cost): This.
38625 (vulcan_tunings): Rename to ...
38626 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
38627 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
38629 2017-01-09 Martin Jambor <mjambor@suse.cz>
38633 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
38634 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
38635 (propagate_vr_accross_jump_function): Use the above function for all
38636 value range computations for pass-through jump functions and type
38637 converasion from explicit value range values.
38638 (ipcp_propagate_stage): Do not attempt to deduce types of formal
38639 parameters from TYPE_ARG_TYPES.
38640 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
38641 (ipa_write_node_info): Stream type of the actual argument.
38642 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
38644 2017-01-09 Martin Liska <mliska@suse.cz>
38647 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
38648 (lookup_compiler): Do not show error message with have_E.
38650 2017-01-09 Jakub Jelinek <jakub@redhat.com>
38652 PR tree-optimization/78938
38653 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
38654 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
38655 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
38656 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
38659 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38661 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
38664 2017-01-09 Richard Biener <rguenther@suse.de>
38666 PR tree-optimization/78997
38667 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
38668 name condition properly.
38670 2017-01-09 Richard Biener <rguenther@suse.de>
38673 * dwarf2out.c (is_cxx): New overload with context.
38674 (is_naming_typedef_decl): Use it.
38676 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38678 * invoke.texi (Option Summary): Correct spacing in option lists
38679 and add line breaks to fix over-long lines.
38681 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38683 PR middle-end/17660
38685 * extend.texi (Common Variable Attributes): Add xref to GCC
38686 Internals manual to explain mode attribute keywords.
38688 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38691 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
38692 and Preprocessor Options.
38693 (Options for Linking): Document -pthread here....
38694 (RS/6000 and PowerPC Options): ...not here.
38695 (Solaris 2 Options): ...or here.
38696 * doc/cppopts.texi: Document -pthread.
38698 2017-01-08 Martin Sebor <msebor@redhat.com>
38700 PR middle-end/77708
38701 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
38702 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
38703 New member functions.
38704 (format_directive): Used them.
38706 (pass_sprintf_length::handle_gimple_call): Same.
38707 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
38708 to avoid truncation for any argument.
38709 (extract_affine_mul): Same.
38710 * tree.c (get_file_function_name): Same.
38712 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38714 PR middle-end/77484
38715 * predict.def (PRED_INDIR_CALL): Set to 86.
38717 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38719 PR preprocessor/54124
38720 * doc/cppopts.texi: Reformat -d subtable to list the full name
38721 of the options. Add cross-reference to the docs for the general
38722 compiler -d options.
38723 * doc/invoke.texi (Developer Options): Add cross-reference to the
38724 preprocessor-specific -d option documentation.
38726 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38728 PR preprocessor/13498
38729 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
38730 redudant material, and reflect new command-line options.
38731 (System Headers): Likewise.
38733 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38735 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
38736 -isystem, and -idirafter. Copy-edit.
38737 * doc/cppopts.texi: Copy-edit. Remove contradiction about
38738 default for -ftrack-macro-expansion. Delete obsolete and
38739 badly-formatted implementation details about -fdebug-cpp output.
38740 * doc/cppwarnopts.texi: Copy-edit.
38742 2017-01-07 David Malcolm <dmalcolm@redhat.com>
38745 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
38746 that the transition from a max line width >= 1<<10 to narrower
38747 lines works correctly.
38749 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
38751 * doc/options.texi (PerFunction): New.
38752 * opt-functions.awk (switch_flags): Map both Optimization and
38753 PerFunction to CL_OPTIMIZATION.
38754 * opth-gen.awk: Test for PerFunction flag along with
38756 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
38757 it only when the latter is present. Skip those that don't in
38758 the hash function generator.
38759 * common.opt (fvar-tracking): Mark as PerFunction instead of
38761 (fvar-tracking-assignments): Likewise.
38762 (fvar-tracking-assignments-toggle): Likewise.
38763 (fvar-tracking-uninit): Likewise.
38765 2017-01-07 Jakub Jelinek <jakub@redhat.com>
38767 PR translation/79018
38768 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
38771 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
38774 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
38775 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
38776 TARGET_LONG_JUMP_TABLE_OFFSETS.
38777 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
38778 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
38779 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
38780 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
38781 * config/m68k/m68k.md (tablejump expander): Likewise.
38782 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
38783 TARGET_LONG_JUMP_TABLE_OFFSETS.
38784 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
38785 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
38787 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
38788 David Holsgrove <david.holsgrove@xilinx.com>
38790 * common/config/microblaze/microblaze-common.c
38791 (TARGET_EXCEPT_UNWIND_INFO): Remove.
38792 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
38794 * config/microblaze/microblaze.c (microblaze_must_save_register)
38795 (microblaze_expand_epilogue, microblaze_return_addr): Handle
38797 (microblaze_eh_return): New function.
38798 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
38799 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
38800 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
38801 * config/microblaze/microblaze.md (eh_return): New pattern.
38803 2017-01-06 Jakub Jelinek <jakub@redhat.com>
38805 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
38806 GCC_DIAGNOSTIC_STRINGIFY): Define.
38808 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
38810 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38812 * config/arm/arm.md (<mcrr>): New.
38814 * config/arm/arm.c (arm_arch5te): New.
38815 (arm_option_override): Set arm_arch5te.
38816 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
38818 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
38819 (arm_mcrr_qualifiers): ... this. New.
38820 (MRRC_QUALIFIERS): Define to...
38821 (arm_mrrc_qualifiers): ... this. New.
38822 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
38824 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
38825 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
38826 (MRRCI, mrrc, MRRC): New.
38827 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
38828 VUNSPEC_MRRC2): New.
38830 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38832 * config/arm/arm.md (<mcr>): New.
38834 * config/arm/arm.c (arm_coproc_builtin_available): Add
38835 support for mcr, mrc, mcr2 and mrc2.
38836 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
38837 (arm_mcr_qualifiers): ... this. New.
38838 (MRC_QUALIFIERS): Define to ...
38839 (arm_mrc_qualifiers): ... this. New.
38840 (MCR_QUALIFIERS): Define to ...
38841 (arm_mcr_qualifiers): ... this. New.
38842 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
38844 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
38845 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
38846 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
38847 VUNSPEC_MRC2): New.
38849 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38851 * config/arm/arm.md (*ldc): New.
38855 * config/arm/arm.c (arm_coproc_builtin_available): Add
38856 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
38857 (arm_coproc_ldc_stc_legitimate_address): New.
38858 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
38859 'qualifier_const_pointer'.
38860 (LDC_QUALIFIERS): Define to...
38861 (arm_ldc_qualifiers): ... this. New.
38862 (STC_QUALIFIERS): Define to...
38863 (arm_stc_qualifiers): ... this. New.
38864 * config/arm/arm-protos.h
38865 (arm_coproc_ldc_stc_legitimate_address): New.
38866 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
38867 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
38868 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
38869 stc2, stcl, stc2l): New.
38870 * config/arm/constraints.md (Uz): New.
38871 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
38872 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
38873 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
38874 VUNSPEC_STC2L): New.
38876 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38878 * config/arm/arm.md (<cdp>): New.
38879 * config/arm/arm.c (neon_const_bounds): Rename this ...
38880 (arm_const_bounds): ... this.
38881 (arm_coproc_builtin_available): New.
38882 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
38883 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
38884 (CDP_QUALIFIERS): Define to...
38885 (arm_cdp_qualifiers): ... this. New.
38887 (arm_expand_builtin_args): Add case for 6 arguments.
38888 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
38889 (arm_const_bounds): ... this.
38890 (arm_coproc_builtin_available): New.
38891 * config/arm/arm_acle.h (__arm_cdp): New.
38893 * config/arm/arm_acle_builtins.def (cdp): New.
38895 * config/arm/iterators.md (CDPI,CDP,cdp): New.
38896 * config/arm/neon.md: Rename all 'neon_const_bounds' to
38897 'arm_const_bounds'.
38898 * config/arm/types.md (coproc): New.
38899 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
38900 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
38901 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
38902 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
38904 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38906 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
38907 (UBINOP_QUALIFIERS): New.
38909 (acle_builtin_data): New. Change comment.
38910 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
38911 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
38912 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
38913 arm_acle_builtins.def.
38914 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
38915 (arm_init_acle_builtins): New.
38916 (CRC32_BUILTIN): Remove.
38917 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
38918 crc32cb, crc32ch and crc32cw.
38919 (arm_init_crc32_builtins): Remove.
38920 (arm_init_builtins): Use arm_init_acle_builtins rather
38921 than arm_init_crc32_builtins.
38922 (arm_expand_acle_builtin): New.
38923 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
38924 * config/arm/arm_acle_builtins.def: New.
38926 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38928 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
38929 (arm_builtin_datum): ... this.
38930 (arm_init_neon_builtin): Rename to ...
38931 (arm_init_builtin): ... this. Add a new parameters PREFIX
38932 and USE_SIG_IN_NAME.
38933 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
38934 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
38935 'arm_builtin_datum'.
38936 (arm_init_vfp_builtins): Likewise.
38937 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
38938 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
38939 (arm_expand_neon_args): Rename to ...
38940 (arm_expand_builtin_args): ... this. Rename builtin_arg
38941 enum values and differentiate between ARG_BUILTIN_MEMORY
38942 and ARG_BUILTIN_NEON_MEMORY.
38943 (arm_expand_neon_builtin_1): Rename to ...
38944 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
38945 values, arm_expand_builtin_args and add bool parameter NEON.
38946 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
38947 (arm_expand_vfp_builtin): Likewise.
38948 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
38950 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38952 PR middle-end/77484
38953 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
38954 * predict.c (tree_estimate_probability_bb): Reverse direction of
38955 polymorphic call predictor.
38957 2017-01-06 David Malcolm <dmalcolm@redhat.com>
38959 * passes.c (execute_one_pass): Split out pass-skipping logic into...
38960 (determine_pass_name_match): ...this new function and...
38961 (should_skip_pass_p): ...this new function.
38963 2017-01-06 Nathan Sidwell <nathan@acm.org>
38965 * ipa-visibility.c (function_and_variable_visibility): Reformat
38966 comments and long lines. Remove extrneous if.
38967 * symtab.c (symtab_node::make_decl_local): Fix code format.
38968 (symtab_node::set_section_for_node): Fix comment typo.
38970 2017-01-06 Martin Liska <mliska@suse.cz>
38973 * lra-constraints.c: Rename invariant to lra_invariant.
38974 * predict.c (set_even_probabilities): Initialize e to NULL.
38976 2017-01-05 Martin Sebor <msebor@redhat.com>
38978 PR tree-optimization/78910
38979 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
38980 (format_integer): Correct off-by-one error in the handling
38981 of precision with negative numbers in signed conversions..
38983 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
38985 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
38987 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38989 PR tree-optimization/71016
38990 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
38991 factor_out_conditional_conversion. Formatting fix.
38992 (factor_out_conditional_conversion): Add cond_stmt argument.
38993 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
38994 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
38997 2017-01-05 David Malcolm <dmalcolm@redhat.com>
38999 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
39000 read-rtl-function.o, and selftest-rtl.o.
39001 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
39002 (selftest::aarch64_test_loading_full_dump): New function.
39003 (selftest::aarch64_run_selftests): New function.
39004 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
39005 selftest::aarch64_run_selftests.
39006 * config/i386/i386.c
39007 (selftest::ix86_test_loading_dump_fragment_1): New function.
39008 (selftest::ix86_test_loading_call_insn): New function.
39009 (selftest::ix86_test_loading_full_dump): New function.
39010 (selftest::ix86_test_loading_unspec): New function.
39011 (selftest::ix86_run_selftests): Call the new functions.
39012 * emit-rtl.c (maybe_set_max_label_num): New function.
39013 * emit-rtl.h (maybe_set_max_label_num): New decl.
39014 * function.c (instantiate_decls): Guard call to
39015 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
39016 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
39018 * gensupport.c (gen_reader::gen_reader): Pass "false"
39019 for new "compact" param of rtx_reader.
39020 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
39021 rather than an empty string for NULL strings.
39022 * read-md.c: Potentially include config.h rather than bconfig.h.
39023 Wrap include of errors.h with #ifdef GENERATOR_FILE.
39024 (have_error): New global, copied from errors.c.
39025 (md_reader::read_name): Rename to...
39026 (md_reader::read_name_1): ...this, adding "out_loc" param,
39027 and converting "missing name or number" to returning false, rather
39029 (md_reader::read_name): Reimplement in terms of read_name_1.
39030 (md_reader::read_name_or_nil): New function.
39031 (md_reader::read_string): Handle "(nil)" by returning NULL.
39032 (md_reader::md_reader): Add new param "compact".
39033 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
39034 (md_reader::read_file): New method.
39035 * read-md.h (md_reader::md_reader): Add new param "compact".
39036 (md_reader::read_file): New method.
39037 (md_reader::is_compact): New accessor.
39038 (md_reader::read_name): Convert return type from void to file_location.
39039 (md_reader::read_name_or_nil): New decl.
39040 (md_reader::read_name_1): New decl.
39041 (md_reader::m_compact): New field.
39042 (noop_reader::noop_reader): Pass "false" for new "compact" param
39044 (rtx_reader::rtx_reader): Add new "compact" param.
39045 (rtx_reader::read_rtx_operand): Make virtual and convert return
39046 type from void to rtx.
39047 (rtx_reader::read_until): New decl.
39048 (rtx_reader::handle_any_trailing_information): New virtual function.
39049 (rtx_reader::postprocess): New virtual function.
39050 (rtx_reader::finalize_string): New virtual function.
39051 (rtx_reader::m_in_call_function_usage): New field.
39052 (rtx_reader::m_reuse_rtx_by_id): New field.
39053 * read-rtl-function.c: New file.
39054 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
39055 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
39056 (selftest::verify_three_block_rtl_cfg): New decl.
39057 * read-rtl-function.h: New file.
39058 * read-rtl.c: Potentially include config.h rather than bconfig.h.
39059 For host, include function.h, memmodel.h, and emit-rtl.h.
39060 (one_time_initialization): New function.
39061 (struct compact_insn_name): New struct.
39062 (compact_insn_names): New array.
39063 (find_code): Handle insn codes in compact dumps.
39064 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
39065 (bind_subst_iter_and_attr): Likewise.
39066 (add_condition_to_string): Likewise.
39067 (add_condition_to_rtx): Likewise.
39068 (apply_attribute_uses): Likewise.
39069 (add_current_iterators): Likewise.
39070 (apply_iterators): Likewise.
39071 (initialize_iterators): Guard usage of apply_subst_iterator with
39072 #ifdef GENERATOR_FILE.
39073 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
39074 (md_reader::read_mapping): Likewise.
39075 (add_define_attr_for_define_subst): Likewise.
39076 (add_define_subst_attr): Likewise.
39077 (read_subst_mapping): Likewise.
39078 (check_code_iterator): Likewise.
39079 (rtx_reader::read_rtx): Likewise. Move one-time initialization
39081 (one_time_initialization): New function.
39082 (rtx_reader::read_until): New method.
39083 (read_flags): New function.
39084 (parse_reg_note_name): New function.
39085 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
39086 Handle reuse_rtx ids.
39087 Wrap iterator lookup within #ifdef GENERATOR_FILE.
39088 Add parsing support for RTL dumps, mirroring the special-cases in
39089 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
39090 values, and calling handle_any_trailing_information.
39091 (rtx_reader::read_rtx_operand): Convert return type from void
39092 to rtx, returning return_rtx. Handle case 'e'. Call
39093 finalize_string on XSTR and XTMPL fields.
39094 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
39095 "(nil)" values were omitted. Call the postprocess vfunc on the
39097 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
39098 class ctor. Initialize m_in_call_function_usage. Call
39099 one_time_initialization.
39100 * rtl-tests.c (selftest::test_uncond_jump): Call
39101 set_new_first_and_last_insn.
39102 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
39103 * selftest-rtl.c: New file.
39104 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
39105 (selftest::get_insn_by_uid): New decl.
39106 * selftest-run-tests.c (selftest::run_tests): Call
39107 read_rtl_function_c_tests.
39108 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
39109 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
39112 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
39114 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
39115 operands in a special way. Assert that pos+len <= mode precision.
39117 2017-01-05 Jakub Jelinek <jakub@redhat.com>
39119 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
39120 3 argument Alias with unlimited for the negative form.
39121 (fno-vect-cost-model): Removed.
39123 2017-01-05 Martin Liska <mliska@suse.cz>
39125 * hsa-gen.c (gen_hsa_divmod): New function.
39126 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
39128 2017-01-05 Martin Liska <mliska@suse.cz>
39131 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
39134 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39136 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
39137 small constant length operands.
39139 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39141 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
39142 between loop iterations.
39144 2017-01-05 Martin Liska <mliska@suse.cz>
39147 * gimplify.c (gimplify_decl_expr): Compare to
39148 asan_poisoned_variables instread of checking flags.
39149 (gimplify_target_expr): Likewise.
39150 (gimplify_expr): Likewise.
39151 (gimplify_function_tree): Conditionally initialize
39152 asan_poisoned_variables.
39154 2017-01-04 Jeff Law <law@redhat.com>
39156 PR tree-optimizatin/78812
39157 * rtl.h (contains_mem_rtx_p): Prototype.
39158 * ifcvt.c (containts_mem_rtx_p): Move from here to...
39159 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
39160 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
39161 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
39162 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
39164 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39166 * input.c (assert_char_at_range): Default-initialize actual_range.
39168 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39170 * df-scan.c (df_ref_create_structure): Make regno unsigned,
39171 to match the caller.
39173 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39175 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
39176 insns after final jump in test to emit dummy move.
39178 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39180 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
39181 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
39183 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39185 * multiple_target.c (create_dispatcher_calls): Init e_next.
39186 * tree-ssa-loop-split.c (split_loop): Init border.
39187 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
39190 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
39195 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
39196 (altivec_register_operand): Do not return true if the operand
39197 contains a SUBREG mixing SImode and SFmode.
39198 (vsx_register_operand): Likewise.
39199 (vsx_reg_sfsubreg_ok): New predicate.
39200 (vfloat_operand): Do not return true if the operand contains a
39201 SUBREG mixing SImode and SFmode.
39202 (vint_operand): Likewise.
39203 (vlogical_operand): Likewise.
39204 (gpc_reg_operand): Likewise.
39205 (int_reg_operand): Likewise.
39206 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
39207 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
39208 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
39210 (rs6000_emit_move_si_sf_subreg): New helper function.
39211 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
39212 fixup SUBREGs involving SImode and SFmode.
39213 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
39214 numbers for the new peephole2 optimization.
39215 (peephole2 for SFmode unions): New peephole2 to optimize cases in
39216 the GLIBC math library that do AND/IOR/XOR operations on single
39217 precision floating point.
39218 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
39219 target macros to say whether we need to avoid SUBREGs mixing
39221 (TARGET_ALLOW_SF_SUBREG): Likewise.
39222 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
39223 (UNSPEC_SI_FROM_SF): Likewise.
39224 (iorxor): Change spacing.
39225 (and_ior_xor): New iterator for AND, IOR, and XOR.
39226 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
39227 (movdi_from_sf_zero_ext): Likewise.
39228 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
39229 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
39230 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
39231 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
39232 (fms<mode>4): Likewise.
39233 (fnma<mode>4): Likewise.
39234 (fnms<mode>4): Likewise.
39235 (nfma<mode>4): Likewise.
39236 (nfms<mode>4): Likewise.
39238 2017-01-04 Marek Polacek <polacek@redhat.com>
39241 * doc/invoke.texi: Document -Wpointer-compare.
39243 2017-01-04 Jakub Jelinek <jakub@redhat.com>
39245 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
39248 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
39249 descriptions for -gdwarf-5 and emit them as uleb128 instead of
39252 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
39255 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
39256 documentation of the powerpc_popcntb_ok attribute.
39257 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
39258 code to issue warning messages if a requested CPU configuration is
39259 not supported by the binary (assembler and loader) toolchain.
39260 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
39261 made to define a built-in function that has been disabled.
39262 (paired_init_builtins): Add assertion to prevent ICE if attempt is
39263 made to define a built-in function that has been disabled.
39264 (altivec_init_builtins): Add comment explaining why definition
39265 of the DST built-in functions is not preceded by an assertion
39266 check. Add assertions to prevent ICE if attempts are made to
39267 define an altivec predicate or an abs* built-in function that has
39269 (htm_init_builtins): Add comment explaining why definition of the
39270 htm built-in functions is not preceded by an assertion check.
39272 2017-01-04 Jeff Law <law@redhat.com>
39274 PR tree-optimizatin/67955
39275 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
39276 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
39277 the points-to solution does not include pt_null. Use DECL_PT_UID
39280 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
39282 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
39283 Use gen_int_mode instead of gen_lopwart for const_int operands.
39285 2017-01-04 Jakub Jelinek <jakub@redhat.com>
39287 PR tree-optimization/71563
39288 * match.pd: Simplify X << Y into X if Y is known to be 0 or
39289 out of range value - has low bits known to be zero.
39291 2017-01-04 Alan Modra <amodra@gmail.com>
39293 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
39294 * configure: Regenerate.
39295 * config.in: Regenerate.
39297 2017-01-04 Jakub Jelinek <jakub@redhat.com>
39300 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
39301 a substring of the message, but strcmp with the whole message. Ifdef
39302 ENABLE_NLS, translate the message first using dgettext.
39304 2017-01-03 Jeff Law <law@redhat.com>
39306 PR tree-optimizatin/78856
39307 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
39308 (mark_threaded_blocks): Remove code to truncate thread paths that
39309 cross multiple loop headers. Instead invalidate the cached loop
39310 iteration information and handle case of a thread path walking
39311 into an irreducible region.
39313 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
39316 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
39317 assertions. Add support for doing the signbit if the IEEE 128-bit
39318 floating point value is in a GPR.
39319 * config/rs6000/rs6000.md (Fsignbit): Delete.
39320 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
39321 Update the length attribute if the value is in a GPR.
39322 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
39323 the sign or zero extension instruction, since the value is always 0/1.
39324 (signbit<mode>2_dm2): Delete using <Fsignbit>.
39327 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
39328 extracting SImode to a GPR register so that we can generate a
39329 store, limit the vector to be in a traditional Altivec register
39330 for the vextuwrx instruction.
39332 2017-01-03 Ian Lance Taylor <iant@google.com>
39334 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
39336 2017-01-03 Martin Sebor <msebor@redhat.com>
39338 PR tree-optimization/78696
39339 * gimple-ssa-sprintf.c (format_floating): Correct handling of
39340 precision. Use MPFR for %f for greater fidelity. Correct handling
39342 (pass_sprintf_length::compute_format_length): Set width and precision
39343 specified by asrerisk to void_node for vararg functions.
39344 (try_substitute_return_value): Adjust dump output.
39346 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
39348 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
39350 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
39352 * doc/invoke.texi (SPARC options): Document -mlra as the default.
39353 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
39354 -mlra/-mno-lra was passed to the compiler.
39356 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
39358 PR rtl-optimization/65618
39359 * emit-rtl.c (try_split): Move initialization of "before" and
39360 "after" to just before the call to emit_insn_after_setloc.
39362 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
39364 * doc/md.texi (Standard Names): Remove reference to Java frontend.
39366 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
39368 * dwarf2out.c (gen_enumeration_type_die): When
39369 -gno-strict-dwarf, add a DW_AT_encoding attribute.
39371 2017-01-03 Jakub Jelinek <jakub@redhat.com>
39373 PR tree-optimization/78965
39374 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
39375 Change first argument from const call_info & to call_info &. For %n
39376 set info.nowrite to false.
39378 PR middle-end/78901
39379 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
39380 possibly throwing calls.
39382 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
39383 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
39384 and fns handling, rather than in a separate case SSA_NAME.
39386 2017-01-02 Jeff Law <law@redhat.com>
39388 * config/darwin-driver.c (darwin_driver_init): Const-correctness
39389 fixes for first_period and second_period variables.
39391 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
39394 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
39395 (*insvqi_1): New insn pattern.
39396 (*insvqi_1_mem_rex64): Ditto.
39397 (*insvqi_2): Ditto.
39398 (*insvqi_3): Rename from *insvqi.
39400 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
39402 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
39404 * doc/cfg.texi (Edges): Remove reference to Java.
39405 (Maintaining the CFG): Ditto.
39407 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39409 PR middle-end/77674
39410 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
39411 transparent aliases.
39413 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39415 PR middle-end/77484
39416 * predict.def (PRED_CALL): Update hitrate.
39417 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
39418 * predict.c (tree_estimate_probability_bb): Split CALL predictor
39419 into direct/indirect/polymorphic variants.
39421 2017-01-01 Jakub Jelinek <jakub@redhat.com>
39423 Update copyright years.
39425 * gcc.c (process_command): Update copyright notice dates.
39426 * gcov-dump.c (print_version): Ditto.
39427 * gcov.c (print_version): Ditto.
39428 * gcov-tool.c (print_version): Ditto.
39429 * gengtype.c (create_file): Ditto.
39430 * doc/cpp.texi: Bump @copying's copyright year.
39431 * doc/cppinternals.texi: Ditto.
39432 * doc/gcc.texi: Ditto.
39433 * doc/gccint.texi: Ditto.
39434 * doc/gcov.texi: Ditto.
39435 * doc/install.texi: Ditto.
39436 * doc/invoke.texi: Ditto.
39438 Copyright (C) 2017 Free Software Foundation, Inc.
39440 Copying and distribution of this file, with or without modification,
39441 are permitted in any medium without royalty provided the copyright
39442 notice and this notice are preserved.